System for estimating amount of cloud water and amount of water vapor using dual-frequency radar

Abstract

To provide a system for estimating an amount of cloud water and an amount of water vapor using a dual-frequency radar, capable of timely and appropriately observing the amount of water vapor contained in a cloud or a cumulus cloud before the occurrence of a rainfall, the distribution of a place where the amount of cloud water is large, or the like, and quickly coping with a disaster.SOLUTION: A system for estimating an amount of cloud water and an amount of water vapor includes: weather simulation means for calculating three-dimensional distribution weather data based on a cloud resolution numerical weather model; machine learning data creation means for creating machine learning data including one-dimensional distribution data of a temperature, an atmospheric pressure, the amount of cloud water, the amount of water vapor, a dual-frequency radar reflection factor, and a dual-frequency ratio with respect to a distance from a radar; machine learning means for creating an estimation model for estimating the one-dimensional distribution data of the amount of cloud water and the amount of water vapor with respect to the distance from the radar by performing machine learning using the machine learning data as teacher data; and cloud water amount and water vapor amount estimation means for estimating the one-dimensional distribution data of the amount of cloud water and the amount of water vapor with respect to the distance from the radar.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to a cloud water content and water vapor content estimation system using dual-frequency radar. More specifically, the present invention relates to a cloud water content and water vapor content estimation system using dual-frequency radar, a cloud water content and water vapor content estimation method using dual-frequency radar, a cloud water content and water vapor content estimation program using dual-frequency radar, a computer-readable recording medium having recorded thereon a cloud water content and water vapor content estimation program using dual-frequency radar, and a data structure for estimating cloud water content and water vapor content using dual-frequency radar. [Background technology]

[0002] Recent research into climate change has assessed that there is a high possibility that the frequency of heavy rainfall and the proportion of heavy rainfall in total precipitation will increase. Furthermore, in recent years, localized, sudden, and sporadic heavy rainfall that is difficult to predict with current technology has occurred frequently, and the occurrence of so-called "linear rain bands" and "guerilla downpours" has been widely reported.

[0003] In order to take disaster prevention measures against extreme weather, which is a weather phenomenon that is significantly different from weather phenomena that have occurred in the past, a technology has been proposed that identifies areas where rain is likely to start falling and observes the surrounding area with radar to detect the start of rain (first echo) more quickly.

[0004] For example, Patent Document 1 discloses a weather radar control device that controls a weather radar that rotates to emit radio waves into its surroundings and receives reflected waves of the radio waves to detect rainfall conditions, and that includes a prediction means that predicts rainfall based on information corresponding to the amount of water vapor in the atmosphere, and a control means that, when the prediction means predicts rainfall, causes the weather radar to emit the radio waves based on the results of the prediction, and the prediction means determines the information corresponding to the amount of water vapor in the atmosphere based on the amount of delay in the atmosphere of a GPS signal.

[0005] Furthermore, Patent Document 2 discloses a water vapor observation device that calculates the amount of water vapor contained in the atmosphere, characterized in that it comprises a wave transmitting unit that transmits a first transmission wave and a second transmission wave having different frequencies, a wave receiving unit that receives as received waves the reflected waves of the transmission waves that have passed through the water vapor and are reflected by the ground surface or the water surface and return, and a calculation processing unit that calculates the amount of water vapor contained in the area through which the transmission waves have passed, based on first reception information generated from the first reception wave as the received wave obtained from the first transmission wave and second reception information generated from the second reception wave as the received wave obtained from the second transmission wave. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] International Publication No. 2016 / 035496 [Patent Document 2] International Publication No. 2017 / 018062 Summary of the Invention [Problem to be solved by the invention]

[0007] However, the weather radar control device in Patent Document 1 aims to detect precipitation early by detecting areas with large amounts of water vapor from GPS radio waves and concentrating observation on areas of normal weather radar, so it cannot observe the state of clouds before precipitation forms. Furthermore, since Patent Document 1 uses GPS for observation, it can determine the vertical integrated amount of water vapor, but it cannot observe the distribution of cloud locations or the state of cloud spread, especially the spread of clouds in the vertical direction (height direction).

[0008] Furthermore, the water vapor observation device in Patent Document 2 receives waves reflected from the ground surface or water surface as received waves, and therefore cannot observe the amount of cloud water or water vapor in the sky where clouds form.

[0009] Therefore, the techniques of Patent Documents 1 and 2 may not be able to timely and appropriately observe the distribution of locations with large amounts of cloud water or water vapor contained in clouds or cumulus clouds before precipitation occurs.

[0010] The present invention has been made in light of the above circumstances, and its main object is to provide a cloud water content and water vapor content estimation system using dual-frequency radar, a cloud water content and water vapor content estimation method using dual-frequency radar, a cloud water content and water vapor content estimation program using dual-frequency radar, a computer-readable recording medium having recorded thereon the cloud water content and water vapor content estimation program using dual-frequency radar, and a data structure for estimating cloud water content and water vapor content using dual-frequency radar, which are capable of timely and appropriately observing the amount of water vapor contained in clouds or cumulus clouds before precipitation occurs and the distribution of areas with high cloud water content, thereby enabling prompt disaster response. [Means for solving the problem]

[0011] As a result of intensive research conducted by the inventors in order to achieve the above-mentioned object, they have succeeded in making it possible to timely and appropriately observe the amount of water vapor contained in clouds or cumulus clouds before precipitation occurs, the distribution of areas with a large amount of cloud water, etc., and to take prompt disaster response measures, thereby completing the present invention.

[0012] That is, in a first aspect, the present invention provides: a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factors of two frequencies, and a two-frequency ratio of the two frequencies relative to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; and cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar.

[0013] The cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to the first aspect of the present invention comprises: The apparatus may further include a cloud water content / water vapor three-dimensional distribution estimation means for estimating a three-dimensional distribution of cloud water content and water vapor content with respect to the distance from the radar based on the one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar estimated by the cloud water content / water vapor content estimation means.

[0014] In the cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to the first aspect of the present invention, The cloud-resolving numerical weather model may be a bin-based model or a bulk-based model.

[0015] In the cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to the first aspect of the present invention, One-dimensional distribution data of the observation altitude with respect to the distance from the radar may be added as the machine learning data, The one-dimensional distribution data with respect to the distance from the radar at the observation altitude may be added as an input of the estimation model created by the machine learning means, The cloud water content and water vapor content estimation means may input one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, radar reflectivity factor of the two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimate one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

[0016] In a second aspect of the present invention, a meteorological simulation calculation step of calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation step for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies relative to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation calculation step; a machine learning step of inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to distance from the radar, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar, by performing machine learning on the machine learning data created in the machine learning data creation step as training data; and a cloud water content and water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and dual-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar.

[0017] In a third aspect of the present invention, A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factor of two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; The present invention provides a cloud water content and water vapor content estimation program using a dual-frequency radar, which functions as a cloud water content and water vapor content estimation means that inputs one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and the two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimates one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar.

[0018] In a fourth aspect of the present invention, A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factor of two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; A computer-readable recording medium is provided which stores a cloud water content and water vapor content estimation program using a dual-frequency radar, and which functions as a cloud water content and water vapor content estimation means which inputs one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and the two-frequency ratio of the two frequencies versus distance from the radar into the estimation model, and estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar.

[0019] In a fifth aspect of the present invention, Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distributed meteorological data calculated based on a cloud-resolving numerical meteorological model; Machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, dual-frequency radar reflectivity factor, and a two-frequency ratio of the dual frequencies relative to the distance from the radar, which data is created based on the calculated three-dimensional distribution meteorological data; One-dimensional distribution estimation data of cloud water content and water vapor content with respect to distance from the radar, which is estimated based on one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and dual-frequency ratio with respect to distance from the radar by machine learning the machine learning data as teacher data; The present invention provides a data structure for estimating cloud water content and water vapor content using a dual-frequency radar, which includes one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of a dual frequency, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and cloud water content and water vapor content estimation data related to one-dimensional distribution data of estimated cloud water content and water vapor content with respect to the distance from the radar, based on the one-dimensional distribution estimation data.

[0020] In addition, as a sixth aspect, the present invention provides: a meteorological simulation means based on the bin method model, which calculates three-dimensional distribution data of temperature, atmospheric pressure, water vapor content, and cloud particle size distribution based on the bin method model; one-dimensional distribution data conversion means for interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation means based on the bin method model into a radar coordinate system and converting it into one-dimensional distribution data relative to the distance from the radar; cloud water content calculation means for calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted by the one-dimensional distribution data conversion means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted by the one-dimensional distribution data converting means, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated by the cloud water content calculating means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0021] In the cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to a sixth aspect of the present invention, The radar reflectivity factor / two-frequency ratio calculation means may include a radio wave scattering simulation means for calculating attenuation and scattering of radio waves, and a radio wave propagation simulation means for calculating propagation of the radio waves.

[0022] In the cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to a sixth aspect of the present invention, One-dimensional distribution data of the observation altitude with respect to the distance from the radar may be added as data included in the data set, The one-dimensional distribution data with respect to the distance from the radar at the observation altitude may be added as an explanatory variable in the machine learning means, The one-dimensional distribution data with respect to the distance from the radar at the observation altitude may be added as an input of the estimation model created by the machine learning means, The cloud water content and water vapor content estimation means may input one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, radar reflectivity factor of the two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimate one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

[0023] In a seventh aspect of the present invention, a meteorological simulation step based on the bin method model, which calculates three-dimensional distribution data of temperature, pressure, water vapor content, and cloud particle size distribution based on the bin method model; a one-dimensional distribution data conversion step of interpolating the three-dimensional distribution data of the temperature, pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation step based on the bin method model into a radar coordinate system and converting the data into one-dimensional distribution data relative to the distance from the radar; a cloud water content calculation step of calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted in the one-dimensional distribution data conversion step; a radar reflectivity factor / two-frequency ratio calculation step of calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted in the one-dimensional distribution data conversion step; a data set creation step of creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted in the one-dimensional distribution data conversion step, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated in the cloud water content calculation step, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies calculated in the radar reflectivity factor / two-frequency ratio calculation step versus distance from the radar; a machine learning step of using the dataset created in the dataset creation step as training data, and using temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; a cloud water content and water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor with respect to distance from the radar; and a three-dimensional distribution data conversion step of interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into a three-dimensional grid and converting it into three-dimensional distribution data of the cloud water content and water vapor content.

[0024] In an eighth aspect of the present invention, A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on the bin method model, which calculates three-dimensional distribution data of temperature, atmospheric pressure, water vapor content, and cloud particle size distribution based on the bin method model; one-dimensional distribution data conversion means for interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation means based on the bin method model into a radar coordinate system and converting it into one-dimensional distribution data relative to the distance from the radar; cloud water content calculation means for calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted by the one-dimensional distribution data conversion means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted by the one-dimensional distribution data converting means, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated by the cloud water content calculating means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0025] In a ninth aspect of the present invention, A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on the bin method model, which calculates three-dimensional distribution data of temperature, atmospheric pressure, water vapor content, and cloud particle size distribution based on the bin method model; one-dimensional distribution data conversion means for interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation means based on the bin method model into a radar coordinate system and converting it into one-dimensional distribution data relative to the distance from the radar; cloud water content calculation means for calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted by the one-dimensional distribution data conversion means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted by the one-dimensional distribution data converting means, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated by the cloud water content calculating means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0026] In a tenth aspect of the present invention, Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distribution data of temperature, pressure, water vapor content, and cloud particle size distribution calculated based on the bottle method model, one-dimensional distribution data obtained by interpolating the three-dimensional distribution data of the calculated temperature, atmospheric pressure, water vapor content, and cloud particle size distribution into a radar coordinate system and converting it into a one-dimensional distribution with respect to the distance from the radar; and one-dimensional distribution data of cloud liquid water volume with respect to distance from the radar calculated based on the one-dimensional distribution data of the converted cloud particle size distribution with respect to distance from the radar; one-dimensional distribution data of a radar reflectivity factor of a dual frequency and a two-frequency ratio of the dual frequencies with respect to the distance from the radar, calculated based on the one-dimensional distribution data of the converted air temperature, air pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar; and a data set including the one-dimensional distribution data of the converted air temperature, air pressure, and water vapor content versus distance from the radar, the one-dimensional distribution data of the calculated cloud water content versus distance from the radar, and the one-dimensional distribution data of the calculated dual-frequency radar reflectivity factor and dual-frequency ratio of the dual-frequency versus distance from the radar; One-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is estimated based on one-dimensional distribution data of the temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies with the data set as training data, and one-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is obtained by machine learning using the data set as training data, and using temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables; cloud water content and water vapor content estimation data relating to the one-dimensional distribution data of cloud water content and water vapor content estimated based on one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar, and the one-dimensional distribution estimation data; The cloud water content and water vapor content estimation data is interpolated into a three-dimensional grid and processed, and the converted cloud water content and water vapor content three-dimensional distribution data is used to provide a data structure for estimating cloud water content and water vapor content using a dual-frequency radar.

[0027] Furthermore, in an eleventh aspect of the present invention, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; cloud particle size distribution calculation means for calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of cloud water amount with respect to distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated by the cloud particle size distribution calculation means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0028] In the cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to an eleventh aspect of the present invention, The radar reflectivity factor / two-frequency ratio calculation means may include a radio wave scattering simulation means for calculating attenuation and scattering of radio waves, and a radio wave propagation simulation means for calculating propagation of the radio waves.

[0029] In the cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to an eleventh aspect of the present invention, One-dimensional distribution data of the observation altitude with respect to the distance from the radar may be added as data included in the data set, The one-dimensional distribution data with respect to the distance from the radar at the observation altitude may be added as an explanatory variable in the machine learning means, The one-dimensional distribution data with respect to the distance from the radar at the observation altitude may be added as an input of the estimation model created by the machine learning means, The cloud water content and water vapor content estimation means may input one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, radar reflectivity factor of the two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimate one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

[0030] In a twelfth aspect of the present invention, a meteorological simulation step based on a bulk method model, which calculates three-dimensional distribution data of temperature, pressure, cloud water content, and water vapor content based on the bulk method model; a one-dimensional distribution data conversion step of interpolating the three-dimensional distribution data of the temperature, pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation step based on the bulk method model into a radar coordinate system and converting the data into one-dimensional distribution data relative to the distance from the radar; a cloud particle size distribution calculation step of calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of the cloud water amount with respect to distance from the radar converted in the one-dimensional distribution data conversion step; a radar reflectivity factor / two-frequency ratio calculation step of calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted in the one-dimensional distribution data conversion step and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated in the cloud particle size distribution calculation step; a data set creation step of creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud liquid water content versus distance from the radar converted in the one-dimensional distribution data conversion step, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated in the radar reflectivity factor / two-frequency ratio calculation step; a machine learning step of using the dataset created in the dataset creation step as training data, and using temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; a cloud water content and water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor with respect to distance from the radar; and a three-dimensional distribution data conversion step of interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into a three-dimensional grid and converting it into three-dimensional distribution data of the cloud water content and water vapor content.

[0031] In a thirteenth aspect of the present invention, A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; cloud particle size distribution calculation means for calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of cloud water amount with respect to distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated by the cloud particle size distribution calculation means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0032] In a fourteenth aspect of the present invention, A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; cloud particle size distribution calculation means for calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of cloud water amount with respect to distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated by the cloud particle size distribution calculation means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0033] In a fifteenth aspect of the present invention, Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distribution data of temperature, pressure, cloud water content, and water vapor content calculated based on the bulk method model, one-dimensional distribution data obtained by interpolating the three-dimensional distribution data of the calculated temperature, atmospheric pressure, water vapor amount, and cloud water content into a radar coordinate system and converting it into one-dimensional distribution data with respect to the distance from the radar; and one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar, calculated by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount, based on the one-dimensional distribution data of the converted cloud water amount with respect to distance from the radar; one-dimensional distribution data of a radar reflectivity factor of a dual frequency and a two-frequency ratio of the dual frequencies with respect to the distance from the radar, calculated based on the one-dimensional distribution data of the converted air temperature, air pressure, and water vapor content with respect to the distance from the radar and the one-dimensional distribution data of the calculated cloud particle size distribution with respect to the distance from the radar; a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud liquid water content versus distance from the radar converted by the one-dimensional distribution data conversion means, and the one-dimensional distribution data of the calculated radar reflectivity factor of the dual frequencies and the dual-frequency ratio of the dual frequencies versus distance from the radar; One-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is estimated based on one-dimensional distribution data of the temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies with the data set as training data, and one-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is obtained by machine learning using the data set as training data, and using temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables; cloud water content and water vapor content estimated data relating to one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar, which is estimated based on one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar; The cloud water content and water vapor content estimation data is interpolated into a three-dimensional grid and processed, and the converted cloud water content and water vapor content three-dimensional distribution data is used to provide a data structure for estimating cloud water content and water vapor content using a dual-frequency radar.

[0034] Furthermore, in a sixteenth aspect of the present invention, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0035] In the cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to a sixteenth aspect of the present invention, The radar reflectivity factor / two-frequency ratio calculation means may include a radio wave scattering simulation means for calculating attenuation and scattering of radio waves, and a radio wave propagation simulation means for calculating propagation of the radio waves.

[0036] In the cloud liquid water content and water vapor content estimation system using a dual-frequency radar according to a sixteenth aspect of the present invention, One-dimensional distribution data of the observation altitude with respect to the distance from the radar may be added as data included in the data set, The one-dimensional distribution data with respect to the distance from the radar at the observation altitude may be added as an explanatory variable in the machine learning means, The one-dimensional distribution data with respect to the distance from the radar at the observation altitude may be added as an input of the estimation model created by the machine learning means, The cloud water content and water vapor content estimation means may input one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, radar reflectivity factor of the two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimate one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

[0037] In a seventeenth aspect of the present invention, a meteorological simulation step based on a bulk method model, which calculates three-dimensional distribution data of temperature, pressure, cloud water content, and water vapor content based on the bulk method model; a one-dimensional distribution data conversion step of interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation step based on the bulk method model into a radar coordinate system and converting the data into one-dimensional distribution data relative to the distance from the radar; a radar reflectivity factor / two-frequency ratio calculation step of calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted in the one-dimensional distribution data conversion step, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creation step of creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud liquid water content versus distance from the radar converted in the one-dimensional distribution data conversion step, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated in the radar reflectivity factor / two-frequency ratio calculation step; a machine learning step of using the dataset created in the dataset creation step as training data, and using temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; a cloud water content and water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor with respect to distance from the radar; and a three-dimensional distribution data conversion step of interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into a three-dimensional grid and converting it into three-dimensional distribution data of the cloud water content and water vapor content.

[0038] In an eighteenth aspect of the present invention, A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0039] In a nineteenth aspect of the present invention, A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0040] In a twentieth aspect of the present invention, Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distribution data of temperature, pressure, cloud water content, and water vapor content calculated based on the bulk method model, one-dimensional distribution data obtained by interpolating the three-dimensional distribution data of the calculated temperature, atmospheric pressure, water vapor amount, and cloud water content into a radar coordinate system and converting it into one-dimensional distribution data with respect to the distance from the radar; and one-dimensional distribution data of the converted temperature, atmospheric pressure, water vapor amount, and cloud water content versus distance from the radar; and one-dimensional distribution data of the radar reflectivity factor of dual frequencies and the dual-frequency ratio of the dual frequencies versus distance from the radar, calculated based on empirical relations regarding temperature, atmospheric pressure, cloud water content, water vapor amount, and radio wave attenuation and scattering. a data set including one-dimensional distribution data of the converted air temperature, air pressure, water vapor content, and cloud liquid water content with respect to the distance from the radar, and one-dimensional distribution data of the calculated radar reflectivity factor of the dual frequencies and the dual-frequency ratio of the dual frequencies with respect to the distance from the radar; One-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is estimated based on one-dimensional distribution data of the temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies with the data set as training data, and one-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is obtained by machine learning using the data set as training data, and using temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables; cloud water content and water vapor content estimated data relating to one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar, which is estimated based on one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar; The cloud water content and water vapor content estimation data is interpolated into a three-dimensional grid and processed to convert it into three-dimensional distribution data of cloud water content and water vapor content. [Effects of the Invention]

[0041] According to the present invention, it is possible to timely and appropriately observe the amount of water vapor contained in clouds or cumulus clouds before precipitation occurs, the distribution of areas with a large amount of cloud water, etc., and to take prompt disaster response measures. Note that the effects described herein are not necessarily limited to those described herein, and may be any of the effects described in this specification. [Brief explanation of the drawings]

[0042] [Figure 1] FIG. 1 is a flowchart showing the creation of an estimation model using the bin method model. [Figure 2] FIG. 2 is a flowchart showing the creation of an estimation model (assuming particle size distribution) using the bulk method model. [Figure 3] FIG. 3 is a flowchart showing the creation of an estimation model (using an empirical formula) using the bulk method model. [Figure 4] Figure 4 is a flowchart showing how cloud water content and water vapor content are estimated from radar observation data using an estimation model. [Figure 5] FIG. 5 is a block diagram showing an example of the configuration of a system for estimating cloud water content and water vapor content using a dual-frequency radar. [Figure 6] FIG. 6 is a block diagram showing another example of the configuration of a system for estimating cloud water content and water vapor content using a dual-frequency radar. [Figure 7] Figure 7 is a flowchart showing the relationship between meteorological parameters and parameters observed and analyzed by radar. [Figure 8] FIG. 8 is a block diagram showing the configuration of the estimation model creating means. [Figure 9] Figure 9 is a schematic diagram of the radar coordinate system. [Figure 10] FIG. 10 is a diagram showing the spatial distribution of meteorological parameters interpolated into the radar coordinate system shown in FIG. [Figure 11] FIG. 11 shows the spatial distribution of cloud water content calculated from the particle size distribution shown in FIG. [Figure 12]FIG. 12 is a diagram showing data calculated by a radio wave scattering simulation from the meteorological parameters shown in FIG. [Figure 13] FIG. 13 shows the radar reflectivity factors of the W-band and D-band predicted to be observed by radar and their two-frequency ratios, calculated using the results of a radio wave scattering simulation, a radio wave propagation simulation, and radar equations. [Figure 14] FIG. 14 shows the results of an example of estimating cloud water content and water vapor content using the estimation model. DETAILED DESCRIPTION OF THE INVENTION

[0043] A preferred embodiment for carrying out the present invention will be described below. The embodiment described below shows an example of a typical embodiment of the present invention, and the scope of the present invention should not be construed as being narrow.

[0044] Unless otherwise specified, in the drawings, "upper" means the upper direction or upper side in the drawing, "lower" means the lower direction or lower side in the drawing, "left" means the left direction or left side in the drawing, and "right" means the right direction or right side in the drawing. Furthermore, in the drawings, the same or equivalent elements or members are given the same reference numerals, and redundant explanations will be omitted.

[0045] The explanation will be given in the following order. 1. Overview of the Invention 2. Explanation of the bottle and bulk method models 3. Creating an estimation model using the bin method 4. Creation of an estimation model using the bulk method model (assuming particle size distribution) and creation of an estimation model using the bulk method model (using an empirical formula) 5. Estimation of cloud water content and water vapor content using estimation models from radar observation data 6. Example of generating data for machine learning using the bin method model 7. Example of Estimation of Cloud Water and Water Vapor Using the Estimation Model

[0046] <1. Overview of the present invention> First, an outline of the present invention will be described.

[0047] Using a weather radar that generates radio waves at multiple frequencies that undergo Rayleigh scattering in clouds and have different attenuation characteristics due to at least one of clouds and water vapor, and that observes the distribution of radio wave attenuation per unit distance from the distance derivative of the ratio of the radar reflectivity factors of the two frequencies, and that takes advantage of the fact that radio wave attenuation per unit distance increases in areas with high water vapor and cloud water content, it is possible to predict the occurrence of cumulonimbus clouds.

[0048] Radio wave attenuation is mainly caused by the influence of cloud water and water vapor, but conventional technology is unable to separate these influences, making it impossible to determine the distribution of cloud water and water vapor amounts.

[0049] The present invention provides a method for estimating the distribution of cloud water content and water vapor content from the distribution of radar reflectivity observed by a dual-frequency weather radar.

[0050] In recent years, advances in cloud-resolving numerical weather models have led to the development of a scheme for forecasting the size distribution of cloud and precipitation particles (Spectrum Bin Microphysics). If the size distribution of cloud and precipitation particles is known, scattering simulations can be used to calculate radar reflectivity as well as radio wave attenuation by clouds.

[0051] On the other hand, weather models also forecast temperature, pressure, and water vapor, so they can also calculate radio wave attenuation due to moist atmosphere.These calculation results make it possible to simulate radar reflectivity factors, including the effects of radio wave attenuation, observed at each radar wavelength.However, it is not possible to determine the distribution of cloud water content and water vapor content from the radar reflectivity factor or its two-frequency ratio.

[0052] Therefore, in the present invention, (1) In advance, cloud simulations for various seasons and cases are performed using a cloud-resolving numerical weather model, and a data set of temperature, pressure, cloud water content, water vapor content, radar reflectivity factors at each frequency, and their two-frequency ratios is created. (2) Using this data set as training data, an estimation model was constructed using machine learning with temperature, pressure, radar reflectivity factors and their dual-frequency ratios as explanatory variables, and water vapor and cloud liquid water content as objective variables. (3) By inputting the radar reflectivity factor and its two-frequency ratio actually observed by radar into the estimation model, the amount of cloud water and water vapor can be estimated.

[0053] Next, a description will be given with reference to Fig. 7. Fig. 7 is a diagram showing a flowchart illustrating the relationship between meteorological parameters and parameters observed and analyzed by radar, more specifically, a diagram showing a flowchart 7000 illustrating the relationship between meteorological parameters and parameters observed and analyzed by radar.

[0054] By performing a cloud reproduction simulation using the cloud-resolving numerical weather models S716 (bulk method) and S717 (bin method), it is possible to obtain the atmospheric pressure S711, temperature S712, water vapor content S713 around the cloud, and information about the cloud such as cloud water content S714 and cloud particle size distribution S715.

[0055] There are two types of cloud-resolving weather models (reference numbers S716 and S717), the bulk method and the bottle method. The bulk method model S716 outputs cloud liquid water S714 as information about clouds. The bottle method model S717 outputs cloud particle size distribution S715 as information about clouds. Because cloud liquid water can be calculated from cloud particle size distribution, it is also possible to obtain cloud liquid water from the bottle method model S717.

[0056] From information on the atmospheric pressure S711, the temperature S712, and the water vapor S713, the radio wave attenuation due to dry air S706 and the radio wave attenuation due to water vapor S707 can be calculated.

[0057] Radio wave attenuation by cloud water (S708) and radio wave scattering by cloud water (S709) can be calculated by radio wave scattering simulation using cloud particle size distribution (S715) and temperature (S712).

[0058] Although the output of the bulk method model S716 does not include cloud particle size distribution, it is possible to assume some kind of cloud particle size distribution from the cloud water content S714, and radio wave attenuation due to cloud water S708 and radio wave scattering due to cloud water S709 can be calculated by radio wave scattering simulation using the assumed particle size distribution and temperature S712. In addition, by using an empirical relationship between cloud water content and radio wave attenuation / scattering, it is also possible to calculate radio wave attenuation and scattering due to cloud water from the cloud water content and temperature.

[0059] From the information on radio wave attenuation and scattering obtained as described above, the received power including the influence of radio wave attenuation can be obtained by radio wave propagation simulation for the first frequency (W band in this example, which may indicate a range of 75 GHz to 110 GHz, for example) and the second frequency (D band in this example, which may indicate a range of 110 GHz to 170 GHz, for example) (S704 and S705).

[0060] In addition, since actual radar observation results also contain noise S710, taking noise into consideration makes it possible to express received power that is closer to actual observations.

[0061] From these received powers, radar reflectivity factors (S702 and S703) of the first and second frequencies, including the influence of radio wave attenuation, can be calculated using a radar equation.

[0062] Furthermore, the two-frequency ratio can be calculated from the ratio (or difference when expressed in decibels) of the radar reflectivity factors of each frequency (S701). The observed radar reflectivity factor is calculated by taking into account the effect of radio wave attenuation along the radio wave propagation path from the true radar reflectivity factor, but if the particle size of cloud particles is sufficiently smaller than the wavelength of the radio waves used (Rayleigh scattering), the true radar reflectivity factor does not depend on the frequency of the radio waves. Therefore, when the two-frequency ratio is calculated, the true radar reflectivity factor cancels out, and as a result, the two-frequency ratio is equal to the ratio (or difference when expressed in decibels) of radio wave attenuation at each frequency.

[0063] If the above calculations were reversible, the original weather parameters could be calculated from the radar observation results, but since many of the above calculations are irreversible (inverse functions cannot be defined), the original weather parameters cannot be obtained from the radar observation results.

[0064] Therefore, we will use a cloud-resolving numerical weather model to carry out experiments reproducing various clouds, calculate the radar reflectivity factor and dual-frequency ratio from the results, and create a large number of datasets for machine learning (indicated by stars in Figure 7). From this dataset, we will create an estimation model S750 in which the radar reflectivity factor, dual-frequency ratio, temperature, and atmospheric pressure data are explanatory variables, and the cloud water content and water vapor content are target variables.

[0065] When the explanatory variables, radar reflectivity factor, dual frequency ratio, temperature, and air pressure, are input into the created estimation model S750, the cloud water content and water vapor content are output.

[0066] This will be explained using Fig. 5. Fig. 5 is a block diagram showing an example of the configuration of a cloud water content and water vapor content estimation system using a dual-frequency radar, more specifically, a block diagram showing the configuration of a cloud water content and water vapor content estimation system 5000 using a dual-frequency radar.

[0067] The cloud water content and water vapor content estimation system 5000 using a dual-frequency radar includes an estimation model creation means 501 (corresponding to an estimation model creation means 8000 in FIG. 8 described later), a cloud water content and water vapor content estimation means 502, and may further include a cloud water content and water vapor content three-dimensional distribution estimation means 503.

[0068] This will be explained using Figure 6. Figure 6 is a block diagram showing another example of the configuration of a cloud water content and water vapor content estimation system using dual-frequency radar, more specifically, a block diagram showing the configuration of a cloud water content and water vapor content estimation system 6000 using dual-frequency radar. The cloud water content and water vapor content estimation system 6000 using dual-frequency radar includes an estimation model creation means 601 (corresponding to an estimation model creation means 8000 in Figure 8 described later), a cloud water content and water vapor content estimation means 602, and a three-dimensional distribution data conversion means 604, and may further include an archive means 603 of distance distribution at various azimuths and elevation angles.

[0069] Further explanation will be given with reference to Fig. 8. Fig. 8 is a block diagram showing the configuration of the estimation model creation means, more specifically, a block diagram showing the configuration of estimation model creation means 8000.

[0070] The estimation model creation means 8000 includes a meteorological simulation means 801, a machine learning data creation means 8001, and a machine learning means 810. The machine learning data creation means 8001 can include a one-dimensional distribution data conversion means 802, a radar reflectivity factor / two-frequency ratio calculation means 804, a cloud liquid water content calculation means 807, a cloud particle size distribution calculation means 803, a data set creation means 808, and a plurality of observation case archive means 809. The radar reflectivity factor / two-frequency ratio calculation means 804 can include a radio wave scattering simulation means 805 and a radio wave propagation simulation means 806.

[0071] The meteorological simulation means 801 calculates three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model, which may be a bin method model, a bulk method model (assuming particle size distribution), or a bulk method model (using empirical formulas).

[0072] If the cloud-resolving numerical weather model is a bin method model, 1. In the one-dimensional distribution data conversion means 802, the three-dimensional distribution data of temperature, atmospheric pressure, water vapor content, and cloud particle size distribution calculated by the weather simulation means 801 is interpolated into the radar coordinate system and converted into one-dimensional distribution data relative to the distance from the radar. 2. The cloud water content calculation means 807 calculates one-dimensional distribution data of cloud water content versus distance from the radar based on the one-dimensional distribution data of cloud particle size distribution versus distance from the radar converted by the one-dimensional distribution data conversion means 802. 3. In the radar reflectivity factor / two-frequency ratio calculation means 804, one-dimensional distribution data of the two-frequency radar reflectivity factor and the two-frequency ratio of the two frequencies versus the distance from the radar is calculated based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution versus the distance from the radar converted by the one-dimensional distribution data conversion means 802. 4. A data set creation means 808 creates a data set including one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted by the one-dimensional distribution data conversion means 802, one-dimensional distribution data of the cloud water content versus distance from the radar calculated by the cloud water content calculation means 807, and one-dimensional distribution data of the dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies calculated by the radar reflectivity factor / dual-frequency ratio calculation means versus distance from the radar. 5. In the machine learning means 810, the dataset created by the dataset creation means 808 is used as training data, and machine learning is performed using the temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and the cloud water content and water vapor content as target variables.By inputting one-dimensional distribution data of the temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar, an estimation model can be created that estimates one-dimensional distribution data of the cloud water content and water vapor content versus distance from the radar.

[0073] If the cloud-resolving numerical weather model is a bulk method model (assuming particle size distribution), 1. In the one-dimensional distribution data conversion means 802, the three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means 801 is interpolated into the radar coordinate system and converted into one-dimensional distribution data relative to the distance from the radar. 2. In the cloud particle size distribution calculation means 803, based on the one-dimensional distribution data of the cloud water content versus distance from the radar converted by the one-dimensional distribution data conversion means 802, one-dimensional distribution data of the cloud particle size distribution versus distance from the radar is calculated by assuming a cloud particle size distribution having approximately the same cloud water content as the converted cloud water content. 3. In the radar reflectivity factor / two-frequency ratio calculation means 804, one-dimensional distribution data of the two-frequency radar reflectivity factor and the two-frequency ratio of the two frequencies versus the distance from the radar is calculated based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus the distance from the radar converted by the one-dimensional distribution data conversion means 802, and the one-dimensional distribution data of the cloud particle size distribution versus the distance from the radar calculated by the cloud particle size distribution calculation means 803. 4. The data set creation means 808 creates a data set including one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content versus distance from the radar converted by the one-dimensional distribution data conversion means 802, and one-dimensional distribution data of the dual-frequency radar reflectivity factor and the dual-frequency ratio of the dual-frequency versus distance from the radar calculated by the radar reflectivity factor / dual-frequency ratio calculation means. 5. In the machine learning means 810, the dataset created by the dataset creation means 808 is used as training data, and machine learning is performed using the temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and the cloud water content and water vapor content as target variables, and by inputting one-dimensional distribution data of the temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar, an estimation model is created that estimates one-dimensional distribution data of the cloud water content and water vapor content versus distance from the radar.

[0074] If the cloud-resolving numerical weather model is a bulk method model (using empirical formulas), 1. In the one-dimensional distribution data conversion means 802, the three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means 801 is interpolated into the radar coordinate system and converted into one-dimensional distribution data relative to the distance from the radar. 2. The radar reflectivity factor and dual-frequency ratio calculation means 804 calculates one-dimensional distribution data of the dual-frequency radar reflectivity factor and the dual-frequency ratio of the dual frequencies versus distance from the radar based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content versus distance from the radar converted by the one-dimensional distribution data conversion means 802, and an empirical relational expression (empirical expression 804-1) relating to the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering. 3. The data set creation means 808 creates a data set including one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content versus distance from the radar converted by the one-dimensional distribution data conversion means 802, and one-dimensional distribution data of the dual-frequency radar reflectivity factor and the dual-frequency ratio of the dual-frequency calculated by the radar reflectivity factor / dual-frequency ratio calculation means versus distance from the radar. 4. In the machine learning means 810, the dataset created by the dataset creation means 808 is used as training data, and machine learning is performed using temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as target variables, and an estimation model is created that estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies.

[0075] Hereinafter, embodiments of the present invention will be described specifically and in detail.

[0076] <2. Explanation of the bottle method and bulk method models> Cloud-resolving numerical weather models are used to reproduce cloud-scale weather such as cumulus and cumulonimbus clouds. In cloud-resolving numerical weather models, meteorological parameters such as temperature, pressure, and humidity are given to a three-dimensional grid with a horizontal grid spacing of about 1 km in order to resolve clouds, and their temporal changes are calculated using equations of motion, mass conservation, thermodynamic equations, etc. In this process, the temporal changes in clouds and precipitation are also calculated, and cloud-resolving numerical weather models are classified into two types depending on the method used to calculate clouds and precipitation.

[0077] In real clouds, water vapor condenses to produce tiny droplets (cloud particles). These cloud particles then repeatedly collide and merge, gradually increasing in size. When they reach the size of droplets (raindrops) with a falling velocity, they fall to the ground as rain. To represent this microphysical process, cloud and rain particles ranging in size from a few micrometers to a few millimeters are divided into dozens of categories (bins). This method of representing the growth of cloud particles into raindrops by calculating how the number of particles in each category changes due to processes such as droplet collision and merging is called the bin method (Spectra-Bin Microphysics), and cloud-resolving numerical weather models that use the bin method are called bin method models. The bin method model outputs information about clouds and precipitation, including the number of particles in each category (information describing how many particles of each size are present, known as the particle size distribution).

[0078] However, calculations using the bin method require enormous computational resources, and are therefore rarely used. Usually, the bulk method (Bulk Microphysics) is used, which simplifies calculations of how quantities such as cloud water content (the sum of the masses of cloud droplets contained per unit volume) and rainwater content (the sum of the masses of raindrops contained per unit volume) change due to processes such as collision and coalescence. Cloud-resolving numerical weather models that use the bulk method are called bulk method models. Bulk method models output the mass of each particle type, such as cloud water content and rainwater content, as information on clouds and precipitation.

[0079] Machine learning datasets can be created using either the bin method or the bulk method.

[0080] <3. Creating an estimation model using the bin method> The creation of a machine learning dataset using the bin method model and the creation of an estimation model using the machine learning dataset for estimating the three-dimensional distribution of cloud water content and water vapor content will be explained below with reference to Figure 1.

[0081] FIG. 1 is a flowchart illustrating the creation of an estimation model using a binning model, and more specifically, a flowchart 1000 for the creation of an estimation model using a binning model.

[0082] The bottle method model calculates temperature, pressure, humidity, and cloud particle size distribution, so both scattering by cloud water and the amount of cloud water can be calculated from the cloud particle size distribution.

[0083] Here, particle size distribution refers to the number of particles having a certain category of particle size contained per unit volume, and the category is a division of particle sizes ranging from several micrometers to several millimeters into several dozens of categories.

[0084] In the weather simulation means, predetermined three-dimensional distributed weather data provided by, for example, the Japan Meteorological Agency is input into a cloud-resolving numerical weather model (bin method model) as initial values ​​and boundary values, and three-dimensional distributed weather data (temperature, pressure, water vapor content, cloud particle size distribution) is calculated (S1 to S3).

[0085] Since the radar is a device that emits radio waves in a specified azimuth and elevation direction and measures the one-dimensional distribution of the received power versus the distance from the radar, the three-dimensional distribution weather data (temperature, pressure, water vapor content, cloud particle size distribution) calculated by the cloud-resolving numerical weather model is interpolated into the radar coordinate system by the one-dimensional data conversion means and converted into one-dimensional distribution weather data (temperature, pressure, water vapor content, cloud particle size distribution) versus the distance from the radar (S4 to S5).

[0086] From the converted one-dimensional distribution data of cloud particle size distribution with respect to the distance from the radar, the cloud water content calculation means calculates one-dimensional distribution data of cloud water content with respect to the distance from the radar (S6 to S7).

[0087] From the one-dimensional distribution meteorological data (temperature, atmospheric pressure, water vapor content, cloud particle size distribution) converted by the one-dimensional data conversion means, one-dimensional distribution data of attenuation and scattering versus distance from the radar (attenuation per unit distance due to dry air, attenuation per unit distance due to water vapor, attenuation per unit distance due to cloud water, and radio wave scattering due to cloud water) is calculated by the radio wave scattering simulation means (S8 to S9).

[0088] As a method for calculating the attenuation per unit distance of radio waves due to dry air and water vapor, for example, recommendations by the International Telecommunication Union Radiocommunication Sector (ITU-R) can be used.

[0089] ITU-R Recommendation P.676-12, “Attenuation by atmospheric gases and related effects”, 2019.

[0090] For example, the T-Matrix method is used to calculate the attenuation and scattering of radio waves per unit distance by clouds from the cloud particle size distribution.

[0091] [Mishchenko, MI and LD Travis, “T-matrix computations of light scattering by large spheroidal particles”, Opt. Commun., vol. 109, pp16-21, 1994.

[0092] From the one-dimensional distribution data of attenuation and scattering versus distance from the radar calculated by the radio wave scattering simulation means (attenuation per unit distance due to dry air, attenuation per unit distance due to water vapor, attenuation per unit distance due to cloud water, radio wave scattering due to cloud water), one-dimensional distribution data of received power versus distance from the radar (received power of the first frequency, received power of the second frequency) is calculated by the radio wave propagation simulation means, and further from the calculated one-dimensional distribution data of received power versus distance from the radar (received power of the first frequency, received power of the second frequency), one-dimensional distribution data of reflectivity factor and dual-frequency ratio versus distance from the radar (radar reflectivity factor observed at the first frequency, radar reflectivity factor observed at the second frequency, dual-frequency ratio) is calculated by the radar equation (S10 to S11).

[0093] Since noise exists in real observations, adding one-dimensional distribution data of noise with respect to the distance from the radar to the radio wave propagation simulation makes it possible to make the radio wave propagation simulation closer to real observations.

[0094] The calculated one-dimensional distribution data (temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factor observed at the first frequency, radar reflectivity factor observed at the second frequency, and dual-frequency ratio) relative to the distance from the radar is associated with a radar coordinate system by a dataset creation means to create a dataset for machine learning (S12 to S13).

[0095] Furthermore, a large amount of data sets for a plurality of cases and a plurality of observation directions are created by the data set creation means, and an archive is created as a machine learning data set (the data sets are saved) (S14).

[0096] Next, using the archived (saved) machine learning dataset as training data, an estimation model is created using an estimation machine learning means with temperature, atmospheric pressure, radar reflectivity factor observed at the first frequency, radar reflectivity factor observed at the second frequency, and dual-frequency ratio as explanatory variables, and cloud water content and water vapor content as objective variables (S15-S16).

[0097] This estimation model estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, pressure, radar reflectivity factor observed at the first frequency, radar reflectivity factor observed at the second frequency, and the two-frequency ratio versus distance from the radar.

[0098] In addition, by adding one-dimensional distribution data of the observation altitude relative to the distance from the radar to the machine learning dataset and adding the observation altitude to the explanatory variables, seasonal changes, etc. can also be taken into account in the machine learning.

[0099] As a method for creating such an estimation model, for example, XGBoost or a Gaussian process regression model can be used.

[0100] XGBoost: T. Chen and C. Guestrin, “XGBoost: A scalable tree boosting system”, Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD '16, New York, NY, USA, pp785-794, 2016.

[0101] Gaussian process regression model: C.M. Bishop, Motoda, et al. (translator), “Pattern Recognition and Machine Learning”, Maruzen Publishing, 2012.

[0102] <4. Creation of an estimation model using the bulk method model (assuming particle size distribution) and creation of an estimation model using the bulk method model (using an empirical formula)> The creation of a machine learning dataset using a bulk method model and the creation of an estimation model using this machine learning dataset to estimate the three-dimensional distribution of cloud water content and water vapor content are explained below with reference to Figures 2 and 3.

[0103] FIG. 2 is a flowchart showing the creation of an estimation model using a bulk method model (assuming particle size distribution), and more specifically, a flowchart 2000 relating to the creation of an estimation model using a bulk method model (assuming particle size distribution).

[0104] FIG. 3 is a flowchart showing the creation of an estimation model by a bulk method model (using an empirical formula), and more specifically, a flowchart 3000 relating to the creation of an estimation model by a bulk method model (using an empirical formula).

[0105] To calculate the attenuation and scattering caused by cloud water, the particle size distribution of the cloud water is required. However, the bulk method model does not calculate the cloud particle size distribution, but calculates the amount of cloud water. Therefore, the attenuation and scattering caused by cloud water are calculated by assuming a function of the particle size distribution of cloud water that has the same amount of cloud water as the calculated amount of cloud water, or by using empirical relationships between meteorological parameters and the attenuation and scattering of radio waves.

[0106] First, the calculation of attenuation and scattering due to cloud water by assuming a particle size distribution of the cloud water will be explained below.

[0107] In the weather simulation means, predetermined three-dimensional distributed weather data provided by, for example, the Japan Meteorological Agency is input into a cloud-resolving numerical weather model (bulk method model) as initial values ​​and boundary values, and three-dimensional distributed weather data (temperature, pressure, cloud water content, water vapor content) is calculated (S21 to S23).

[0108] The three-dimensional distribution weather data (temperature, pressure, cloud water content, water vapor content) calculated by the cloud-resolving numerical weather model (bulk method model) is interpolated to the radar coordinate system by one-dimensional data conversion means, and converted into one-dimensional distribution weather data (temperature, pressure, cloud water content, water vapor content) relative to the distance from the radar (S24-S25).

[0109] A cloud particle size distribution calculation means assumes a function such as an exponential function that has the same cloud water content as the converted cloud water content, and calculates one-dimensional distribution data of cloud particle size distribution with respect to the distance from the radar (S26 to S27).

[0110] For example, as a function of particle size distribution, N(D)=N0exp(-λD) An exponential function such as (Marshall and Palmer distribution) can be assumed, where D is the particle size, N(D) is the particle size distribution, and N0, λ, and λ are coefficients related to the shape of the particle size distribution.

[0111] The particle size distribution with a certain cloud water content Qc is Qc=∫0 ∞ ρ V(D) N(D) dD where ρ is the density of water and V(D) is the volume of a water droplet with diameter D.

[0112] The one-dimensional distribution data of cloud particle size distribution versus distance from the radar is calculated from one-dimensional distribution meteorological data (temperature, atmospheric pressure, cloud water content, water vapor content) versus distance from the radar converted by the one-dimensional data conversion means, and then the radio wave scattering simulation means calculates one-dimensional distribution data of attenuation and scattering versus distance from the radar (attenuation per unit distance due to dry air, attenuation per unit distance due to water vapor, attenuation per unit distance due to cloud water, radio wave scattering by cloud water) (S28-S29).

[0113] From the one-dimensional distribution data of attenuation and scattering versus distance from the radar calculated by the radio wave scattering simulation means (attenuation per unit distance due to dry air, attenuation per unit distance due to water vapor, attenuation per unit distance due to cloud water, radio wave scattering due to cloud water), one-dimensional distribution data of received power versus distance from the radar (received power of the first frequency, received power of the second frequency) is calculated by the radio wave propagation simulation means, and further from the calculated one-dimensional distribution data of received power versus distance from the radar (received power of the first frequency, received power of the second frequency), one-dimensional distribution data of the reflectivity factor and dual-frequency ratio versus distance from the radar (radar reflectivity factor observed at the first frequency, radar reflectivity factor observed at the second frequency, dual-frequency ratio) is calculated by the radar equation (S30 to S31).

[0114] Since actual observations contain noise, adding one-dimensional distribution data of the noise versus distance from the radar to the radio wave propagation simulation makes it possible to make the radio wave propagation simulation closer to actual observations.

[0115] Reference symbols S32 to S36 are the same as reference symbols S12 to S16 described in the above section <3. Creation of estimation model using bin method model>.

[0116] Next, we will explain how to calculate attenuation and scattering due to cloud water by using meteorological parameters (temperature, atmospheric pressure, cloud water content, water vapor content) and empirical relationships regarding radio wave attenuation and scattering.

[0117] In this case, the calculation of attenuation and scattering due to cloud water is the same as when calculating attenuation and scattering due to cloud water by assuming a particle size distribution, except that the calculation is performed using empirical relationships between meteorological parameters (temperature, atmospheric pressure, cloud water content, water vapor content) and radio wave attenuation and scattering (S46-S47).In addition, by adding one-dimensional distribution data for the distance from the radar at the observation altitude to the machine learning dataset and adding the observation altitude to the explanatory variables, it is also possible to take seasonal changes, etc., into account in the machine learning.

[0118] For example, the empirical formula proposed by Benoit (1968) can be used to calculate the amount of radio wave attenuation by cloud water, and the empirical formula proposed by Atlas (1954) can be used to calculate the scattering of radio waves by cloud water.

[0119] Benoit, A., “Signal attenuation due to neutral oxygen and water vapor, rain and clouds”, Microwave J., 11, 73-80, 1968 Atlas, D., “The estimation of cloud parameters by radar”, J. Meteor., 11, pp309-317, 1954.

[0120] <5. Estimating cloud water content and water vapor content using estimation models from radar observation data> The cloud water content and water vapor three-dimensional distribution estimation method, which estimates the three-dimensional distribution of cloud water content and water vapor content using an estimation model from radar observation data and environmental temperature and pressure data, is explained below with reference to Figure 4.

[0121] FIG. 4 is a flowchart showing the process of estimating cloud water content and water vapor content using an estimation model from radar observation data, and more specifically, a flowchart 4000 showing the process of estimating cloud water content and water vapor content using an estimation model from radar observation data.

[0122] By observing the dual-frequency radar, the system obtains one-dimensional distribution data for the radar reflectivity factor of the first frequency, the radar reflectivity factor of the second frequency, the dual-frequency ratio, and the distance from the radar at the observation altitude. At the same time, the system stores the time, azimuth angle, and elevation angle at which the above data was observed (S61 to S62).

[0123] Three-dimensional distribution data of the temperature and pressure in the environmental field is acquired from the Japan Meteorological Agency or the like, and the acquired three-dimensional distribution data of the temperature and pressure is interpolated into the radar coordinate system by one-dimensional data conversion means to convert it into one-dimensional distribution data with respect to the distance from the radar (S63 to S65).

[0124] The cloud water content and water vapor content estimation means inputs the one-dimensional distribution data of the temperature, pressure, first frequency radar reflectivity factor, second frequency radar reflectivity factor, and dual frequency ratio with respect to the distance from the radar into the created estimation model, and estimates the one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar (S66 to S68).

[0125] In addition, in an estimation model in which one-dimensional distribution data for the distance from the radar at the observation altitude is added to the machine learning dataset and the observation altitude is added as an explanatory variable, seasonal changes, etc. can be taken into account in the machine learning by inputting the one-dimensional distribution data for the distance from the radar at the observation altitude into the estimation model.

[0126] Based on the observation data at various azimuth and elevation angles, one-dimensional distribution data of cloud water content and water vapor content at various azimuth and elevation angles versus distance from the radar is estimated and an archive is created (the data is saved) (S69).

[0127] Archived (saved) one-dimensional distribution data of cloud water content and water vapor content at various azimuth and elevation angles versus distance from the radar is interpolated into a three-dimensional grid using a three-dimensional distribution data conversion method, and the three-dimensional distribution of cloud water content and water vapor content is estimated (S70-S71).

[0128] <6. Example of generating data for machine learning using the bin method model> An example of generation for machine learning using the bin method model will be described with reference to FIGS.

[0129] The data handled in this specification is described in a radar coordinate system (one-dimensional distribution with respect to distance from the radar). FIG. 9 is a schematic diagram of a radar coordinate system 9000, showing that a radar P transmits radio waves in a direction with a certain azimuth angle Q and elevation angle R, and acquires a one-dimensional distribution U with respect to the distance from the radar P in that direction. Note that reference symbol T shown in FIG. 9 indicates the distance on the Earth's surface. For example, if the curvature of the Earth is not taken into consideration (for example, under conditions where the curvature of the Earth can be ignored), the following relational expression is obtained. (Distance on the ground surface T) = (Distance from radar P) x cos(Elevation angle R)

[0130] The parameters shown in this figure are plotted assuming an RHI vertical cross section S obtained when the radar antenna is fixed at its azimuth angle and scanned from horizontal to vertical (RHI observation). In other words, the data shown in this figure displays a collection of one-dimensional distributions from horizontal to vertical obtained from observations in one direction.

[0131] Figures 10(a) to 10(h) show the spatial distribution of temperature, pressure, water vapor content, and cloud particle size distribution (corresponding to data in S5), created by assuming that the surrounding area of ​​a cloud simulated using a weather simulation model is observed by a weather radar with an RHI. The horizontal axis of the figure represents the distance from the radar on the ground surface, and the vertical axis represents altitude. Figures 10(a) to 10(c) show the values ​​of pressure, temperature, and water vapor mixing ratio (water vapor content), respectively, interpolated to the radar coordinate system 9000. Figures 10(d) to 10(h) similarly show the particle size distribution. In this example, the number of particles per unit volume with particle sizes of 1 μm, 2 μm, 4 μm, 8 μm, and 16 μm is shown as representative particle size categories. However, because the dynamic range of the number is very wide, the common logarithm of the number is calculated.

[0132] Figure 11(i) shows the spatial distribution of cloud water content calculated from particle size distributions such as those in Figures 10(d) to 10(h) (S7).

[0133] Figures 12(j) to 12(n) show data (S9) calculated by radio wave scattering simulation using the meteorological parameters (S5) shown in Figures 10(a) to 10(h). Figure 12(j) shows radio wave scattering by cloud water. Figures 12(k) and 12(l) show radio wave attenuation of the first frequency (assuming W-band frequencies in this example) and the second frequency (assuming D-band frequencies in this example) by moist air (air with both dry air and water vapor added). Figures 12(m) and 12(n) show radio wave attenuation of the W-band and D-band frequencies by clouds.

[0134] Figures 13(o) to 13(q) show the radar reflectivity factors of the W-band and D-band that would be observed by radar (presumed to be observed by radar) and their two-frequency ratios, calculated using the results of radio wave scattering simulations, radio wave propagation simulations, and radar equations (S11).

[0135] The data used in S15 machine learning are marked with stars, with white stars indicating explanatory variables and black stars indicating target variables.

[0136] 7. Example of Estimating Cloud Water and Water Vapor Using the Estimation Model An example of estimating cloud water content and water vapor content using an estimation model will be described with reference to Fig. 14. Fig. 14 is a diagram showing the results of an example of estimating cloud water content and water vapor content using an estimation model. The horizontal axis of the graph shown in Fig. 14 represents the distance from the radar.

[0137] Figures 14(a) to 14(c) show the distance distribution of radar reflectivity factors at the first and second frequencies observed by dual-frequency weather radar, as well as the distance distribution of the dual-frequency ratio (reference numerals 140a to 140c) as shown in S62. With weather radar, radio waves are scattered only where there are clouds, so values ​​can only be obtained where clouds exist and where the strength of the received radio waves is higher than the radar's minimum receiving sensitivity. Noise is observed in other locations, but such noise data is hidden in the figures above.

[0138] 14(d) and 14(e) show the distance distribution of atmospheric pressure and temperature in the environmental field interpolated into the radar coordinate system shown in S65 (reference numerals 140d and 140e).

[0139] When the distance distributions shown in Figures 14(a) to 14(e) (reference symbols 140a to 140e) are input into an appropriately trained estimation model, the distance distributions of cloud water content and water vapor content shown in Figures 14(f) and 14(g) are output (reference symbols 140f and 140g).

[0140] Although the embodiments of the present invention have been described above, it goes without saying that the present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the gist of the present invention.

[0141] The present invention can have the following configuration. [1] a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factors of two frequencies, and a two-frequency ratio of the two frequencies relative to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; and a cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar.

[0142] [2] 2. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 1, further comprising a cloud water content and water vapor content three-dimensional distribution estimation means for estimating a three-dimensional distribution of cloud water content and water vapor content with respect to distance from the radar based on the one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar estimated by the cloud water content and water vapor content estimation means.

[0143] [3] The cloud-resolving numerical meteorological model is a bin method model or a bulk method model. [1] or [2].

[0144] [4] adding one-dimensional distribution data of the observation altitude with respect to the distance from the radar as the machine learning data; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an input of the estimation model created by the machine learning means; The cloud water content and water vapor content estimation system using a dual-frequency radar described in any one of [1] to [3], wherein the cloud water content and water vapor content estimation means inputs one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimates one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

[0145] [5] a meteorological simulation calculation step of calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation step for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies relative to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation calculation step; a machine learning step of inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to distance from the radar, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar, by performing machine learning on the machine learning data created in the machine learning data creation step as training data; a cloud water content and water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar.

[0146] [6] A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factor of two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; A cloud water content and water vapor content estimation program using a dual-frequency radar, which functions as a cloud water content and water vapor content estimation means that inputs one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and the two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimates one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar.

[0147] [7] A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factor of two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; A computer-readable recording medium having recorded thereon a cloud water content and water vapor content estimation program using a dual-frequency radar, which functions as a cloud water content and water vapor content estimation means that inputs one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and the two-frequency ratio of the two frequencies versus distance from the radar into the estimation model, and estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar.

[0148] [8] Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distributed meteorological data calculated based on a cloud-resolving numerical meteorological model; Machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, dual-frequency radar reflectivity factors, and a two-frequency ratio of the dual frequencies relative to the distance from the radar, which data is created based on the calculated three-dimensional distribution meteorological data; One-dimensional distribution estimation data of cloud water content and water vapor content with respect to distance from the radar, estimated based on one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and dual-frequency ratio with respect to distance from the radar, by machine learning the machine-learning data as teacher data; A data structure for estimating cloud water content and water vapor content using a dual-frequency radar, comprising: one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of a dual frequency, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar; and cloud water content and water vapor content estimation data relating to one-dimensional distribution data of estimated cloud water content and water vapor content with respect to the distance from the radar, based on the one-dimensional distribution estimation data.

[0149] [9] a meteorological simulation means based on the bin method model, which calculates three-dimensional distribution data of temperature, atmospheric pressure, water vapor content, and cloud particle size distribution based on the bin method model; one-dimensional distribution data conversion means for interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation means based on the bin method model into a radar coordinate system and converting it into one-dimensional distribution data relative to the distance from the radar; cloud water content calculation means for calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted by the one-dimensional distribution data conversion means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted by the one-dimensional distribution data converting means, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated by the cloud water content calculating means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0150]

[10] The radar reflectivity factor / dual frequency ratio calculation means comprises a radio wave scattering simulation means for calculating the attenuation and scattering of radio waves, and a radio wave propagation simulation means for calculating the propagation of the radio waves. [9] A system for estimating cloud water content and water vapor content using a dual frequency radar.

[0151]

[11] adding one-dimensional distribution data of the observation altitude with respect to the distance from the radar as data included in the data set; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an explanatory variable in the machine learning means; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an input of the estimation model created by the machine learning means; The cloud water content and water vapor content estimation system using a dual-frequency radar described in [9] or

[10] , wherein the cloud water content and water vapor content estimation means inputs one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model, and estimates one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

[0152]

[12] a meteorological simulation step based on the bin method model, which calculates three-dimensional distribution data of temperature, pressure, water vapor content, and cloud particle size distribution based on the bin method model; a one-dimensional distribution data conversion step of interpolating the three-dimensional distribution data of the temperature, pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation step based on the bin method model into a radar coordinate system and converting the data into one-dimensional distribution data relative to the distance from the radar; a cloud water content calculation step of calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted in the one-dimensional distribution data conversion step; a radar reflectivity factor / two-frequency ratio calculation step of calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted in the one-dimensional distribution data conversion step; a data set creation step of creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted in the one-dimensional distribution data conversion step, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated in the cloud water content calculation step, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies calculated in the radar reflectivity factor / two-frequency ratio calculation step versus distance from the radar; a machine learning step of using the dataset created in the dataset creation step as training data, and using temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; a cloud water content and water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor with respect to distance from the radar; and a three-dimensional distribution data conversion step of interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into a three-dimensional grid to convert it into three-dimensional distribution data of the cloud water content and water vapor content.

[0153]

[13] A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on the bin method model, which calculates three-dimensional distribution data of temperature, atmospheric pressure, water vapor content, and cloud particle size distribution based on the bin method model; one-dimensional distribution data conversion means for interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation means based on the bin method model into a radar coordinate system and converting it into one-dimensional distribution data relative to the distance from the radar; cloud water content calculation means for calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted by the one-dimensional distribution data conversion means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted by the one-dimensional distribution data converting means, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated by the cloud water content calculating means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content relative to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0154]

[14] A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on the bin method model, which calculates three-dimensional distribution data of temperature, atmospheric pressure, water vapor content, and cloud particle size distribution based on the bin method model; one-dimensional distribution data conversion means for interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation means based on the bin method model into a radar coordinate system and converting it into one-dimensional distribution data relative to the distance from the radar; cloud water content calculation means for calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted by the one-dimensional distribution data conversion means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted by the one-dimensional distribution data converting means, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated by the cloud water content calculating means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0155]

[15] Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distribution data of temperature, pressure, water vapor content, and cloud particle size distribution calculated based on the bottle method model, one-dimensional distribution data obtained by interpolating the three-dimensional distribution data of the calculated temperature, atmospheric pressure, water vapor content, and cloud particle size distribution into a radar coordinate system and converting it into a one-dimensional distribution with respect to the distance from the radar; and one-dimensional distribution data of cloud liquid water volume with respect to distance from the radar calculated based on the one-dimensional distribution data of the converted cloud particle size distribution with respect to distance from the radar; one-dimensional distribution data of a radar reflectivity factor of a dual frequency and a two-frequency ratio of the dual frequencies with respect to the distance from the radar, calculated based on the one-dimensional distribution data of the converted air temperature, air pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar; and a data set including the one-dimensional distribution data of the converted air temperature, air pressure, and water vapor content versus distance from the radar, the one-dimensional distribution data of the calculated cloud water content versus distance from the radar, and the one-dimensional distribution data of the calculated dual-frequency radar reflectivity factor and dual-frequency ratio of the dual-frequency versus distance from the radar; One-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is estimated based on one-dimensional distribution data of the temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies with the data set as training data, and one-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is obtained by machine learning using the data set as training data, and using temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables; cloud water content and water vapor content estimation data relating to the one-dimensional distribution data of cloud water content and water vapor content estimated based on one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar, and the one-dimensional distribution estimation data; A data structure for estimating cloud water content and water vapor content using a dual-frequency radar, which includes the cloud water content and water vapor content estimated data interpolated and processed into a three-dimensional grid, and three-dimensional distribution data of the converted cloud water content and water vapor content.

[0156]

[16] a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; cloud particle size distribution calculation means for calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of cloud water amount with respect to distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated by the cloud particle size distribution calculation means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0157]

[17] The radar reflectivity factor / dual frequency ratio calculation means comprises a radio wave scattering simulation means for calculating the attenuation and scattering of radio waves, and a radio wave propagation simulation means for calculating the propagation of the radio waves.

[16] A system for estimating cloud water content and water vapor content using a dual frequency radar.

[0158]

[18] adding one-dimensional distribution data of the observation altitude with respect to the distance from the radar as data included in the data set; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an explanatory variable in the machine learning means; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an input of the estimation model created by the machine learning means; The cloud water content and water vapor content estimation system using a dual-frequency radar described in

[16] or

[17] , wherein the cloud water content and water vapor content estimation means inputs one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model to estimate one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

[0159]

[19] a meteorological simulation step based on a bulk method model, which calculates three-dimensional distribution data of temperature, pressure, cloud water content, and water vapor content based on the bulk method model; a one-dimensional distribution data conversion step of interpolating the three-dimensional distribution data of the temperature, pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation step based on the bulk method model into a radar coordinate system and converting the data into one-dimensional distribution data relative to the distance from the radar; a cloud particle size distribution calculation step of calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of the cloud water amount with respect to distance from the radar converted in the one-dimensional distribution data conversion step; a radar reflectivity factor / two-frequency ratio calculation step of calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted in the one-dimensional distribution data conversion step and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated in the cloud particle size distribution calculation step; a data set creation step of creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud liquid water content versus distance from the radar converted in the one-dimensional distribution data conversion step, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated in the radar reflectivity factor / two-frequency ratio calculation step; a machine learning step of using the dataset created in the dataset creation step as training data, and using temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; a cloud water content and water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor with respect to distance from the radar; and a three-dimensional distribution data conversion step of interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into a three-dimensional grid to convert it into three-dimensional distribution data of the cloud water content and water vapor content.

[0160]

[20] A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; cloud particle size distribution calculation means for calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of cloud water amount with respect to distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated by the cloud particle size distribution calculation means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content relative to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0161] [twenty one] A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; cloud particle size distribution calculation means for calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of cloud water amount with respect to distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated by the cloud particle size distribution calculation means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0162] [twenty two] Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distribution data of temperature, pressure, cloud water content, and water vapor content calculated based on the bulk method model, one-dimensional distribution data obtained by interpolating the three-dimensional distribution data of the calculated temperature, atmospheric pressure, water vapor amount, and cloud water content into a radar coordinate system and converting it into one-dimensional distribution data with respect to the distance from the radar; and one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar, calculated by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount, based on the one-dimensional distribution data of the converted cloud water amount with respect to distance from the radar; and one-dimensional distribution data of a radar reflectivity factor of a dual frequency and a two-frequency ratio of the dual frequencies with respect to the distance from the radar, calculated based on the one-dimensional distribution data of the converted air temperature, air pressure, and water vapor content with respect to the distance from the radar and the one-dimensional distribution data of the calculated cloud particle size distribution with respect to the distance from the radar; a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud liquid water content versus distance from the radar converted by the one-dimensional distribution data conversion means, and the one-dimensional distribution data of the calculated radar reflectivity factor of the dual frequencies and the dual-frequency ratio of the dual frequencies versus distance from the radar; One-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is estimated based on one-dimensional distribution data of the temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies with the data set as training data, and one-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is obtained by machine learning using the data set as training data, and using temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables; cloud water content and water vapor content estimated data relating to one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar, which is estimated based on one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar; A data structure for estimating cloud water content and water vapor content using a dual-frequency radar, which includes the cloud water content and water vapor content estimated data interpolated and processed into a three-dimensional grid, and three-dimensional distribution data of the converted cloud water content and water vapor content.

[0163] [twenty three] a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0164] [twenty four] The radar reflectivity factor / dual frequency ratio calculation means comprises a radio wave scattering simulation means for calculating the attenuation and scattering of radio waves, and a radio wave propagation simulation means for calculating the propagation of the radio waves.

[23] A system for estimating cloud water content and water vapor content using a dual frequency radar.

[0165] [twenty five] adding one-dimensional distribution data of the observation altitude with respect to the distance from the radar as data included in the data set; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an explanatory variable in the machine learning means; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an input of the estimation model created by the machine learning means; The cloud water content and water vapor content estimation system using a dual-frequency radar described in

[23] or

[24] , wherein the cloud water content and water vapor content estimation means inputs one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and a two-frequency ratio of the two frequencies with respect to the distance from the radar into the estimation model to estimate one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

[0166]

[26] a meteorological simulation step based on a bulk method model, which calculates three-dimensional distribution data of temperature, pressure, cloud water content, and water vapor content based on the bulk method model; a one-dimensional distribution data conversion step of interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation step based on the bulk method model into a radar coordinate system and converting the data into one-dimensional distribution data relative to the distance from the radar; a radar reflectivity factor / two-frequency ratio calculation step of calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted in the one-dimensional distribution data conversion step, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creation step of creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud liquid water content versus distance from the radar converted in the one-dimensional distribution data conversion step, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated in the radar reflectivity factor / two-frequency ratio calculation step; a machine learning step of using the dataset created in the dataset creation step as training data, and using temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; a cloud water content and water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor with respect to distance from the radar; and a three-dimensional distribution data conversion step of interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into a three-dimensional grid to convert it into three-dimensional distribution data of the cloud water content and water vapor content.

[0167]

[27] A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content relative to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

[0168]

[28] A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data converting means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculating means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content.

[0169]

[29] Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distribution data of temperature, pressure, cloud water content, and water vapor content calculated based on the bulk method model, one-dimensional distribution data obtained by interpolating the three-dimensional distribution data of the calculated temperature, atmospheric pressure, water vapor amount, and cloud water content into a radar coordinate system and converting it into one-dimensional distribution data with respect to the distance from the radar; and one-dimensional distribution data of the converted temperature, atmospheric pressure, water vapor amount, and cloud water content versus distance from the radar; and one-dimensional distribution data of the radar reflectivity factor of dual frequencies and the dual-frequency ratio of the dual frequencies versus distance from the radar, calculated based on empirical relations regarding temperature, atmospheric pressure, cloud water content, water vapor amount, and radio wave attenuation and scattering. a data set including one-dimensional distribution data of the converted air temperature, air pressure, water vapor content, and cloud liquid water content with respect to the distance from the radar, and one-dimensional distribution data of the calculated radar reflectivity factor of the dual frequencies and the dual-frequency ratio of the dual frequencies with respect to the distance from the radar; One-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is estimated based on one-dimensional distribution data of the temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies with the data set as training data, and one-dimensional distribution estimation data of the cloud water content and water vapor content with respect to the distance from the radar, which is obtained by machine learning using the data set as training data, and using temperature, pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables; cloud water content and water vapor content estimated data relating to one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar, which is estimated based on one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar; and three-dimensional data obtained by interpolating the cloud water content and water vapor content estimated data into a three-dimensional grid and converting it into three-dimensional distribution data of cloud water content and water vapor content. [Explanation of symbols]

[0170] Flowchart for creating an estimation model using the 1000···bin method model. 2000···Flowchart for creating an estimation model using the bulk method model (assuming particle size distribution) 3000···Flowchart for creating an estimation model using the bulk method model (using empirical formulas) 4000... Flowchart for estimating cloud water content and water vapor content using an estimation model from radar observation data. 5000....A system for estimating cloud water content and water vapor content using dual-frequency radar, 6000....Dual-frequency radar cloud water content and water vapor content estimation system, 7000....A flowchart showing the relationship between meteorological parameters and parameters observed and analyzed by radar; 8000···Means for creating estimation models, 9000···Radar coordinate system.

Claims

1. a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factors of two frequencies, and a two-frequency ratio of the two frequencies relative to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; and a cloud water content / water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and a two-frequency ratio of the two frequencies versus distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar.

2. 2. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 1, further comprising a cloud water content and water vapor content three-dimensional distribution estimation means for estimating a three-dimensional distribution of the cloud water content and water vapor content with respect to the distance from the radar, based on the one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar estimated by the cloud water content and water vapor content estimation means.

3. 2. The system for estimating cloud water content and water vapor content using a dual-frequency radar according to claim 1, wherein the cloud-resolving numerical meteorological model is a bin method model or a bulk method model.

4. adding one-dimensional distribution data of the observation altitude with respect to the distance from the radar as the machine learning data; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an input of the estimation model created by the machine learning means; 4. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 1, wherein the cloud water content and water vapor content estimation means inputs one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, a dual-frequency radar reflectivity factor, and a two-frequency ratio of the dual frequencies with respect to the distance from the radar into the estimation model, and estimates one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

5. a meteorological simulation calculation step of calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation step for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies relative to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation calculation step; a machine learning step of inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to distance from the radar, and creating an estimation model that estimates one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar, by performing machine learning on the machine learning data created in the machine learning data creation step as training data; a cloud water content / water vapor content estimation step of inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and a two-frequency ratio of the two frequencies versus distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar.

6. A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factor of two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; A cloud water content and water vapor content estimation program using a dual-frequency radar, which functions as a cloud water content and water vapor content estimation means that inputs one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and the two-frequency ratio of the two frequencies versus distance from the radar into the estimation model, and estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar.

7. A computer in a system for estimating cloud water content and water vapor content using a dual-frequency radar, a meteorological simulation means for calculating three-dimensional distribution meteorological data based on a cloud-resolving numerical meteorological model; a machine learning data creation means for creating machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, radar reflectivity factor of two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the three-dimensional distribution meteorological data calculated by the meteorological simulation means; a machine learning means for inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factors, and a dual-frequency ratio of the dual frequencies with respect to the distance from the radar, and creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content with respect to the distance from the radar, by performing machine learning on the machine learning data created by the machine learning data creation means as training data; A computer-readable recording medium having recorded thereon a cloud water content and water vapor content estimation program using a dual-frequency radar, which functions as a cloud water content and water vapor content estimation means that inputs one-dimensional distribution data of predetermined temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and the two-frequency ratio of the two frequencies versus distance from the radar into the estimation model, and estimates one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar.

8. Used by a computer in a system for estimating cloud water content and water vapor content using dual-frequency radar, Three-dimensional distributed meteorological data calculated based on a cloud-resolving numerical meteorological model; Machine learning data including one-dimensional distribution data of temperature, atmospheric pressure, cloud water content, water vapor content, dual-frequency radar reflectivity factors, and a two-frequency ratio of the dual frequencies relative to the distance from the radar, which data is created based on the calculated three-dimensional distribution meteorological data; One-dimensional distribution estimation data of cloud water content and water vapor content with respect to distance from the radar, estimated based on one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor, and dual-frequency ratio with respect to distance from the radar, by machine learning the machine-learning data as teacher data; A data structure for estimating cloud water content and water vapor content using a dual-frequency radar, comprising: one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of two frequencies, and a two-frequency ratio of the two frequencies with respect to the distance from the radar; and cloud water content and water vapor content estimation data relating to one-dimensional distribution data of estimated cloud water content and water vapor content with respect to the distance from the radar, based on the one-dimensional distribution estimation data.

9. a meteorological simulation means based on the bin method model, which calculates three-dimensional distribution data of temperature, atmospheric pressure, water vapor content, and cloud particle size distribution based on the bin method model; one-dimensional distribution data conversion means for interpolating the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution calculated by the meteorological simulation means based on the bin method model into a radar coordinate system and converting it into one-dimensional distribution data relative to the distance from the radar; cloud water content calculation means for calculating one-dimensional distribution data of cloud liquid water versus distance from the radar based on the one-dimensional distribution data of the cloud particle size distribution versus distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud particle size distribution with respect to the distance from the radar converted by the one-dimensional distribution data conversion means; a data set creating means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content versus distance from the radar converted by the one-dimensional distribution data converting means, the one-dimensional distribution data of the cloud water content versus distance from the radar calculated by the cloud water content calculating means, and the one-dimensional distribution data of the radar reflectivity factor of the dual frequencies and the dual frequency ratio of the dual frequencies calculated by the radar reflectivity factor / dual frequency ratio calculating means versus distance from the radar; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of the two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

10. 10. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 9, wherein the radar reflectivity factor / dual-frequency ratio calculation means comprises: a radio wave scattering simulation means for calculating attenuation and scattering of radio waves; and a radio wave propagation simulation means for calculating propagation of the radio waves.

11. adding one-dimensional distribution data of the observation altitude with respect to the distance from the radar as data included in the data set; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an explanatory variable in the machine learning means; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an input of the estimation model created by the machine learning means; 11. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 9 or 10, wherein the cloud water content and water vapor content estimation means inputs one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, radar reflectivity factor of the dual frequencies, and a two-frequency ratio of the dual frequencies with respect to the distance from the radar into the estimation model, and estimates one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

12. a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, water vapor amount, and cloud liquid water amount calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; cloud particle size distribution calculation means for calculating one-dimensional distribution data of cloud particle size distribution with respect to distance from the radar by assuming a particle size distribution of clouds having substantially the same amount of cloud water as the converted cloud water amount based on the one-dimensional distribution data of cloud water amount with respect to distance from the radar converted by the one-dimensional distribution data conversion means; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of a dual frequency and a two-frequency ratio of the dual frequency with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means and the one-dimensional distribution data of the cloud particle size distribution with respect to the distance from the radar calculated by the cloud particle size distribution calculation means; a data set creation means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data conversion means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculation means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of the two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

13. 13. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 12, wherein the radar reflectivity factor / dual-frequency ratio calculation means comprises: a radio wave scattering simulation means for calculating attenuation and scattering of radio waves; and a radio wave propagation simulation means for calculating propagation of the radio waves.

14. adding one-dimensional distribution data of the observation altitude with respect to the distance from the radar as data included in the data set; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an explanatory variable in the machine learning means; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an input of the estimation model created by the machine learning means; 14. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 12 or 13, wherein the cloud water content and water vapor content estimation means inputs one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, a dual-frequency radar reflectivity factor, and a two-frequency ratio of the dual frequencies with respect to the distance from the radar into the estimation model, and estimates one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

15. a meteorological simulation means based on a bulk method model for calculating three-dimensional distribution data of temperature, atmospheric pressure, cloud water content, and water vapor content based on the bulk method model; one-dimensional distribution data conversion means for converting the three-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content calculated by the meteorological simulation means based on the bulk method model into one-dimensional distribution data relative to the distance from the radar by interpolating the data into a radar coordinate system; a radar reflectivity factor / two-frequency ratio calculation means for calculating one-dimensional distribution data of a radar reflectivity factor of two frequencies and a two-frequency ratio of the two frequencies with respect to the distance from the radar, based on the one-dimensional distribution data of the temperature, atmospheric pressure, cloud water content, and water vapor content with respect to the distance from the radar converted by the one-dimensional distribution data conversion means, and empirical relational expressions regarding the temperature, atmospheric pressure, cloud water content, water vapor content, and radio wave attenuation and scattering; a data set creation means for creating a data set including the one-dimensional distribution data of the temperature, atmospheric pressure, water vapor content, and cloud water content versus distance from the radar converted by the one-dimensional distribution data conversion means, and the one-dimensional distribution data of the radar reflectivity factors of the two frequencies and the two-frequency ratio of the two frequencies versus distance from the radar calculated by the radar reflectivity factor / two-frequency ratio calculation means; a machine learning means for performing machine learning using the dataset created by the dataset creation means as training data, temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies as explanatory variables, and cloud water content and water vapor content as objective variables, and for creating an estimation model for estimating one-dimensional distribution data of cloud water content and water vapor content versus distance from the radar by inputting one-dimensional distribution data of temperature, atmospheric pressure, dual-frequency radar reflectivity factor and dual-frequency ratio of the dual frequencies versus distance from the radar; cloud water content and water vapor content estimation means for inputting one-dimensional distribution data of predetermined temperature, atmospheric pressure, radar reflectivity factor of the two frequencies, and two-frequency ratio of the two frequencies with respect to distance from the radar into the estimation model, and estimating one-dimensional distribution data of cloud water content and water vapor content with respect to distance from the radar; and a three-dimensional distribution data conversion means for converting the one-dimensional distribution data of the estimated cloud water content and water vapor content with respect to the distance from the radar into three-dimensional distribution data of the cloud water content and water vapor content by interpolating the data into a three-dimensional grid.

16. 16. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 15, wherein the radar reflectivity factor / dual-frequency ratio calculation means comprises: a radio wave scattering simulation means for calculating attenuation and scattering of radio waves; and a radio wave propagation simulation means for calculating propagation of the radio waves.

17. adding one-dimensional distribution data of the observation altitude with respect to the distance from the radar as data included in the data set; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an explanatory variable in the machine learning means; adding the one-dimensional distribution data with respect to the distance from the radar at the observation altitude as an input of the estimation model created by the machine learning means; 17. The cloud water content and water vapor content estimation system using a dual-frequency radar according to claim 15 or 16, wherein the cloud water content and water vapor content estimation means inputs one-dimensional distribution data of a predetermined observation altitude, temperature, atmospheric pressure, radar reflectivity factor of the dual frequencies, and a two-frequency ratio of the dual frequencies with respect to the distance from the radar into the estimation model, and estimates one-dimensional distribution data of the cloud water content and water vapor content with respect to the distance from the radar.

Images