Rainfall prediction device and rainfall prediction method

Abstract

The purpose of the present disclosure is to provide a rainfall prediction device and rainfall prediction method that are capable of predicting rainfall including newly occurring rainfall. A rainfall prediction device according to the present disclosure acquires, from a measurement machine that measures an index indicating the water vapor amount in the atmosphere, measurement values of the index. The rainfall prediction device predicts, on the basis of the variation trends of the water vapor amount indicated by the time-series data of the measurement values, the occurrence time of rainfall that may occur in the future.

Description

Rainfall prediction device and rainfall prediction method

[0001] This disclosure relates to a rainfall prediction device and a rainfall prediction method.

[0002] In recent years, mobile communication systems have advanced, making mobile services available in most parts of the world. Furthermore, ultra-high coverage is one of the requirements for the fifth-generation (Beyond 5G) or sixth-generation mobile communication systems that are expected to be commercialized in the future.

[0003] Ultra-coverage refers to expanding service areas to locations where the cost of laying existing base stations is high or difficult, such as mountainous areas, at sea, and in the air. Furthermore, national resilience against natural disasters is also needed, and the emergence of communication systems that are robust against ground-based disasters is desired.

[0004] To realize such wireless communication systems, NTN (Non-Terrestrial Network), which uses aerial relay stations such as high-altitude pseudo-satellites, communication satellites, drones, and unmanned aerial vehicles, is attracting attention.

[0005] NTN anticipates the use of high-frequency band (millimeter waves, etc.) wireless signals to expand communication capacity, but atmospheric attenuation of wireless signals due to rainfall is significant in the high-frequency band. Therefore, proactive communication control utilizing rainfall forecast data is required.

[0006] Non-patent document 1 discloses a site diversity technique in which an aerial relay station switches to a connected ground antenna in advance to avoid rainfall based on rainfall forecast data. In this technique, high-resolution precipitation nowcast data provided by the Japan Meteorological Agency is used as the rainfall forecast data.

[0007] Kitazono, Nobuaki; Suzuki, Jun; Sotozono, Yuki; Kishiyama, Yoshihisa; Fukazawa, Kenji; 'Development of a 38GHz Band Wireless Communication System in Collaboration with a 5G Network using a High Altitude Platform (HAPS) - Development of Proactive Site Diversity -'; 2023 IEICE General Conference, B-3-21, March 2023.

[0008] However, conventional rainfall forecasting techniques, such as high-resolution precipitation nowcasting, are limited to predicting the movement of existing rainfall areas and do not address the prediction of newly formed rainfall areas.

[0009] This disclosure aims to provide a rainfall forecasting device and method that can predict rainfall, including newly occurring rainfall, in order to solve the above-mentioned problems.

[0010] A first aspect of this disclosure is preferably a rainfall forecasting device configured to perform the following: a process of acquiring measured values ​​of an index indicating the amount of water vapor in the atmosphere from a measuring instrument; and a prediction process of predicting the time of rainfall that may occur in the future based on the trend of change in the amount of water vapor indicated by the time-series data of the measured values.

[0011] Furthermore, a second embodiment is preferably a rainfall prediction method that includes: obtaining measured values ​​of an index indicating the amount of water vapor in the atmosphere from a measuring instrument; and predicting the time of occurrence of rainfall that may occur in the future based on the trend of change in the amount of water vapor indicated by the time-series data of the measured values.

[0012] According to the first and second aspects of this disclosure, it is possible to provide a rainfall forecasting device and a rainfall forecasting method that can forecast rainfall, including newly occurring rainfall.

[0013] This is a diagram illustrating the conventional rainfall forecasting technology related to the comparative example. This is a diagram showing the measured values ​​of atmospheric water vapor amount acquired by the rainfall forecasting device according to Embodiment 1 in a time series. This is a pattern in which no trigger occurs according to Embodiment 1. This is time series data of the rate of increase of water vapor amount calculated by the rainfall forecasting device according to Embodiment 1. This is an example of the configuration of the rainfall forecasting device according to Embodiment 1. This is a diagram showing the hardware configuration of the rainfall forecasting device according to Embodiment 1. This is a flowchart illustrating the processing performed by the CPU of the rainfall forecasting device according to Embodiment 1.

[0014] <Comparative Example> Here, we will first explain in detail the challenges faced by conventional technology.

[0015] FIG. 1 is a diagram for explaining a conventional technique of rainfall prediction according to a comparative example. On the left side of FIG. 1, the position P1 of the rain cloud 10 at a certain point in time and the position P2 of the rain cloud 10 predicted to move after n minutes from a certain point in time are shown. On the right side of FIG. 1, the state in which the rain cloud 10 moves from the position P1 to the position P2 after n minutes from a certain point in time is shown. Conventional techniques of rainfall prediction such as high-resolution precipitation nowcast predict the movement of such a rain cloud 10.

[0016] In the conventional rainfall prediction, it only predicts the movement of the existing rainfall area and does not deal with the prediction of a newly generated rainfall area. Therefore, even though the movement of the existing rain cloud 10 can be predicted, the generation of a new rain cloud 20 cannot be predicted.

[0017] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. The same or corresponding components may be denoted by the same reference numerals, and the repeated description may be omitted.

[0018] Embodiment 1 The rainfall prediction method of the present disclosure includes the following.

[0019] 1. Determination of rainfall prediction start trigger FIG. 2 is a diagram showing the measured values of the amount of water vapor in the atmosphere acquired by the rainfall prediction device 100 according to Embodiment 1 in time series. The horizontal axis represents time, and the vertical axis represents the amount of water vapor. The rainfall prediction device 100 is capable of acquiring the measured value of the real-time amount of water vapor in the atmosphere.

[0020] As the rainfall prediction device 100 acquires the measured value of the amount of water vapor, it determines whether the amount of water vapor exceeds a preset first threshold value V １ . The first threshold value V １ is set to a value lower than the amount of water vapor corresponding to the state in which the rain cloud 10 exists. As an example, the first threshold value V １ is 4.5 g / m ３ .

[0021] When it is recognized that the amount of water vapor exceeds the first threshold value V １ , the rainfall prediction device 100 starts timing from the time t １ when the amount of water vapor exceeds the first threshold value V １ . Further, the rainfall prediction device 100 uses the time t １For the amount of water vapor measured until time m has elapsed, it is determined whether the amount of water vapor exceeds the second threshold value V ２ . The second threshold value V ２ is set to a value lower than the first threshold value V １ (V ２ < V １ ). As an example, the second threshold value V ２ is 4.0 g / m ３ . Also, as an example, the time m is 5 minutes.

[0022] At time t １ , when the amount of water vapor does not fall below the second threshold value V ２ during the time m, the rainfall prediction device 100 starts rainfall prediction. For example, as shown in FIG. 2, when the amount of water vapor has an increasing trend after time t １ , the trigger for rainfall prediction occurs because the amount of water vapor does not fall below the second threshold value V ２ .

[0023] On the other hand, FIG. 3 shows a pattern in which no trigger occurs. In FIG. 3, the amount of water vapor fluctuates around the first threshold value V １ , and the amount of water vapor falls below the second threshold value V ２ before the time m elapses from time t １ . In this case, since it can be said that the amount of water vapor has a decreasing trend at time m, the rainfall prediction device 100 does not start rainfall prediction. <​​​​​​​​​​​​

[0027] 2. Time of rainfall t ２ As the prediction trigger occurs, the rainfall prediction device 100 analyzes the trend of change in water vapor amount shown in the acquired time-series data of water vapor amount (hereinafter, time-series data) to determine the time of occurrence of rainfall that may occur in the future. ２ Predict (predictive processing).

[0028] Time of rainfall t ２ As an example of a prediction method, the rainfall prediction device 100 has a database in which multiple time-series data measured before and after past rainfall events and the actual rainfall occurrence times corresponding to each of the past time-series data are stored. The rainfall prediction device 100 identifies the past time-series data that has the closest trend of change to the time-series data acquired from the database. Furthermore, the rainfall prediction device 100 identifies the rainfall occurrence time of the time-series data acquired based on the rainfall occurrence time corresponding to the identified past time-series data. ２ To decide.

[0029] 3. Precipitation forecast Figure 4 shows time-series data of the rate of increase of water vapor amount calculated by the rainfall forecasting device according to Embodiment 1. The horizontal axis is time, and the vertical axis is the rate of increase of water vapor amount. The rainfall forecasting device 100 is triggered at time t １ The rainfall prediction device 100 calculates the rate of increase of water vapor (time derivative, etc.) based on the amount of water vapor measured from +m until time n has elapsed. By analyzing the trend of change in the rate of increase of water vapor, the rainfall prediction device 100 determines the time of rainfall occurrence t ２ Predict future precipitation. For example, time n is 30 minutes.

[0030] As an example of a precipitation forecasting method, the rainfall forecasting device 100 calculates the average precipitation (e.g., 6 mm / h) corresponding to the water vapor increase rate (e.g., 0.15) using a known linear relationship between the maximum value of the water vapor increase rate (e.g., 0.15) and the average precipitation. Alternatively, the rainfall forecasting device 100 stores a table relating the maximum value of the water vapor increase rate to the average precipitation, and determines the precipitation corresponding to the water vapor increase rate based on this table.

[0031] Thus, in this embodiment, rainfall prediction can be made based on the amount of water vapor in the atmosphere. Therefore, the formation of new rain clouds 20 can be predicted earlier than in the conventional technology.

[0032] Figure 5 shows an example configuration of the rainfall prediction device 100 according to Embodiment 1. The rainfall prediction device 100 is capable of communicating with an external measuring instrument 200. The measuring instrument 200 can be any measuring instrument capable of continuously measuring the amount of water vapor in the atmosphere, such as a microwave radiometer, lidar, weather satellite, or weather radar.

[0033] The determination circuit 1 acquires real-time water vapor quantity measurements from the measuring device 200. The determination circuit 1 applies a first threshold V to the water vapor quantity measurement. １ and the second threshold V ２ The system determines whether a rainfall forecast start trigger has occurred by performing a two-stage threshold determination using the specified method. If a trigger occurs, the determination circuit 1 notifies the first analysis circuit 2 and the second analysis circuit 3 that rainfall forecasting has started.

[0034] Upon receiving notification of the start of rainfall forecasting, the first analysis circuit 2 analyzes the time-series data obtained from the judgment circuit 1 using statistical or machine learning methods to determine the time of rainfall occurrence t ２ The first analysis circuit 2 predicts the rainfall occurrence time t. ２ The prediction result is notified to the communication device 4.

[0035] Furthermore, machine learning can determine the timing of rainfall occurrence t ２ When calculating the rainfall time t, the first analysis circuit 2 further includes a first trained model. The first analysis circuit 2 inputs time series data into the first trained model, thereby determining the rainfall time t in the first trained model. ２ The output is generated. The first trained model determines the rainfall occurrence time t. ２ The rainfall occurrence time t is output based on the training time series data for which the known values ​​are t. ２ This is a pre-trained model designed to approximate the correct answer. While known machine learning methods such as Random Forest, SVM (Support Vector Machine), K-Nearest Neighbors, and Neural Networks are examples, they are not limited to this method.

[0036] Upon receiving the rainfall forecast start notification, the second analysis circuit 3 determines the trigger time t １ Time series data is acquired from +m until time n has elapsed. Furthermore, the second analysis circuit 3 calculates the rate of increase of water vapor using the acquired time series data. The rainfall prediction device 100 analyzes the trend of change in the rate of increase of water vapor using statistics or machine learning to determine the time of rainfall t ２ The second analysis circuit 3 predicts the amount of precipitation from that point onward. The second analysis circuit 3 notifies the communication device 4 of the precipitation prediction results.

[0037] Furthermore, when calculating precipitation using machine learning, the second analysis circuit 3 is equipped with a second pre-trained model. The second analysis circuit 3 inputs data on the rate of increase of water vapor into the second pre-trained model, causing the second pre-trained model to output precipitation. The second pre-trained model is a pre-trained model that has been trained so that the output precipitation amount approaches the correct answer based on training data on the rate of increase of water vapor, for which precipitation is known.

[0038] Communication device 4 records the time of rainfall t ２ The precipitation forecast results are then transmitted to the external device 300.

[0039] Thus, the rainfall prediction device 100 of this embodiment predicts rainfall using the water vapor amount measured by the measuring device 200. By installing the measuring device 200, it is possible to predict the occurrence of rainfall in a desired location.

[0040] In this embodiment, the rainfall prediction method does not necessarily have to be based on the amount of water vapor in the atmosphere; rainfall prediction may also be based on the optical thickness of the atmosphere, etc. Optical thickness is a physical quantity that indicates the amount of absorption when electromagnetic waves propagate through the atmosphere, and can be calculated by combining the amount of water vapor with physical quantities such as temperature. In this embodiment, rainfall prediction is possible by using an index that indicates the amount of water vapor in the atmosphere.

[0041] Figure 6 shows the hardware configuration of the rainfall forecasting device 100 according to Embodiment 1. The processing performed by the rainfall forecasting device 100 may be executed by a program using a computer equipped with a CPU and memory, in which a rainfall forecasting program is stored. Alternatively, it may be executed by a program using an integrated circuit such as an FPGA (Field Programmable Gate Array). The rainfall forecasting program may be provided by recording it on a storage medium or by providing it via a network.

[0042] The rainfall forecasting device 100 has an input unit 40, an output unit 41, a communication unit 42, a CPU (Central Processing Unit, also called a processor) 43, a memory 44, and an HDD (Hard Disk Drive) 45 connected via a bus 46, and functions as a computer. The rainfall forecasting device 100 is also configured to input and output data to and from a storage medium 47 that can be read by a computer.

[0043] The input unit 40 is, for example, a keyboard and mouse. The output unit 41 is, for example, a display device such as a display.

[0044] The communication unit 42 is a communication interface that communicates with, for example, the measuring instrument 200 and the external device 300.

[0045] Memory 44 refers to volatile or non-volatile semiconductor memory such as RAM, ROM, and flash memory, or magnetic disks, flexible disks, optical disks, and DVDs.

[0046] The CPU 43 controls each component of the rainfall forecasting device 100 and performs predetermined processing. The memory 44 and HDD 45 are storage devices that store, for example, time-series data acquired from the measuring device 200 and rainfall forecasting programs.

[0047] The storage medium 47 is capable of storing rainfall prediction programs and the like that which enable the rainfall prediction device 100 to perform its functions. The storage medium 47 can be a USB (Universal Serial Bus) memory, a CD-ROM (Compact Disc Read Only Memory), or the like.

[0048] Note that the architecture of the rainfall prediction device 100 is not limited to the example shown in the figure.

[0049] Figure 7 is a flowchart illustrating the process performed by the CPU 43 of the rainfall prediction device 100 according to Embodiment 1. The CPU 43 reads the rainfall prediction program stored in the memory 44 or HDD 45 and executes the following processes.

[0050] First, real-time water vapor quantity measurements are obtained from the measuring device 200 (step S01). Next, the water vapor quantity is set to a preset first threshold V. １ Determine if it exceeds the first threshold V (step S02). １ If it is not determined that the limit has been exceeded (No), return to step S01 and continue monitoring the amount of water vapor.

[0051] On the other hand, the amount of water vapor is the first threshold V １ If it is determined that the amount of water vapor exceeds the first threshold V (Yes), the amount of water vapor is determined to be the first threshold V １ Time t exceeding １ Between time m, the amount of water vapor is the second threshold V. ２ Determine if the amount of water vapor did not fall below the second threshold V (step S03). ２ If the value falls below this threshold (No), it means that no trigger will occur. In this case, return to step S01 and continue monitoring the amount of water vapor.

[0052] On the other hand, the amount of water vapor is the second threshold V ２ If it is determined that the water vapor content did not fall below the threshold (Yes), it means that a trigger occurred. In this case, by analyzing the trend of water vapor changes shown in the time series data, the time of future rainfall t can be determined. ２ Predict (Step S04).

[0053] Next, the time t when the trigger occurred. １The rate of increase of water vapor is calculated using the amount of water vapor measured from +m until time n has elapsed (step S05). Furthermore, by analyzing the trend of change in the rate of increase of water vapor, the time of rainfall t is determined. ２ Predict the amount of rainfall from now on (step S06). Furthermore, predict the time t when the rainfall will occur. ２ The precipitation forecast results are output to the external device 300 (step S07). Finally, the process ends.

[0054] As explained above, in this embodiment, it is possible to predict rainfall including newly occurring rainfall.

[0055] This disclosure is not limited to the embodiments described above, and various modifications can be made during implementation without departing from its essence. Furthermore, each embodiment and its modifications may be combined as appropriate, and in that case, the combined effects can be obtained.

[0056] For example, as described above, the rainfall prediction device 100 determines the time of rainfall t based on the trend of changes in the rate of increase of water vapor. ２ Precipitation will be predicted from now on. However, precipitation does not necessarily have to be predicted based on the trend of the rate of increase of water vapor; it is sufficient if it is predicted based on the trend of the change in water vapor.

[0057] The various aspects of this disclosure are described below as appendices. (Appendix 1) A rainfall forecasting device configured to perform the following: a process of acquiring measured values ​​of an index that indicates the amount of water vapor in the atmosphere from a measuring instrument that measures the amount of water vapor in the atmosphere; and a prediction process of predicting the time of occurrence of rainfall that may occur in the future based on the trend of change in the amount of water vapor indicated by the time series data of the measured values. (Appendix 2) The rainfall forecasting device according to Appendix 1, wherein the prediction process further includes a process of predicting the amount of precipitation after the time of occurrence based on the trend of change in the amount of water vapor indicated by the time series data of the measured values. (Appendix 3) The rainfall forecasting device according to Appendix 1 or 2, wherein the prediction process is further performed to determine whether the measured values ​​are on a decreasing trend for a predetermined time based on the time series data of the measured values, and the prediction process is performed when it is not determined that the measured values ​​are on a decreasing trend for a predetermined time. (Note 4) A rainfall forecasting device according to Note 2, comprising: a process of converting time-series data of measured values ​​into time-series data of the rate of increase of the measured values; and a process of predicting the amount of rainfall based on the trend of change in the rate of increase of water vapor shown by the time-series data of the rate of increase. (Note 5) A rainfall forecasting method comprising: obtaining measured values ​​of an index from a measuring instrument that measures an index indicating the amount of water vapor in the atmosphere; and predicting the time of occurrence of rainfall that may occur in the future based on the trend of change in the amount of water vapor shown by the time-series data of the measured values.

[0058] 1: Judgment circuit, 2: First analysis circuit, 3: Second analysis circuit, 4: Communication device, 10: Rain cloud, 20: New rain cloud, 40: Input unit, 41: Output unit, 42: Communication unit, 43: CPU, 44: Memory, 45: HDD, 46: Bus, 47: Storage medium, 100: Rainfall prediction device, 200: Measuring device, 300: External device

Claims

1. A rainfall forecasting device configured to perform the following: a process of acquiring measured values ​​of an index that indicates the amount of water vapor in the atmosphere from a measuring instrument; and a prediction process of predicting the time of rainfall that may occur in the future based on the trend of change in the amount of water vapor indicated by the time-series data of the measured values.

2. The rainfall forecasting device according to claim 1, further comprising a process for predicting the amount of precipitation after the time of occurrence based on the trend of change in the amount of water vapor indicated by the time series data of the measured values.

3. A rainfall forecasting device according to claim 1 or 2, further comprising a process to determine whether the measured value is decreasing over a predetermined period of time based on the time-series data of the measured value, wherein the forecasting process is performed when it is determined that the measured value is not decreasing over a predetermined period of time.

4. A rainfall prediction method comprising: obtaining measured values ​​of an index indicating the amount of water vapor in the atmosphere from a measuring instrument; and predicting the time of future rainfall based on the trend of changes in the amount of water vapor indicated by the time-series data of the measured values.

Images