Weather forecasting device, weather forecasting method, and program

Abstract

To accurately predict weather information.SOLUTION: A weather forecasting device includes an acquisition unit that acquires weather forecast data including multiple members based on an ensemble forecast and actual weather data for which at least a portion of a time interval corresponds to the weather forecast data, a similarity calculation unit that calculates the similarity between each member and the actual weather data for the corresponding time interval, and an integration unit that generates weather forecast data that integrates the members based on the similarity.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present disclosure relates to a weather forecasting device, a weather forecasting method, and a program. [Background technology]

[0002] There are known techniques for predicting weather information based on weather forecast data. For example, Patent Literature 1 discloses a photovoltaic output prediction support device that acquires information on the solar radiation distribution of a target prediction day, extracts similar days from past days that have solar radiation distributions similar to the target prediction day, and acquires prediction errors for each prediction model for similar days based on measurement results of photovoltaic power generation output for the past days and prediction results for each prediction model for the past days. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7128106 Summary of the Invention [Problem to be solved by the invention]

[0004] However, there is room for improvement in the accuracy of weather information forecasts in the conventional techniques. For example, the conventional technique disclosed in Patent Document 1 uses a single forecast value and cannot take into account the uncertainty of the forecast value, so that the forecast accuracy may decrease if the forecast value is incorrect.

[0005] One aspect of the present disclosure provides a technology that can accurately predict weather information. [Means for solving the problem]

[0006] A weather forecasting device according to one aspect of the present disclosure includes an acquisition unit that acquires weather forecast data including multiple members based on an ensemble forecast and actual weather data for at least a portion of a time period corresponding to the weather forecast data, a similarity calculation unit that calculates the similarity between each member in the corresponding time period and the actual weather data, and an integration unit that generates weather forecast data that integrates the members based on the similarity. [Effects of the Invention]

[0007] According to one aspect of the present disclosure, weather information can be predicted with high accuracy. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram showing an example of the overall configuration of a weather forecasting system. [Figure 2] FIG. 2 is a block diagram illustrating an example of a hardware configuration of a computer. [Figure 3] FIG. 2 is a block diagram illustrating an example of a functional configuration of the weather prediction device. [Figure 4] FIG. 10 is a diagram illustrating an example of similarity. [Figure 5] FIG. 10 is a diagram illustrating an example of integration processing. [Figure 6] 1 is a flowchart illustrating an example of a weather forecasting method. [Figure 7] 10 is a flowchart showing a first example of an integration process. [Figure 8] 10 is a flowchart showing a second example of the integration process. [Figure 9] 10 is a flowchart showing a third example of the integration process. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. In this specification and drawings, components having substantially the same functional configurations are designated by the same reference numerals, and redundant description will be omitted.

[0010] [Embodiment] An embodiment of the present disclosure is an example of an information processing system that predicts weather information based on weather forecast data. Hereinafter, the information processing system according to this embodiment will be referred to as a "weather forecasting system." In this embodiment, the weather forecast data may be an ensemble forecast distributed from a weather information system operated by the Japan Meteorological Agency.

[0011] The Japan Meteorological Agency (JMA) forecasts weather information such as solar radiation, temperature, or humidity. Weather forecasts by the JMA can be broadly divided into single forecasts and ensemble forecasts. Single forecasts are weather forecasts that forecast weather information for up to several tens of hours into the future every few hours. Examples of single forecasts include global models (GSM; Global Spectral Model), mesoscale models (MSM; Meso Scale Model), and local forecast models (LFM; Local Forecast Model). Single forecasts have the problem that their forecast values ​​are prone to be inaccurate when atmospheric conditions are unstable, such as when a front passes or a typhoon forms.

[0012] An ensemble forecast is a collection of forecast values ​​consisting of multiple members. Each member in an ensemble forecast is a weather forecast recalculated by adding perturbations to a single forecast. Each member is composed of forecast values ​​for meteorological information such as solar radiation, temperature, and air pressure. The number of members in an ensemble forecast is, for example, between 21 and 51. Examples of ensemble forecasts include the Global Spectral Ensemble Prediction System (GEPS) and the Meso-scale Ensemble Prediction System (MEPS). Ensemble forecasts are also used, for example, to create forecast circles for typhoons.

[0013] An information processing system is in operation that distributes weather forecasts from the Japan Meteorological Agency to users. Hereinafter, the information processing system that distributes weather forecasts will be referred to as the "weather information system." Weather forecasts from the Japan Meteorological Agency are distributed after the forecast values ​​are created, and it takes a certain amount of time for users to receive the information. For example, an ensemble forecast can take up to three hours for users to receive it.

[0014] Patent Document 1 discloses a technology for selecting a prediction model with high prediction accuracy for solar power generation from multiple prediction models using a single weather forecast value and weather satellite images as input. Solar power generation is a value obtained by converting the amount of solar radiation from meteorological information based on rating, efficiency, etc., so Patent Document 1 can be said to be a technology related to predicting the amount of solar radiation, which is an example of meteorological information. However, Patent Document 1 uses a single forecast value and cannot take into account the uncertainty of the forecast value, which poses a problem of reduced prediction accuracy when the forecast value itself is incorrect.

[0015] For example, using multiple members included in an ensemble forecast can take into account the uncertainty of forecast values, which is expected to improve the accuracy of weather forecasts. However, because it is difficult to evaluate the forecast accuracy of each member included in an ensemble forecast, Patent Document 1 cannot be directly applied to ensemble forecasts. Patent Document 1 selects a forecast model to use for weather forecasting from multiple forecast models based on the forecast error of a past date similar to the target forecast date. However, because the perturbations used to generate each member of the ensemble forecast are generated each time a forecast is made, the forecast accuracy of each member of a newly delivered ensemble forecast is unrelated to the forecast accuracy of each member of previously delivered ensemble forecasts. Therefore, ensemble forecasts cannot refer to past forecast accuracy, and applying Patent Document 1 to ensemble forecasts does not improve the forecast accuracy of weather information.

[0016] This embodiment aims to accurately predict weather information. To this end, this embodiment acquires weather forecast data including multiple members based on ensemble forecasting and actual weather data corresponding to at least a portion of the time interval of the weather forecast data, and generates weather forecast data integrating the members based on the similarity between each member and the actual weather data for the corresponding time interval. In one aspect, this embodiment can take into account the uncertainty of forecast values, thereby enabling accurate prediction of weather information.

[0017] <Overall structure> The overall configuration of a weather forecasting system according to this embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the overall configuration of a weather forecasting system.

[0018] 1, the weather prediction system 1000 includes a weather information system 10 and a weather prediction device 20. The weather information system 10 and the weather prediction device 20 are connected to each other so as to be able to communicate data with each other via a communication path N. The communication path N may be, for example, a dedicated line, a VPN (Virtual Private Network), the Internet, or another communication path.

[0019] The weather information system 10 is an example of an information processing system that distributes weather forecast data and actual weather data. The weather information system 10 may be an information communication system operated by the Japan Meteorological Agency.

[0020] The weather forecast data is electronic data indicating forecast values ​​of weather information. The weather forecast data may include multiple members based on an ensemble forecast. The multiple members may include forecast values ​​calculated based on one or more forecast models and recalculated using different perturbations. Each of the multiple members may be time-series data indicating forecast values ​​for multiple weather information. The weather information indicated in the weather forecast data may include, for example, solar radiation, temperature, or humidity.

[0021] The actual weather data is electronic data indicating actual measured values ​​of weather information. The actual weather data may be data indicating actual measured values ​​of weather information at a single point in time, or may be time-series data including actual measured values ​​of weather information at multiple points in time. The weather information indicated in the actual weather data may include, for example, solar radiation, temperature, or humidity. It is sufficient that at least one of the weather information indicated in the actual weather data matches the weather information indicated in the weather forecast data. The actual weather data does not have to include any of the weather information indicated in the weather forecast data, and may include one or more weather information not indicated in the weather forecast data.

[0022] The weather information system 10 distributes weather forecast data at predetermined time intervals. The weather information system 10 may also transmit the weather forecast data to the weather prediction device 20 at predetermined time intervals. The weather information system 10 may also publish the weather forecast data electronically and transmit the weather forecast data to the weather prediction device 20 in response to a request from the weather prediction device 20.

[0023] The weather information system 10 distributes actual weather data at predetermined time intervals. The weather information system 10 may transmit the actual weather data to the weather forecasting device 20 at predetermined time intervals. The weather information system 10 may electronically publish the actual weather data and transmit the actual weather data to the weather forecasting device 20 in response to a request from the weather forecasting device 20.

[0024] The weather information system 10 takes a certain amount of time from creating weather forecast data to distributing it. In this embodiment, it takes three hours from creating weather forecast data to distributing it. On the other hand, the weather information system 10 distributes actual weather data immediately after creating it. In other words, the time it takes the weather information system 10 to distribute actual weather data from creating it is longer than the time it takes the weather forecast data from creating it to distributing it. Therefore, the weather forecast data and actual weather data correspond to each other at least in part of the time interval.

[0025] The weather forecasting device 20 is an example of an information processing device such as a personal computer, workstation, or server that predicts weather information. The weather forecasting device 20 acquires weather forecast data and actual weather data distributed by the weather information system 10. The weather forecasting device 20 generates weather forecast data based on the weather forecast data and actual weather data, and outputs the weather forecast data.

[0026] The overall configuration of the weather forecasting system 1000 shown in Fig. 1 is one example, and various system configuration examples are possible depending on the application and purpose. For example, the weather forecasting system 1000 may include multiple units of one or more of the weather information system 10 and the weather forecasting device 20. For example, the weather forecasting device 20 may be realized by multiple computers, or may be realized as a cloud computing service. The classification of devices such as the weather information system 10 and the weather forecasting device 20 shown in Fig. 1 is one example.

[0027] <Hardware configuration> The hardware configuration of each device included in the weather prediction system 1000 will be described with reference to Fig. 2. The weather information system 10 or the weather prediction device 20 included in the weather prediction system 1000 is realized by, for example, a computer. Fig. 2 is a block diagram showing an example of the hardware configuration of a computer.

[0028] 2, the computer 500 includes a CPU (Central Processing Unit) 501, a ROM (Read Only Memory) 502, a RAM (Random Access Memory) 503, a HDD (Hard Disk Drive) 504, an input device 505, a display device 506, a communication I / F (Interface) 507, and an external I / F 508. The CPU 501, the ROM 502, and the RAM 503 form a so-called computer. The hardware components of the computer 500 are connected to each other via a bus line 509. The input device 505 and the display device 506 may be connected to the external I / F 508 for use.

[0029] The CPU 501 is a computing device that reads programs and data from a storage device such as the ROM 502 or the HDD 504 onto the RAM 503 and executes the processes, thereby realizing the overall control and functions of the computer 500. The computer 500 may have a GPU (Graphics Processing Unit) in addition to or instead of the CPU 501.

[0030] The ROM 502 is an example of a non-volatile semiconductor memory (storage device) that can retain programs and data even when the power is turned off. The ROM 502 functions as a main storage device that stores various programs, data, etc. required for the CPU 501 to execute various programs installed in the HDD 504. Specifically, the ROM 502 stores boot programs such as a Basic Input Output System (BIOS) and an Extensible Firmware Interface (EFI) that are executed when the computer 500 starts up, as well as data such as OS (Operating System) settings and network settings.

[0031] The RAM 503 is an example of a volatile semiconductor memory (storage device) in which programs and data are erased when the power is turned off. The RAM 503 is, for example, a dynamic random access memory (DRAM) or a static random access memory (SRAM). The RAM 503 provides a working area in which various programs installed in the HDD 504 are expanded when executed by the CPU 501.

[0032] The HDD 504 is an example of a non-volatile storage device that stores programs and data. The programs and data stored in the HDD 504 include an OS, which is basic software that controls the entire computer 500, and applications that provide various functions on the OS. Note that the computer 500 may use a storage device that uses flash memory as a storage medium (e.g., an SSD (Solid State Drive)) instead of the HDD 504.

[0033] The input device 505 includes a touch panel, operation keys and buttons, a keyboard and mouse, a microphone for inputting sound data such as voice, and the like, which are used by the user to input various signals.

[0034] The display device 506 is configured with a display such as a liquid crystal display or organic EL (Electro-Luminescence) display for displaying a screen, a speaker for outputting sound data such as voice, and the like.

[0035] The communication I / F 507 is an interface that connects to a communication network and enables the computer 500 to perform data communication.

[0036] The external I / F 508 is an interface with external devices, such as a drive device 510.

[0037] The drive device 510 is a device for loading a recording medium 511. The recording medium 511 here includes media that record information optically, electrically, or magnetically, such as a CD-ROM, a flexible disk, or a magneto-optical disk. The recording medium 511 may also include semiconductor memories that record information electrically, such as ROMs and flash memories. This allows the computer 500 to read from and / or write to the recording medium 511 via the external I / F 508.

[0038] The various programs to be installed in the HDD 504 are installed, for example, by setting the distributed recording medium 511 in a drive device 510 connected to the external I / F 508 and reading out the various programs recorded on the recording medium 511 by the drive device 510. Alternatively, the various programs to be installed in the HDD 504 may be installed by being downloaded via the communication I / F 507 from a network different from the communication network.

[0039] <Functional configuration> The functional configuration of the weather prediction system 1000 will be described with reference to Fig. 3. Fig. 3 is a block diagram showing an example of the functional configuration of a weather prediction device.

[0040] As shown in FIG. 3, the weather forecasting device 20 includes a storage unit 110, an acquisition unit 120, a similarity calculation unit 130, an integration unit 140, and an output unit 150.

[0041] The storage unit 110 is realized by the HDD 504 shown in FIG.

[0042] The acquisition unit 120, the similarity calculation unit 130, the integration unit 140, and the output unit 150 are realized by the processing that the CPU 501 executes by a program loaded onto the RAM 503 from the HDD 504 shown in FIG.

[0043] The memory unit 110 stores weather forecast data and actual weather data. The memory unit 110 may also store weather forecast data and actual weather data that have been distributed in the past. When the weather information system 10 distributes actual weather data indicating actual values ​​of weather information at a given point in time, the actual weather data is accumulated in time series in the memory unit 110 to generate time-series data indicating actual values ​​of the weather information. The memory unit 110 may also store other data, such as values ​​generated by the weather forecasting device 20 during calculation.

[0044] The acquisition unit 120 acquires weather forecast data and actual weather data. The acquisition unit 120 may acquire weather forecast data and actual weather data stored in the storage unit 110. When multiple pieces of weather forecast data are stored in the storage unit 110, the acquisition unit 120 may acquire the latest weather forecast data.

[0045] The acquisition unit 120 may acquire at least one of weather forecast data or actual weather data from the weather information system 10. The acquisition unit 120 may receive at least one of weather forecast data or actual weather data distributed from the weather information system 10. The acquisition unit 120 may request at least one of weather forecast data or actual weather data from the weather information system 10. The acquisition unit 120 may accept weather forecast data or actual weather data input via the input device 505 provided in the weather prediction device 20. When the acquisition unit 120 acquires weather forecast data or actual weather data from the weather information system 10, the acquisition unit 120 may store the weather forecast data or actual weather data in the memory unit 110.

[0046] The similarity calculation unit 130 calculates the similarity between each member included in the weather forecast data and the actual weather data. The similarity calculation unit 130 may calculate the similarity between each member included in the weather forecast data and the actual weather data for a time period corresponding to the weather forecast data and the actual weather data. In this embodiment, it takes about three hours from when the weather information system 10 creates the weather forecast data until the weather forecast data is distributed to the weather forecasting device 20, so the similarity calculation unit 130 only needs to calculate the similarity between the first three hours of the weather forecast data and the most recent three hours of the actual weather data.

[0047] For example, the similarity calculation unit 130 may calculate the similarity based on one or a combination of at least two of Euclidean distance, cosine similarity, coefficient of determination, dynamic time warping, and mutual information. For example, the similarity calculation unit 130 may calculate the similarity using a machine learning method such as a neural network. In a distance measure such as Euclidean distance, a larger value indicates a lower similarity, so when a distance measure is used as the similarity, the similarity calculation unit 130 may calculate the reciprocal of the distance as the similarity.

[0048] The similarity calculation unit 130 may calculate the similarity for one or more pieces of weather information. The similarity calculation unit 130 may calculate the similarity using multiple types of weather information as indices. The similarity calculation unit 130 may calculate the similarity by weighting each indices. In this embodiment, the similarity calculation unit 130 may calculate the similarity using the amount of solar radiation and the temperature as indices.

[0049] For example, the similarity calculation unit 130 may calculate the similarity between each member using formula (1). Formula (1) is an example of a formula for calculating the similarity by Euclidean distance using the amount of solar radiation and temperature as indicators.

[0050]

number

[0051] However, t i is the time included in the corresponding time interval between the weather forecast data and the weather record data, m is the index of the member included in the weather forecast data, and Sim m is the similarity of member m, and s a (t i ) is time t i is the actual value of solar radiation at s m (t i ) is time t i is the forecast value of solar radiation for member m at f a (t i ) is time t i is the actual temperature at f m(t i ) is time t i is the forecast value of the temperature for member m in , and α is a weight that adjusts the influence of each index on the similarity.

[0052] The integrating unit 140 generates weather forecast data based on the similarity calculated by the similarity calculating unit 130. The integrating unit 140 may generate weather forecast data by integrating members included in the weather forecast data based on the similarity.

[0053] The integrating unit 140 may integrate the members by taking a weighted average of the members included in the weather forecast data, using the similarity as a weight. As a preprocessing step for taking the weighted average, the integrating unit 140 may normalize the similarity of each member so that the sum of the similarities is 1. As an example, the integrating unit 140 may normalize the similarity of each member using a linear transformation or a softmax function.

[0054] For example, the integration unit 140 may normalize the similarity of each member using equation (2): Equation (2) is an example of a mathematical formula for normalizing the similarity of each member using linear transformation.

[0055]

number

[0056] However, Sim m is the similarity of member m, and Sim i is the similarity of each member included in the weather forecast data, and Sim′ m is the normalized similarity of member m.

[0057] Fig. 4 is a diagram showing an example of similarity. Fig. 4 shows the result of normalizing the similarity calculated for each of M members included in the weather forecast data. As shown in Fig. 4, the similarity is calculated for each of members 1 to M, and the similarity for each member is normalized to a value between 0 and 1 so that the sum is 1. Note that M is the number of members in the weather forecast data.

[0058] For example, the integrating unit 140 may calculate a weighted average of each member using equation (3).

[0059]

number

[0060] However, t j is the time included in the weather forecast data, and Sim′ m is the normalized similarity of member m, and s m (t j ) is time t j is the forecast value of solar radiation for member m at ^s(t j ) is time t j This is the predicted value of solar radiation at . Note that the "^" symbol should be written directly above the character immediately following it, but due to limitations in text notation, it is written immediately before it in the text. In the formula, it is written directly above the actual character.

[0061] The integrating unit 140 may integrate members by selecting members included in the weather forecast data based on the similarity and averaging the selected members. The integrating unit 140 may select a predetermined number of members in descending order of similarity. The integrating unit 140 may select members whose similarity is equal to or greater than a predetermined threshold.

[0062] The integrating unit 140 may select members included in the weather forecast data based on the similarity, and integrate the members by taking a weighted average of the selected members using the similarity as a weight. As a preprocessing step for taking the weighted average, the integrating unit 140 may normalize the similarity of each member so that the sum of the similarities becomes 1. The integrating unit 140 may select a predetermined number of members in descending order of normalized similarity. The integrating unit 140 may select members whose normalized similarity is equal to or greater than a predetermined threshold.

[0063] Fig. 5 is a diagram illustrating an example of the integration process. As shown in Fig. 5, the weather forecast data includes multiple members that are time-series data of forecast values ​​of insolation, and the similarity between each member and the actual weather data is calculated for the time period during which actual weather data is acquired. Then, the members of the weather forecast data are integrated based on their respective similarities to generate weather forecast data that is time-series data of predicted values ​​of insolation.

[0064] The output unit 150 outputs the weather forecast data generated by the integrating unit 140. The output unit 150 may present the weather forecast data to a user of the weather forecasting system 1000. The output unit 150 may display the weather forecast data on a display device 506 provided in the weather forecasting device 20. The output unit 150 may transmit the weather forecast data to a terminal device operated by a user of the weather forecasting system 1000.

[0065] The output unit 150 may transmit the weather forecast data to another information processing device or information processing system that performs predetermined information processing based on the forecasted values ​​of the weather information. As an example, the predetermined information processing may be a solar power generation amount prediction based on the solar radiation amount prediction. As another example, the predetermined information processing may be a power market price prediction based on the solar power generation amount prediction.

[0066] <Processing Procedure> The processing procedure of the weather forecasting method executed by the weather forecasting system 1000 will be described with reference to Fig. 6. Fig. 6 is a flowchart showing an example of the weather forecasting method.

[0067] In step S1, the weather information system 10 distributes weather forecast data to the weather prediction device 20. The weather prediction device 20 stores the weather forecast data distributed from the weather information system 10 in the memory unit 110. The acquisition unit 120 of the weather prediction device 20 reads the latest weather forecast data from the memory unit 110. The acquisition unit 120 sends the acquired weather forecast data to the similarity calculation unit 130 and the integration unit 140.

[0068] In step S2, the weather information system 10 distributes actual weather data to the weather prediction device 20. The weather prediction device 20 stores the actual weather data distributed from the weather information system 10 in the memory unit 110. The acquisition unit 120 of the weather prediction device 20 reads out, from the memory unit 110, actual weather data whose time interval at least corresponds to the weather forecast data acquired in step S1. The acquisition unit 120 sends the acquired actual weather data to the similarity calculation unit 130.

[0069] In step S3, the similarity calculation unit 130 of the weather prediction device 20 receives the weather forecast data and actual weather data from the acquisition unit 120. The similarity calculation unit 130 calculates the similarity between each member included in the weather forecast data and the actual weather data for the time interval corresponding to the weather forecast data and the actual weather data. The similarity calculation unit 130 sends the similarity calculated for each member to the integration unit 140.

[0070] In step S4, the integrating unit 140 of the weather forecasting device 20 receives the weather forecast data from the acquiring unit 120. The integrating unit 140 receives the similarity for each member from the similarity calculating unit 130. The integrating unit 140 executes a predetermined integration process to integrate the members included in the weather forecast data based on the similarity for each member. The integrating unit 140 sends the weather forecast data generated by the integration process to the output unit 150.

[0071] Integration Processing The integration process (step S4 in FIG. 6) executed by the integration unit 140 will be described in more detail with reference to FIGS.

[0072] 7 is a flowchart showing a first example of the integration process. Step S4A, which is the first example of the integration process, is an example of integration process in which a weighted average of members is calculated using the similarity as a weight.

[0073] In step S11, the integrating unit 140 receives the similarity for each member from the similarity calculation unit 130. The integrating unit 140 normalizes the similarity for each member so that the sum of the similarities becomes 1. Specifically, the integrating unit 140 divides each similarity for each member by the sum of the similarities.

[0074] In step S12, the integration unit 140 calculates a weighted average of the members included in the weather forecast data using the similarity normalized in step S11 as a weight. This generates weather forecast data that integrates the members. The integration unit 140 sends the weather forecast data to the output unit 150.

[0075] 8 is a flowchart showing a second example of the integrating process. Step S4B, which is the second example of the integrating process, is an example of integrating process in which members selected based on the similarity are averaged.

[0076] In step S21, the integrating unit 140 receives the similarity for each member from the similarity calculation unit 130. The integrating unit 140 arranges the members in descending order of similarity. The integrating unit 140 selects multiple members with high similarity. The integrating unit 140 may select a predetermined number of members with high similarity. The integrating unit 140 may select members with similarity equal to or greater than a predetermined threshold.

[0077] In step S22, the integrating unit 140 calculates the average of the members selected in step S21. As a result, weather forecast data integrating the members is generated. The integrating unit 140 sends the weather forecast data to the output unit 150.

[0078] 9 is a flowchart showing a third example of the integrating process. Step S4C, which is the third example of the integrating process, is an example of integrating process in which members selected based on similarity are weighted and averaged using similarity as a weight.

[0079] In step S31, the integrating unit 140 receives the similarity for each member from the similarity calculation unit 130. The integrating unit 140 normalizes the similarity for each member so that the sum of the similarities becomes 1. Specifically, the integrating unit 140 divides each similarity for each member by the sum of the similarities.

[0080] In step S32, the integrating unit 140 arranges the members in descending order of similarity after normalization. The integrating unit 140 selects multiple members with high similarity after normalization. The integrating unit 140 may select a predetermined number of members with high similarity after normalization. The integrating unit 140 may select members with similarity after normalization equal to or greater than a predetermined threshold.

[0081] In step S33, the integration unit 140 calculates a weighted average of the members selected in step S32 using the similarity normalized in step S31 as a weight. This generates weather forecast data integrating the members. The integration unit 140 sends the weather forecast data to the output unit 150.

[0082] Returning to Figure 6, in step S5, the output unit 150 of the weather forecasting device 20 receives the weather forecast data from the integrating unit 140. The output unit 150 outputs the weather forecast data. The output unit 150 may present the weather forecast data to a user of the weather forecasting system 1000. The output unit 150 may transmit the weather forecast data to another information processing device or information processing system.

[0083] <Effects of the embodiment> The weather forecasting device 20 of this embodiment acquires weather forecast data including multiple members based on an ensemble forecast and actual weather data for at least a portion of the time period corresponding to the weather forecast data, and generates weather forecast data that integrates the members based on the similarity between each member and the actual weather data for the corresponding time period.

[0084] According to one aspect, this embodiment enables weather information to be predicted with high accuracy. Since ensemble forecasts require a certain amount of time to be distributed from the weather information system, actual weather data can be obtained for some time intervals. Since ensemble forecasts generate perturbations for each forecast, the forecast accuracy of other ensemble forecasts distributed in the past cannot be evaluated. On the other hand, the forecast accuracy for some time intervals of the same ensemble forecast can be used as the forecast accuracy for other time intervals. Therefore, according to this embodiment, the uncertainty of forecast values ​​can be taken into account based on ensemble forecasts, allowing weather information to be predicted with high accuracy.

[0085] The weather forecasting device 20 may integrate members by taking a weighted average of the members, with the similarity as a weight. The weather forecasting device 20 may integrate members by taking an average of the members selected based on the similarity. The weather forecasting device 20 may integrate members by taking a weighted average of the members selected based on the similarity, with the similarity as a weight. In one aspect, according to this embodiment, weather forecast data is generated by prioritizing multiple members with high forecast accuracy, so that forecast values ​​can be calculated with reduced influence of fluctuations in forecast values.

[0086] The weather prediction device 20 may acquire weather forecast data and actual weather data distributed by the weather information system 10. The time between the generation and distribution of the weather forecast data in the weather information system 10 may be longer than the time between the generation and distribution of the actual weather data. In one aspect, according to this embodiment, a portion of the actual weather data can be acquired at the time the weather forecast data is distributed, allowing future weather information to be predicted immediately.

[0087] The weather forecasting device 20 may calculate the similarity based on one or a combination of at least two of the following: Euclidean distance, cosine similarity, coefficient of determination, dynamic time warping, and mutual information. In one aspect, according to this embodiment, the similarity between the weather forecast data and actual weather data can be evaluated based on various measures.

[0088] [supplement] Each function of the above-described embodiments can be realized by one or more processing circuits. Here, the term "processing circuit" in this specification includes a processor programmed to perform each function by software, such as a processor implemented by an electronic circuit, as well as devices such as an ASIC (Application Specific Integrated Circuit), a DSP (Digital Signal Processor), an FPGA (Field Programmable Gate Array), and conventional circuit modules designed to perform each of the above-described functions.

[0089] Although the embodiments of the present invention have been described in detail above, the present invention is not limited to these embodiments, and various modifications and changes are possible within the scope of the gist of the present invention described in the claims. [Explanation of symbols]

[0090] 1000: Weather Forecasting System 10: Weather information system 20: Weather forecasting device 110: Storage section 120: Acquisition Department 130: Similarity calculation part 140: Integration Department 150: Output section

Claims

1. an acquisition unit configured to acquire weather forecast data including a plurality of members based on an ensemble forecast and actual weather data corresponding to at least a portion of a time interval of the weather forecast data; a similarity calculation unit configured to calculate a similarity between each of the members and the weather record data in a corresponding time period; an integration unit configured to generate weather forecast data by integrating the members based on the similarity; A weather forecasting device comprising:

2. the integrating unit is configured to integrate the members by taking a weighted average of the members using the similarity as a weight. The weather forecasting device according to claim 1 .

3. the integrating unit is configured to integrate the members by averaging the members selected based on the similarity. The weather forecasting device according to claim 1 .

4. the integrating unit is configured to integrate the members by taking a weighted average of the members selected based on the similarity, with the similarity used as a weight. The weather forecasting device according to claim 1 .

5. the acquisition unit is configured to acquire the weather forecast data and the weather record data distributed by a predetermined weather information system, the weather information system is configured such that the time from generation to distribution of the weather forecast data is longer than the time from generation to distribution of the weather record data; The weather forecasting device according to any one of claims 1 to 4.

6. the similarity calculation unit is configured to calculate the similarity based on any one or a combination of at least two of Euclidean distance, cosine similarity, coefficient of determination, dynamic time warping, or mutual information. The weather forecasting device according to any one of claims 1 to 4.

7. The computer A step of acquiring weather forecast data including a plurality of members based on an ensemble forecast and actual weather data corresponding to at least a part of the time interval of the weather forecast data; A procedure for calculating a similarity between each of the members and the weather record data in a corresponding time period; generating weather forecast data that integrates the members based on the similarity; A weather forecasting method that performs

8. On the computer, A step of acquiring weather forecast data including a plurality of members based on an ensemble forecast and actual weather data corresponding to at least a part of the time interval of the weather forecast data; A procedure for calculating a similarity between each of the members and the weather record data in a corresponding time period; generating weather forecast data that integrates the members based on the similarity; A program to execute.

Images