Method and device for forecasting weather
By adopting a data assimilation framework based on deep learning models in the weather forecasting system, processing and fusion of background data and satellite observation data, the challenges of existing systems in data processing and forecast accuracy are solved, and more efficient and accurate weather forecasts are achieved.
Patent Information
- Application Number
- CN202411795899.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-06
- Publication Date
- 2025-03-14
AI Technical Summary
The existing weather forecasting system faces challenges such as insufficient data quality, high computing resources and time consumption, and difficulty in effectively utilizing cloud and precipitation data when processing large amounts of observation data, resulting in limited accuracy and efficiency of weather forecasts.
Using a deep learning model-based data assimilation (DA) framework, background data and satellite observation data are processed through acquisition, encoding, combining, cropping and mixing steps, data weights are automatically learned, and information containing analytical increments is output to improve data processing efficiency and weather forecast accuracy.
Improve data processing efficiency, simplify the data assimilation process, enhance the accuracy and efficiency of weather forecasts, enable more efficient use of satellite data, and reduce manual preprocessing steps.
Smart Images
Figure CN119620241A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to a method, device, medium, electronic device, and computer program product for forecasting weather. Background Art
[0002] Accurate weather forecasts are essential for saving lives, emergency management, mitigating the impact of disasters, and preventing economic losses due to severe weather events. Currently, weather forecasting relies mainly on numerical weather prediction (NWP) models. These traditional models generate weather forecasts by solving governing partial differential equations for initial conditions based on the current state of the atmosphere and the surface. In recent years, a new generation of deep learning (DL) models for medium-range weather forecasting have been recognized for their potential in medium-range weather forecasting, demonstrating forecast performance comparable to leading NWP models. Weather forecasting is essentially an "initial value" problem, where the largest uncertainty comes from the initial conditions. Therefore, the accuracy of these initial conditions is critical regardless of the forecasting method used. Currently, both NWP and DL-based weather prediction models rely on analytical states, which are derived from NWP outputs and observations through data assimilation (DA) systems, ensuring the highest possible accuracy of initial conditions. DA systems lay the foundation for reliable weather forecasts and are equally important as weather forecast models.
[0003] Over the past few decades, advances in NWP models and DA systems have significantly improved the accuracy of weather forecasts. Modern operational DA systems need to process a large amount of observational data, including satellite data, sounding balloons, aircraft and ground observations, to produce analytical states for forecast models within a one-hour time window. Despite these advances, the development of DA systems faces several challenges. One major challenge is the huge volume of observational data, many of which are compromised due to insufficient quality. This issue, coupled with the large amount of computing resources and time required for data processing, forces operational DA systems to find a balance between increasing the amount of assimilated observations and maintaining the timeliness of forecasts. For example, while the European Centre for Medium-Range Weather Forecasts (ECMWF) receives 800 million observations per day, only 60 million pass quality control for integration into the Integrated Forecast System (IFS). In addition, satellite observations are mainly assimilated under clear sky conditions, ignoring valuable cloud and precipitation data, resulting in only 5-10% of total satellite data being used in global weather forecasts. Notably, this small fraction of data contributes the majority of relevant information for weather forecasting. Another challenge lies in the complexity of traditional DA workflows, which leads to substantial delays from the acquisition of new observations to their operational application. This challenge is amplified by the ever-increasing volume of available observations, which presents even greater scientific challenges for DA systems. Notably, satellite-measured variables such as radiance, brightness temperature, or reflectivity do not directly correspond to model variables such as temperature, humidity, or ozone. Therefore, observation operators must be developed to convert model variables into model equivalents of observed data for comparison, a process that not only needs to be tailored for each new type of observation but also introduces additional errors. Finally, widely adopted DA methods such as four-dimensional variational (4D-Var) and ensemble-variational (En-Var) methods are computationally expensive. These challenges highlight the need for continued innovation in DA techniques to improve the efficiency and accuracy of weather forecasts in the face of growing data volumes and demands. Summary of the invention
[0004] Embodiments of the present application provide a method, apparatus, medium, electronic device, and computer program product for forecasting weather.
[0005] In a first aspect, an embodiment of the present application provides a method for forecasting weather, which is used in an electronic device, and the method includes:
[0006] A collection step, obtaining a regular grid area including an observation area from the background area, wherein the regular grid area includes a plurality of grid points, and calculating grid point observation data of each grid point in a predetermined observation time period;
[0007] An encoding step, encoding the grid point observation data of each grid point to obtain additional observation data of each grid point;
[0008] a combining step of combining the grid point observation data and the additional observation data of each grid point to obtain combined observation data of each grid point;
[0009] A cutting step, cutting the background data in the background area to match the regular grid area, thereby obtaining background input data;
[0010] A mixing step, mixing the background input data and the combined observation data to obtain mixed input data;
[0011] The forecasting step is to input the combined observation data, the background input data, and the mixed input data into a deep learning model, and forecast the weather based on the output data.
[0012] In a second aspect, an embodiment of the present application provides a device for forecasting weather, the device comprising:
[0013] A collection unit, which acquires a regular grid area including an observation area from a background area, wherein the regular grid area includes a plurality of grid points, and calculates grid point observation data of each grid point in a predetermined observation time period;
[0014] An encoding unit encodes the grid point observation data of each grid point to obtain additional observation data of each grid point;
[0015] a combining unit, combining the grid point observation data and the additional observation data of each grid point to obtain combined observation data of each grid point;
[0016] A cutting unit, cutting the background data in the background area to match the regular grid area, thereby obtaining background input data;
[0017] A mixing unit, mixing the background input data and the combined observation data to obtain mixed input data;
[0018] The forecasting unit inputs the combined observation data, the background input data, and the mixed input data into a deep learning model, and forecasts the weather based on the output data.
[0019] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, on which instructions are stored. When the instructions are executed on a computer, the computer executes the method described in the first aspect above.
[0020] In a fourth aspect, an embodiment of the present application provides an electronic device, comprising: one or more processors; one or more memories; the one or more memories storing one or more programs, and when the one or more programs are executed by the one or more processors, the electronic device executes the method described in the first aspect above.
[0021] In a fifth aspect, an embodiment of the present application provides a computer program product, comprising computer executable instructions, wherein the instructions are executed by a processor to implement the method described in the first aspect.
[0022] In the present invention, a data assimilation (DA) framework based on a deep learning model is used to forecast weather, which can efficiently process and fuse background data and satellite observation data, automatically learn the weights of background data and satellite observation data, and output information containing analysis increments. In this way, not only the data processing efficiency is improved, but also the data assimilation process is simplified by reducing the manual preprocessing steps. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 A flow chart of a method for forecasting weather according to an embodiment of the present application is shown;
[0024] Figure 2 A structural diagram of a device for forecasting weather according to an embodiment of the present application is shown;
[0025] Figure 3 A block diagram of an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION
[0026] The following will describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application; it is obvious that the described embodiments are only part of the embodiments of the present application, not all of the embodiments. According to the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0027] Figure 1 A flow chart of a method for forecasting weather according to an embodiment of the present invention is shown. The method is used in electronic devices, such as computers, servers, mobile terminals, etc.
[0028] like Figure 1 As shown, in the acquisition step S11, a regular grid area including the observation area is acquired from the background area, the regular grid area includes a plurality of grid points, and grid point observation data of each grid point in a predetermined observation time period is calculated.
[0029] The background area is, for example, the earth, from which a regular grid area including the observation area is cut out. The regular grid area includes a plurality of grid points, and each grid point is, for example, an area of 25 km×25 km.
[0030] Specifically, satellite observation data within a predetermined range of each grid point is acquired, and mean calculation is performed on the satellite observation data, thereby obtaining grid point observation data of each grid point.
[0031] In this embodiment, the predetermined range is, for example, ±12.5 kilometers, and satellite observation data within a range of ±12.5 kilometers around each grid point is obtained. It can be understood that the satellite observation data includes cloud layer, temperature, humidity, and the like.
[0032] The satellite observation data is AGRI (Advanced Geostationary Radiation Imager) data, which is high-resolution meteorological data obtained by meteorological satellites and is mainly used to monitor and analyze the state of the atmosphere, surface and ocean. AGRI can provide continuous, real-time images and data, and is mainly used in weather forecasting, climate monitoring, disaster warning and other fields. For AGRI data, the corresponding background size is 640 (first latitude) × 640 (first longitude).
[0033] There are 8 variables in the AGRI data, and the mean is calculated for each variable. For example, the mean of all temperatures within a range of ±12.5 kilometers around a grid point is calculated to obtain the temperature mean. Similarly, the mean is calculated for the remaining 7 variables, thereby obtaining the means of 8 variables. It can be understood that the grid point observation data of a grid point includes the means of the 8 variables. In other words, the first number of variables is 8.
[0034] In addition, the grid point observation data is for a predetermined observation time period, such as 2 hours. If 15 minutes is set as 1 time frame, then 2 hours contains 8 time frames. In other words, the first number of time frames determined by the predetermined range is 8. In this way, the size of the grid point observation data of each grid point is 8 (first number of time frames) × 8 (first number of variables) × 640 (first latitude) × 640 (first longitude).
[0035] Furthermore, if no satellite observation data within the range of ±12.5 km around a grid point is obtained, the grid point is masked and will not be processed later.
[0036] In the encoding step S12, the grid point observation data of each grid point is encoded to obtain additional observation data of each grid point.
[0037] Considering that the deviation of satellite observation data (grid point observation data) usually varies with the scanning angle and geographical location, the present invention optimizes the input data of the model by encoding. Specifically, the grid point observation data of each grid point includes the satellite zenith angle and time information, and the time information is converted into the day of the year and the minute of the day.
[0038] The specific encoding is to calculate the sine value of the first longitude, the cosine value of the first latitude, the cosine value of the satellite zenith angle, the sine value and cosine value of the day, and the sine value and cosine value of the minute, respectively, to obtain 7 variables, that is, the second variable number is 7.
[0039] Through the above encoding, the size of the obtained additional observation data is 8 (the number of first time frames)×7 (the number of second variables)×640 (the first latitude)×640 (the first longitude).
[0040] In the combining step S13, the grid point observation data and the additional observation data of each grid point are combined to obtain the combined observation data of each grid point.
[0041] Specifically, the first number of variables and the second number of variables are added to obtain the third number of variables. In this embodiment, the third number of variables is 7+8=15. In this way, the size of the combined observation data of each grid point is 8 (first time frame number)×15 (third number of variables)×640 (first latitude)×640 (first longitude).
[0042] In the cutting step S14, the background data in the background area is cut to match the regular grid area, thereby obtaining background input data.
[0043] The background area contains background data, for example, the initialization data of the earth. The size of the background data is 1 (time) × 70 (number of the fourth variable) × 721 (initial latitude) × 1440 (initial longitude). For the background data, since there is only one time point, the time dimension is "1", and the time dimension is omitted in the following description.
[0044] In order to adapt the background data to the model, the size of the background data is adjusted by bilinear interpolation to the following size: 70 (the fourth variable number) × 720 (the second latitude) × 1440 (the second longitude).
[0045] Then, the adjusted background data is cropped to match the regular grid area obtained in the acquisition step S11, that is, the longitude and latitude of the data are consistent with the longitude and latitude of the satellite observation data (grid point observation data) in the regular grid area, so as to obtain the background input data. The size of the background input data is 70 (the fourth variable number) × 640 (the first latitude) × 640 (the first longitude).
[0046] In the mixing step S15, the background input data and the combined observation data are mixed to obtain mixed input data.
[0047] Specifically, the product of the first time frame number and the third variable number is added to the fourth variable number to obtain the fifth variable number. In this embodiment, the fifth variable number calculated is 8×15+70=190. In this way, the size of the mixed input data is 190 (fifth variable number)×640 (first latitude)×640 (first longitude).
[0048] In the forecasting step S16, the three types of data, namely the combined observation data, the background input data, and the mixed input data obtained as described above, are input into the deep learning model, and then the weather is forecasted based on the output data.
[0049] The deep learning model is, for example, a U-net model, which is used to process data from three branches, including two downsampling and two upsampling stages.
[0050] The architecture and functions of the model network are as follows: Each downsampling stage includes a 2×2 2D convolution layer (stride 2), a layer normalization layer, a SiLU activation function, and a 3×3 2D convolution layer (stride 1). Each upsampling stage includes a 3×3 2D convolution layer (stride 1), a layer normalization layer, a SiLU activation function, a 3×3 2D convolution layer (stride 1), and a pixel reorganization layer (magnification factor 2). In addition, a skip connection is included to connect the output of the downsampling stage in series with the output of the upsampling stage before entering the next upsampling stage.
[0051] It can be understood that according to the size of the input data (e.g., longitude, latitude), some parameters of the U-net model can be changed accordingly.
[0052] In the present invention, a data assimilation (DA) framework based on a deep learning model is used to forecast weather, which can efficiently process and fuse background data and satellite observation data, automatically learn the weights of background data and satellite observation data, and output information containing analysis increments. In this way, not only the data processing efficiency is improved, but also the data assimilation process is simplified by reducing the manual preprocessing steps.
[0053] The present invention also provides a device for forecasting weather. Figure 2 As shown, the device 20 comprises:
[0054] The acquisition unit 201 acquires a regular grid area including an observation area from a background area, wherein the regular grid area includes a plurality of grid points, and calculates grid point observation data of each grid point in a predetermined observation time period;
[0055] The encoding unit 202 encodes the grid point observation data of each grid point to obtain additional observation data of each grid point;
[0056] A combining unit 203 combines the grid point observation data and the additional observation data of each grid point to obtain combined observation data of each grid point;
[0057] A cutting unit 204 cuts the background data in the background area to match the regular grid area, thereby obtaining background input data;
[0058] A mixing unit 205, mixing the background input data and the combined observation data to obtain mixed input data;
[0059] The forecasting unit 206 inputs the combined observation data, the background input data, and the mixed input data into a deep learning model, and forecasts the weather based on the output data.
[0060] It is understandable that the acquisition unit 201, the encoding unit 202, the combination unit 203, the cutting unit 204, the mixing unit 205, and the prediction unit 206 can be Figure 3 The processor 1404 in the electronic device 1400 having the functions of these modules or units is implemented.
[0061] The present invention also provides a computer-readable storage medium, wherein the storage medium stores instructions, which, when executed on a computer, cause the computer to execute Figure 1 The method shown in .
[0062] The present invention also provides a computer program product, including computer executable instructions, which are executed by the processor 1404 to implement Figure 1 The method shown in .
[0063] Reference now Figure 3 , Figure 3 An example electronic device 1400 according to an embodiment of the present invention is schematically shown. In one embodiment, the electronic device 1400 may include one or more processors 1404, a system control logic 1408 connected to at least one of the processors 1404, a system memory 1412 connected to the system control logic 1408, a non-volatile memory (NVM) 1416 connected to the system control logic 1408, and a network interface 1420 connected to the system control logic 1408.
[0064] In some embodiments, processor 1404 may include one or more single-core or multi-core processors. In some embodiments, processor 1404 may include any combination of general-purpose processors and special-purpose processors (eg, graphics processors, application processors, baseband processors, etc.).
[0065] In some embodiments, system control logic 1408 may include any suitable interface controller to provide any suitable interface to at least one of processors 1404 and / or any suitable device or component in communication with system control logic 1408 .
[0066] In some embodiments, the system control logic 1408 may include one or more memory controllers to provide an interface to the system memory 1412. The system memory 1412 may be used to load and store data and / or instructions. In some embodiments, the memory 1412 of the electronic device 1400 may include any suitable volatile memory, such as a suitable dynamic random access memory (DRAM).
[0067] NVM / memory 1416 may include one or more tangible, non-transitory computer-readable media for storing data and / or instructions. In some embodiments, NVM / memory 1416 may include any suitable non-volatile memory such as flash memory and / or any suitable non-volatile storage device, such as at least one of a HDD (Hard Disk Drive), a CD (Compact Disc) drive, and a DVD (Digital Versatile Disc) drive.
[0068] NVM / storage 1416 may include a portion of storage resources on a device on which electronic device 1400 is installed, or it may be accessible by the device but not necessarily a portion of the device. For example, NVM / storage 1416 may be accessed over a network via network interface 1420.
[0069] In particular, system memory 1412 and NVM / storage 1416 may include, respectively, a temporary copy and a permanent copy of instructions 1424. Instructions 1424 may include instructions that, when executed by at least one of processors 1404, cause electronic device 1400 to implement the following: Figure 1 In some embodiments, instructions 1424, hardware, firmware, and / or software components thereof may additionally / alternatively be placed in system control logic 1408, network interface 1420, and / or processor 1404.
[0070] The network interface 1420 may include a transceiver for providing a radio interface for the electronic device 1400, thereby communicating with any other suitable device (such as a front-end module, an antenna, etc.) through one or more networks. In some embodiments, the network interface 1420 may be integrated with other components of the electronic device 1400. For example, the network interface 1420 may be integrated with at least one of the processor 1404, the system memory 1412, the NVM / storage 1416, and a firmware device (not shown) having instructions. When at least one of the processors 1404 executes the instructions, the electronic device 1400 implements the following. Figure 1 The method shown.
[0071] The network interface 1420 may further include any suitable hardware and / or firmware to provide a multiple-input multiple-output radio interface. For example, the network interface 1420 may be a network adapter, a wireless network adapter, a telephone modem and / or a wireless modem.
[0072] In one embodiment, at least one of the processors 1404 may be packaged together with logic for one or more controllers of the system control logic 1408 to form a system in package (SiP). In one embodiment, at least one of the processors 1404 may be integrated on the same die with logic for one or more controllers of the system control logic 1408 to form a system on a chip (SoC).
[0073] The electronic device 1400 may further include an input / output (I / O) device 1432. The I / O device 1432 may include a user interface to enable a user to interact with the electronic device 1400; the design of the peripheral component interface enables the peripheral components to interact with the electronic device 1400. In some embodiments, the electronic device 1400 further includes a sensor for determining at least one of an environmental condition and location information related to the electronic device 1400.
[0074] In some embodiments, the user interface may include, but is not limited to, a display (e.g., an LCD display, a touch screen display, etc.), a speaker, a microphone, one or more cameras (e.g., a still image camera and / or a video camera), a flashlight (e.g., an LED flash), and a keyboard.
[0075] The various embodiments of the mechanism disclosed in the present application can be implemented in hardware, software, firmware or a combination of these implementation methods. The embodiments of the present application can be implemented as a computer program or program code executed on a programmable system, which includes at least one processor, a storage system (including volatile and non-volatile memory and / or storage elements), at least one input device and at least one output device.
[0076] Program code can be applied to input instructions to perform the functions described in this application and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, a processing system includes any system having a processor such as, for example, a digital signal processor (DSP), a microcontroller, an application specific integrated circuit (ASIC), or a microprocessor.
[0077] Program code can be implemented with high-level programming language or object-oriented programming language to communicate with the processing system. When necessary, program code can also be implemented with assembly language or machine language. In fact, the mechanism described in this application is not limited to the scope of any specific programming language. In either case, the language can be a compiled language or an interpreted language.
[0078] In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried or stored on one or more temporary or non-temporary machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors. For example, instructions may be distributed over a network or through other computer-readable media. Therefore, a machine-readable medium may include any mechanism for storing or transmitting information in a machine (e.g., computer) readable form, including, but not limited to, a floppy disk, an optical disk, an optical disk, a read-only memory (CD-ROM), a magneto-optical disk, a read-only memory (ROM), a random access memory (RAM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic card or an optical card, a flash memory, or a tangible machine-readable memory for transmitting information (e.g., carrier wave, infrared signal digital signal, etc.) using the Internet in an electrical, optical, acoustic or other form of propagation signal. Therefore, a machine-readable medium includes any type of machine-readable medium suitable for storing or transmitting electronic instructions or information in a machine (e.g., computer) readable form.
[0079] In the accompanying drawings, some structural or method features may be shown in a specific arrangement and / or order. However, it should be understood that such a specific arrangement and / or order may not be required. Instead, in some embodiments, these features may be arranged in a manner and / or order different from that shown in the illustrative drawings. In addition, the inclusion of structural or method features in a particular figure does not mean that such features are required in all embodiments, and in some embodiments, these features may not be included or may be combined with other features.
[0080] It should be noted that the units / modules mentioned in the various device embodiments of the present application are all logical units / modules. Physically, a logical unit / module can be a physical unit / module, or a part of a physical unit / module, or can be implemented as a combination of multiple physical units / modules. The physical implementation method of these logical units / modules themselves is not the most important. The combination of functions implemented by these logical units / modules is the key to solving the technical problems proposed by the present application. In addition, in order to highlight the innovative part of the present application, the above-mentioned device embodiments of the present application do not introduce units / modules that are not closely related to solving the technical problems proposed by the present application, which does not mean that there are no other units / modules in the above-mentioned device embodiments.
[0081] It should be noted that in the examples and description of this patent, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including one" do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0082] Although the present application has been illustrated and described with reference to certain preferred embodiments thereof, it will be apparent to those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present application.
Claims
1. A method for forecasting weather, used in electronic equipment, characterized in that: The method comprises: A collection step, obtaining a regular grid area including an observation area from the background area, wherein the regular grid area includes a plurality of grid points, and calculating grid point observation data of each grid point in a predetermined observation time period; An encoding step, encoding the grid point observation data of each grid point to obtain additional observation data of each grid point; a combining step of combining the grid point observation data and the additional observation data of each grid point to obtain combined observation data of each grid point; A cutting step, cutting the background data in the background area to match the regular grid area, thereby obtaining background input data; A mixing step, mixing the background input data and the combined observation data to obtain mixed input data; The forecasting step is to input the combined observation data, the background input data, and the mixed input data into a deep learning model, and forecast the weather based on the output data.
2. The method according to claim 1, characterized in that In the acquisition step, satellite observation data within a predetermined range of each grid point is obtained, and the satellite observation data is averaged to obtain grid point observation data of each grid point. The size of the grid point observation data is the first time frame number×the first variable number×the first latitude×the first longitude determined by the predetermined range.
3. The method according to claim 2, characterized in that If satellite observation data within a predetermined range of a grid point is not acquired, mask processing is performed on the grid point.
4. The method according to claim 2, characterized in that: The grid point observation data includes the satellite zenith angle and time information, and the time information is converted into the day of the year and the minute of the day. Wherein, in the encoding step, the sine value of the first longitude, the cosine value of the first latitude, the cosine value of the satellite zenith angle, the sine value and cosine value of the day, and the sine value and cosine value of the minute are calculated respectively, thereby obtaining a second variable number, The size of the additional observation data is the first number of time frames×the second number of variables×the first latitude×the first longitude.
5. The method according to claim 4, characterized in that In the combining step, the first variable number and the second variable number are added to obtain a third variable number, The size of the combined observation data is the number of the first time frames×the number of the third variables×the first latitude×the first longitude.
6. The method according to claim 5, characterized in that The size of the background input data is a fourth variable number×the first latitude×the first longitude, Wherein, in the mixing step, the product of the first time frame number and the third variable number is added to the fourth variable number to obtain a fifth variable number, The size of the mixed input data is the fifth variable number×the first latitude×the first longitude.
7. The method according to claim 1, characterized in that The deep learning model is a U-net model.
8. A device for forecasting weather, characterized in that: The device comprises: A collection unit, which obtains a regular grid area including an observation area from a background area, wherein the regular grid area includes a plurality of grid points, and calculates grid point observation data of each grid point in a predetermined observation time period; An encoding unit encodes the grid point observation data of each grid point to obtain additional observation data of each grid point; a combining unit, combining the grid point observation data and the additional observation data of each grid point to obtain combined observation data of each grid point; A cutting unit, cutting the background data in the background area to match the regular grid area, thereby obtaining background input data; A mixing unit, mixing the background input data and the combined observation data to obtain mixed input data; The forecasting unit inputs the combined observation data, the background input data, and the mixed input data into a deep learning model, and forecasts the weather based on the output data.
9. A computer-readable storage medium, characterized in that: The storage medium stores instructions, which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 7.
10. An electronic device, characterized in that: include: one or more processors; One or more memories; the one or more memories store one or more programs, and when the one or more programs are executed by the one or more processors, the electronic device executes the method according to any one of claims 1 to 7.
11. A computer program product comprising computer executable instructions, characterized in that: The instructions are executed by a processor to implement the method according to any one of claims 1 to 7.