A rainfall prediction method, system, device and storage medium

Abstract

The application discloses a rainfall prediction method, system, device and storage medium, wherein the method comprises the following steps: acquiring a first historical rainfall data set of a ground observation station and a second historical rainfall data set of a radar observation station; determining first predicted rainfall data of a current day under a background field based on background field data and the first historical rainfall data set, and determining second predicted rainfall data after the current day under the background field based on the background field data and the second historical rainfall data set; inputting the first predicted rainfall data and the second historical rainfall data set into a convolutional neural network to obtain third predicted rainfall data; adjusting parameters of the convolutional neural network according to the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network; and inputting rainfall data before a prediction day into the trained convolutional neural network to obtain target rainfall data of the prediction day. The application can be widely applied in the field of artificial intelligence technology.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a rainfall prediction method, system, device and storage medium. BACKGROUND

[0002] Rainfall prediction is a complex and important research task, and its technical background covers meteorology, statistics, satellite remote sensing technology, and machine learning and deep learning, etc. Many mature technologies have been developed, such as numerical weather prediction, meteorological radar technology, satellite remote sensing technology, etc. Meteorological radar technology uses echo intensity to monitor rainfall distribution in real time, and realizes short-term nowcasting through radar mosaics, but it is mainly suitable for short-term prediction. Satellite remote sensing technology provides spatial and temporal distribution information of clouds and precipitation through polar and geostationary satellites, which complements the deficiency of radar data. In meteorology, rainfall prediction mainly relies on theoretical models based on atmospheric dynamics and thermodynamics. These models simulate the movement, humidity, temperature and other physical variables of the atmosphere through numerical weather prediction (NWP), and achieve long-time scale and large-scale prediction through high-performance computing. On the other hand, early statistical methods such as time series analysis and linear regression also provide a preliminary solution to rainfall prediction by establishing the relationship between historical data and rainfall. With the development of observation technology, meteorological satellites and radars have gradually become important data sources for rainfall prediction. Satellites can provide macro information of clouds and humidity, and radars can capture local precipitation intensity, which greatly improves the accuracy of rainfall prediction.

[0003] For different regions and time ranges, the method of rainfall prediction will be different. For short time (0-6 hours) and regions with relatively stable climate, some relatively simple models can well complete the prediction task. However, for regions with rapid changes in atmospheric state and longer time ranges (4-6 hours or more), the physical model of prediction is required to have strong simulation capability, such as simulating the precipitation process. This kind of model belongs to the category of numerical weather prediction (NWP). NWP model simulates physical changes through partial differential equations and approximates related equations through numerical methods. As one of the NWP models, the Weather Research and Forecasting Model (WRF) can simulate different physical processes by providing parameter configurations, making the simulation conditions more similar to the corresponding region. For different regions, the optimal grid resolution and WRF model parameter configuration may not be small, and it needs to be tried several times. For example, in order to get the best rainfall prediction of British Columbia, Canada, researchers have studied more than 100 possible model configurations; some studies have found that in some complex terrain, high-resolution (3 km) grids are better for rainfall value prediction, but low-resolution (27 km) grids perform better overall.

[0004] In the related art, rainfall prediction technology shows a diversified development trend of combining traditional physical models with modern data-driven models. The traditional numerical weather prediction (NWP) method is based on atmospheric dynamics and thermodynamics principles and can simulate large-scale and long-time rainfall conditions, but due to high computational complexity and sensitivity to initial conditions, its accuracy in local rainfall and short-time rainfall prediction is limited. Therefore, there are still technical problems to be solved in the related art. SUMMARY

[0005] The present application aims to at least partially solve one of the technical problems in the prior art.

[0006] To this end, one purpose of the embodiments of the present application is to provide a rainfall prediction method, system, device and storage medium, which can improve the accuracy of rainfall prediction.

[0007] In order to achieve the above technical purpose, the technical solution adopted by the embodiments of the present application comprises: a rainfall prediction method comprising the following steps: obtaining a first set of historical rainfall data of a ground observation station and a second set of historical rainfall data of a radar observation station; wherein the first set of historical rainfall data is a set of historical rainfall data observed by the ground observation station before the current day; the second set of historical rainfall data is a set of the first set of historical rainfall data and rainfall data of the current day; determining first predicted rainfall data of the current day under the background field based on the background field data and the first set of historical rainfall data, and determining second predicted rainfall data after the current day under the background field based on the background field data and the second set of historical rainfall data; inputting the first predicted rainfall data and the second set of historical rainfall data into a convolutional neural network to obtain third predicted rainfall data after the current day; adjusting the parameters of the convolutional neural network according to the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network; inputting rainfall data before the prediction day into the trained convolutional neural network to obtain target rainfall data of the prediction day.

[0008] The application can obtain a first historical rainfall data set of a ground observation station and a second historical rainfall data set of a radar observation station; the first historical rainfall data set is a set of historical rainfall data observed by the ground observation station before the current day; the second historical rainfall data set is a set of the first historical rainfall data set and rainfall data of the current day; based on the background field data and the first historical rainfall data set, first predicted rainfall data of the current day under the background field is determined, and based on the background field data and the second historical rainfall data set, second predicted rainfall data after the current day under the background field is determined; the first predicted rainfall data and the second historical rainfall data set are input into a convolutional neural network to obtain third predicted rainfall data after the current day; parameters of the convolutional neural network are adjusted according to the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network; rainfall data before a prediction day is input into the trained convolutional neural network to obtain target rainfall data of the prediction day. The application proposes a new prediction method, which fully considers historical rainfall data before the current day and rainfall data of the current day to predict rainfall data of the prediction day, and can improve prediction accuracy.

[0009] In addition, according to the rainfall prediction method of the above-mentioned embodiment of the application, the following additional technical features can also be provided:

[0010] Further, in the embodiment of the application, the determination of the first predicted rainfall data of the current day under the background field based on the background field data and the first historical rainfall data set specifically comprises:

[0011] configuring configuration parameters of a weather research and prediction model and sending the background field data and physical boundary conditions to the weather research and prediction model to obtain a simulation model;

[0012] running the simulation model and inputting the first historical rainfall data set into the simulation model to obtain the first predicted rainfall data.

[0013] Further, in the embodiment of the application, the determination of the second predicted rainfall data after the current day under the background field based on the background field data and the second historical rainfall data set specifically comprises:

[0014] configuring configuration parameters of a weather research and prediction model and sending the background field data and physical boundary conditions to the weather research and prediction model to obtain a simulation model;

[0015] running the simulation model and inputting the second historical rainfall data set into the simulation model to obtain the second predicted rainfall data.

[0016] Further, in the embodiment of the present application, the first predicted rainfall data and the second historical rainfall data set are input into the convolutional neural network to obtain third predicted rainfall data after the current day, specifically including:

[0017] The first predicted rainfall data and the second historical rainfall data set are preprocessed to obtain training data.

[0018] The training data is input into the convolutional neural network to obtain the third predicted rainfall data.

[0019] Further, in the embodiment of the present application, the first predicted rainfall data and the second historical rainfall data set are preprocessed to obtain training data, specifically including:

[0020] The first predicted rainfall data and the second historical rainfall data set are aligned,

[0021] and the first predicted rainfall data and the second historical rainfall data set are adjusted in data format to obtain training data.

[0022] Further, in the embodiment of the present application, the parameters of the convolutional neural network are adjusted according to the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network, specifically including:

[0023] The training error is determined according to the second predicted rainfall data and the third predicted rainfall data.

[0024] When the training error is greater than a preset error, the parameters of the convolutional neural network are adjusted and the convolutional neural network is retrained until the training error is less than or equal to the preset error, and the trained convolutional neural network is obtained.

[0025] Further, in the embodiment of the present application, the training error is determined according to the second predicted rainfall data and the third predicted rainfall data, specifically including:

[0026] The second predicted rainfall data and the third predicted rainfall data are subtracted to obtain a first difference.

[0027] The first difference is taken as the training error.

[0028] On the other hand, the embodiment of the present application also provides a rainfall prediction system, including:

[0029] a first processing unit configured to obtain a first set of historical rainfall data of a ground observation station and a second set of historical rainfall data of a radar observation station, wherein the first set of historical rainfall data is a set of historical rainfall data observed by the ground observation station before a current day, and the second set of historical rainfall data is a set of rainfall data of the current day and the first set of historical rainfall data;

[0030] a second processing unit configured to determine first predicted rainfall data of the current day under a background field based on the background field data and the first set of historical rainfall data, and determine second predicted rainfall data after the current day under the background field based on the background field data and the second set of historical rainfall data;

[0031] a third processing unit configured to input the first predicted rainfall data and the second set of historical rainfall data into a convolutional neural network to obtain third predicted rainfall data after the current day;

[0032] a fourth processing unit configured to adjust parameters of the convolutional neural network according to the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network;

[0033] a fifth processing unit configured to input rainfall data before a prediction day into the trained convolutional neural network to obtain target rainfall data of the prediction day.

[0034] In another aspect, the present application also provides a rainfall prediction device, comprising:

[0035] at least one processor;

[0036] at least one memory configured to store at least one program;

[0037] when the at least one program is executed by the at least one processor, the at least one processor is caused to implement the rainfall prediction method according to any one of the summary.

[0038] In addition, the present application also provides a computer readable storage medium, wherein the computer readable storage medium stores processor executable instructions, and the processor executable instructions are used to execute the rainfall prediction method according to any one of the above when executed by a processor.

[0039] The advantages and beneficial effects of the present application will be partially given in the following description, partially will become obvious from the following description, or will be understood by the practice of the present application:

[0040] The application can obtain a first historical rainfall data set of a ground observation station and a second historical rainfall data set of a radar observation station; the first historical rainfall data set is a set of historical rainfall data observed by the ground observation station before the current day; the second historical rainfall data set is a set of the first historical rainfall data set and rainfall data of the current day; based on the background field data and the first historical rainfall data set, the first predicted rainfall data of the current day under the background field is determined, and based on the background field data and the second historical rainfall data set, the second predicted rainfall data after the current day under the background field is determined; the first predicted rainfall data and the second historical rainfall data set are input into a convolutional neural network to obtain third predicted rainfall data after the current day; the parameters of the convolutional neural network are adjusted according to the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network; and rainfall data before a prediction day is input into the trained convolutional neural network to obtain target rainfall data of the prediction day. The application proposes a new prediction method, which fully considers the historical rainfall data before the current day and the rainfall data of the current day to predict the rainfall data of the prediction day, and can improve the prediction accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 A step schematic diagram of a rainfall prediction method in one embodiment of the application;

[0042] Figure 2 A flowchart of a rainfall prediction method in one embodiment of the application

[0043] Figure 3 A structure schematic diagram of a rainfall prediction system in another embodiment of the application;

[0044] Figure 4 A structure schematic diagram of a rainfall prediction device in one embodiment of the application. DETAILED DESCRIPTION

[0045] The principles and processes of the rainfall prediction method, system, device and storage medium in the embodiments of the application are described below in combination with the drawings.

[0046] First, the terms of the application are explained:

[0047] 1. NWP (Numerical Weather Prediction) model: A numerical weather prediction model is a general term for weather forecasting methods based on physics, mathematics, and computer science. It simulates the future changes in atmospheric state by solving the basic physical equations (such as fluid dynamics equations, thermodynamic equations) that describe the motion of the atmosphere. In the prediction of rainfall, the NWP model predicts the time, intensity, and distribution of precipitation by simulating the evolution of parameters such as humidity, air flow, and temperature in the atmosphere.

[0048] 2. WRF (Weather Research and Forecasting Model): A specific model of NWP, dedicated to weather research and actual weather forecasting tasks.

[0049] 3. RMSE (Root Mean Square Error): A statistical index that measures the deviation between predicted values and actual observations, calculated by taking the square of the prediction error and then taking the square root of the average.

[0050] 4. HR (Hit Rate): A measure of the accuracy of a prediction model, calculated by dividing the number of true positives by the sum of true positives and false negatives.

[0051] 5. FAR (False Alarm Rate): A measure of the error rate of a model, indicating the prediction effect is excellent when the FAR is low. It is calculated by dividing the number of false positives by the sum of true positives and false positives.

[0052] 6. CSI (Critical Success Index): An index for evaluating the performance of a prediction model or classifier, calculated by dividing the number of true positives by the sum of true positives, false positives, and false negatives.

[0053] 7. FSS (Fractional Skill Score): An index for evaluating spatial prediction performance, mainly used in meteorology and environmental modeling, which requires verification of spatial distribution. The calculation method is relatively complex. The closer the FSS is to 1, the closer the prediction result is to the true value.

[0054] 8. Grid: A basic unit used to discretize continuous space in numerical simulation, dividing the study area into a regular grid structure.

[0055] 9. Resolution: The spatial scale of the grid, indicating the size of each grid cell, usually expressed in distance (such as kilometers or meters).

[0056] 10. ECMWF (European Centre for Medium-Range Weather Forecasts): The European Centre for Medium-Range Weather Forecasts is an international organization headquartered in Reading, UK, dedicated to the research, development, and operation of medium-range weather forecasting (i.e., 3-10 days into the future). ECMWF is widely recognized as one of the most advanced numerical weather prediction agencies in the world.

[0057] 11. Machine Learning: A general term for techniques that allow computers to automatically learn patterns from data and make predictions or decisions without explicit programming.

[0058] 12. Deep Learning: Deep learning is a machine learning method that automatically learns features and patterns in data through multiple layers of neural networks, enabling classification, regression, or generation tasks.

[0059] 13. CNN (Convolutional Neural Network): A deep learning model that is particularly good at processing images and spatiotemporal data. In precipitation prediction, CNNs generate AI models to calibrate predicted data by learning the differences between real data and predicted data.

[0060] Precipitation prediction is a complex and important research task that involves multiple fields such as meteorology, statistics, satellite remote sensing technology, and machine learning and deep learning. Various mature technologies have been developed, such as numerical weather prediction, meteorological radar technology, and satellite remote sensing technology. Meteorological radar technology uses echo intensity to monitor rainfall distribution in real time and achieves short-term nowcasting through radar mosaics, but it is mainly suitable for short-term prediction. Satellite remote sensing technology provides cloud and precipitation spatiotemporal distribution information through polar and geostationary satellites, complementing the lack of radar data. In meteorology, precipitation prediction relies on theoretical models based on atmospheric dynamics and thermodynamics, which simulate atmospheric motion, humidity, temperature, and other physical variables through numerical weather prediction (NWP) and high-performance computing for long-term and large-scale prediction. On the other hand, early statistical methods such as time series analysis and linear regression also provide preliminary solutions for precipitation prediction by establishing relationships between historical data and rainfall. With the development of observation technology, meteorological satellites and radars have become important data sources for precipitation prediction. Satellites can provide macro information on clouds and humidity, while radars can capture local precipitation intensity, greatly improving the accuracy of precipitation prediction.

[0061] For different regions and time ranges, the methods of rainfall prediction vary. For short time (0-6 hours) and regions with relatively stable climate, some simpler models can well complete the prediction task. However, for regions with rapid changes in atmospheric state and longer time ranges (4-6 hours or more), the prediction physical model is required to have strong simulation capability, such as simulating the precipitation process, which belongs to the category of numerical weather prediction (NWP). The NWP model simulates physical changes through partial differential equations and approximates the related equations through numerical methods. As one of the NWP models, the Weather Research and Forecasting Model (WRF) can simulate different physical processes by providing parameter configurations, making the simulation conditions more approximate to the corresponding regions. For different regions, the optimal grid resolution and WRF model parameter configuration may not be small, and it needs to be tried several times. For example, in order to obtain the best precipitation prediction of British Columbia, Canada, researchers have studied more than 100 possible model configurations; it has also been found that in some complex terrain, high-resolution (3 km) grids are better for precipitation value prediction, but low-resolution (27 km) grids are better in overall performance.

[0062] In related technologies, rainfall prediction technology shows a diversified development trend of combining traditional physical models with modern data-driven models. The traditional numerical weather prediction (NWP) method is based on atmospheric dynamics and thermodynamics principles, which can simulate large-scale and long-time rainfall, but due to high computational complexity and sensitivity to initial conditions, its accuracy in local rainfall and short-term rainfall prediction is limited. Therefore, there are still technical problems to be solved in related technologies.

[0063] In view of the defects of the above prior art, Figure 1 The present application provides a rainfall prediction method. In Figure 1 The method can include the following steps S101-S105.

[0064] S101, obtaining a first historical rainfall data set of a ground observation station and a second historical rainfall data set of a radar observation station; wherein the first historical rainfall data set is a set of historical rainfall data observed by the ground observation station before the current day; and the second historical rainfall data set is a set of the first historical rainfall data set and the current day rainfall data.

[0065] S102, determining the first predicted rainfall data of the current day under the background field based on the background field data and the first historical rainfall data set, and determining the second predicted rainfall data after the current day under the background field based on the background field data and the second historical rainfall data set.

[0066] S103, input the first predicted rainfall data and the second historical rainfall data set into the convolutional neural network to obtain third predicted rainfall data after the current day.

[0067] S104, adjust the parameters of the convolutional neural network according to the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network.

[0068] S105, input rainfall data before the prediction day into the trained convolutional neural network to obtain target rainfall data of the prediction day.

[0069] It can be understood that the prediction day is one day after the current day, and the prediction day and the current day are adjacent two days. The current day can be any day.

[0070] In some possible embodiments of the present application, the processor can be connected with the acquisition unit through wired or wireless connection. After the connection is established, the processor can acquire the first historical rainfall data set of the ground observation station and the second historical rainfall data set of the radar observation station from the acquisition unit; the first historical rainfall data set is a set of historical rainfall data observed by the ground observation station before the current day; the second historical rainfall data set is a set of the first historical rainfall data set and the rainfall data of the current day; based on the background field data and the first historical rainfall data set, the processor can determine the first predicted rainfall data of the current day under the background field, and based on the background field data and the second historical rainfall data set, the processor can determine the second predicted rainfall data after the current day under the background field; input the first predicted rainfall data and the second historical rainfall data set into the convolutional neural network, and the processor can obtain the third predicted rainfall data after the current day; according to the second predicted rainfall data and the third predicted rainfall data, the processor can adjust the parameters of the convolutional neural network to obtain a trained convolutional neural network; input the rainfall data before the prediction day into the trained convolutional neural network, and the processor can obtain the target rainfall data of the prediction day.

[0071] It should be noted that the rainfall data can be the rainfall of the day. The above wired connection mode can include the connection between the mobile device and the processing module, and can also include the connection between the processing module and the hardware device, as well as the wired connection between other now known or future developed devices and the processing module; and the above wireless connection mode can include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (Ultra Wide Band) connection and other now known or future developed wireless connection modes.

[0072] The application can obtain a first historical rainfall data set of a ground observation station and a second historical rainfall data set of a radar observation station; the first historical rainfall data set is a set of historical rainfall data observed by the ground observation station before the current day; the second historical rainfall data set is a set of the first historical rainfall data set and rainfall data of the current day; based on the background field data and the first historical rainfall data set, first predicted rainfall data of the current day under the background field is determined, and based on the background field data and the second historical rainfall data set, second predicted rainfall data after the current day under the background field is determined; the first predicted rainfall data and the second historical rainfall data set are input into a convolutional neural network to obtain third predicted rainfall data after the current day; based on the second predicted rainfall data and the third predicted rainfall data, parameters of the convolutional neural network are adjusted to obtain a trained convolutional neural network; rainfall data before a prediction day is input into the trained convolutional neural network to obtain target rainfall data of the prediction day. The application proposes a new prediction method, which fully considers historical rainfall data before the current day and rainfall data of the current day to predict rainfall data of the prediction day, and can improve prediction accuracy.

[0073] Further, in the embodiment of the application, the process of determining the first predicted rainfall data of the current day under the background field based on the background field data and the first historical rainfall data set can include steps S201-S202.

[0074] S201, configure parameters of a weather research and forecast model and send the background field data and physical boundary conditions to the weather research and forecast model to obtain a simulation model.

[0075] S202, run the simulation model and input the first historical rainfall data set to the simulation model to obtain the first predicted rainfall data.

[0076] It can be understood that the simulation model can be a weather research and forecast model, and the first predicted rainfall data can be rainfall data of the current day obtained by running the weather research and forecast model.

[0077] Further, in the embodiment of the application, the process of determining the second predicted rainfall data after the current day under the background field based on the background field data and the second historical rainfall data set can include steps S301-S302.

[0078] S301, configure parameters of a weather research and forecast model and send the background field data and physical boundary conditions to the weather research and forecast model to obtain a simulation model.

[0079] S302, run the simulation model and input the second historical rainfall data set to the simulation model to obtain the second predicted rainfall data.

[0080] It can be understood that the simulation model can be a weather research and forecast model, and the second prediction data can be predicted rainfall data of a prediction day obtained by running the weather research and forecast model, the predicted rainfall data being a prediction value.

[0081] Further, in the embodiments of the present application, the process of inputting the first predicted rainfall data and the second historical rainfall data set into the convolutional neural network to obtain the third predicted rainfall data after the current day can include steps S401-S402.

[0082] S401, data preprocessing is performed on the first predicted rainfall data and the second historical rainfall data set to obtain training data.

[0083] S402, the training data is input into the convolutional neural network to obtain the third predicted rainfall data.

[0084] Further, in the embodiments of the present application, the process of data preprocessing the first predicted rainfall data and the second historical rainfall data set to obtain training data can include step S501.

[0085] S501, data alignment processing is performed on the first predicted rainfall data and the second historical rainfall data set,

[0086] and data format adjustment processing is performed on the first predicted rainfall data and the second historical rainfall data set to obtain the training data.

[0087] Further, in the embodiments of the present application, the process of adjusting the parameters of the convolutional neural network according to the second predicted rainfall data and the third predicted rainfall data to obtain the trained convolutional neural network can include steps S601-S602.

[0088] S601, a training error is determined according to the second predicted rainfall data and the third predicted rainfall data.

[0089] S602, when the training error is greater than a preset error, the parameters of the convolutional neural network are adjusted and the convolutional neural network is retrained until the training error is less than or equal to the preset error, and a trained convolutional neural network is obtained.

[0090] Further, in the embodiments of the present application, the process of determining a training error according to the second predicted rainfall data and the third predicted rainfall data can include steps S701-S702.

[0091] S701, a first difference is obtained by subtracting the second predicted rainfall data from the third predicted rainfall data.

[0092] S702, the first difference is taken as the training error.

[0093] The principles of the present application are explained below with reference to the accompanying drawings Figure 2 The principles of the present application are explained below with reference to the accompanying drawings

[0094] With reference to Figure 2 The embodiment proposes a method of selecting the optimal grid resolution through pre-experiment and calibrating the prediction results through convolutional neural network (CNN) training model. The innovation lies in performing pre-experiment to obtain the overall optimal spatial resolution, and calibrating the prediction results through CNN on this basis.

[0095] The specific steps are as follows:

[0096] 1. Data preparation and processing: Collect the precipitation of the prediction area periodically (such as every hour) through the observation station on the ground, collect the corresponding radar observation data (the radar observation data is more accurate, generally as the true value) through the national environmental agency, and obtain the corresponding background field, boundary condition and other data from the database provided by the European Center for Medium-Range Weather Forecasts (ECMWF). Classify the data and take the data for pre-experiment, numerical simulation of rainfall prediction and verification of model effect respectively.

[0097] 2. Pre-experiment and spatial resolution selection: Set several possible spatial resolutions, select a small number of days for pre-experiment, select the optimal resolution considering accuracy and calculation efficiency through analyzing the deviation of rainfall intensity, time distribution and spatial coverage range from observation data, and evaluation indexes such as CSI and FSS, and set the spatial resolution in the following experiment.

[0098] 3. WRF numerical simulation: In this technical solution, the WRF version used is v4.5. Based on the optimal resolution and physical scheme determined by pre-experiment, the WRF model is run to numerically simulate the rainfall process of the target area, and output the rainfall prediction results (such as cumulative rainfall, rainfall intensity time series).

[0099] 4. CNN configuration and training: The rainfall data from WRF simulation output is matched with the corresponding radar observation data to generate input-output sample pairs. In this technical solution, the CNN model is composed of five convolutional layers, one fully connected layer, and one output layer. Each convolutional layer contains at least 32 filters, and ReLu is used as the activation function. The input of the first layer consists of various physical parameters extracted from the WRF model and the grid resolution. In the convolution operation, a randomly initialized convolution kernel is applied to the input features. Feature maps are obtained through the output of the first layer, and these feature maps are used as the input of the second layer. In subsequent layers, this process is repeated. The output of the fifth convolutional layer is passed to the fully connected layer. The last layer converts the data into a precipitation intensity map with the same spatial resolution as the input data and outputs it. The constructed model is used to train the CNN model with the input data described above, optimize the loss function, and verify the performance of the model on an independent test set. According to the evaluation results, the model performance is optimized (by adjusting the hyperparameters).

[0100] Actual application and performance evaluation: In actual rainfall prediction, the data obtained directly through WRF prediction and the prediction data obtained after CNN calibration are compared with radar observation data. The results are presented in the form of high-resolution spatial rainfall maps, and the improvement effect of the invention is evaluated by evaluation indicators such as RMSE, CSI, and FSS. For different rainfall time statistics, the improvement range is analyzed, and the calibrated results are converted into standardized meteorological products (such as rainfall intensity prediction maps and rainfall probability forecasts), which are provided to meteorological departments or related users in the form of images or data files.

[0101] In some embodiments, taking Singapore as an example, the specific application process of the embodiment is demonstrated in detail. In terms of data preparation, we periodically collected rainfall observation data from various observation stations in Singapore from 2020 to 2021, and focused on a day containing heavy rain weather, the data was obtained from the National Environment Agency of Singapore. In addition to ground observation data, we also obtained the same rainfall data through radar observation data provided by the National Environment Agency, which was used as the true data. At the same time, we used the 12 UTC initial and boundary conditions of the high-resolution deterministic model from the European Centre for Medium-Range Weather Forecasts (ECMWF) for all experiments. The ECMWF model input data is at 3-hour intervals with a spatial resolution of 0.1°. To test the effect of WRF spatial resolution on Singapore rainfall simulation, multiple domain resolutions are set. In this embodiment, the version of WRF is set to v4.5. When training, we selected the most representative 15 days from the obtained data, which contained different seasons and rainfall levels, to conduct pre-experiments.

[0102] After obtaining the data, the domain resolution of 1 km, 3 km, 9 km and 12 km is taken respectively, the pre-experiment is carried out by using the WRF model, the prediction results are evaluated by using the evaluation indexes such as CSI and FSS, and the spatial resolution with the best comprehensive effect is selected. In this embodiment, the spatial resolution of 1 km has better effect, and the scores of CSI and FSS are the highest in each type of weather. Therefore, 1 km is taken as the spatial resolution of WRF, the numerical simulation is carried out through the observation data, and the prediction data is obtained. Then, the rainfall data output by the WRF simulation is matched with the corresponding radar observation data to generate the input-output sample pair through the obtained prediction data and the corresponding radar observation data. After the data processing is completed, the CNN model is used for training, the configuration of the CNN is the same as that in the technical scheme, and the model capable of calibrating the prediction result is obtained. Finally, the comparative experiment is carried out in the data set set as the verification set, the results of directly carrying out the numerical prediction by using the WRF model and the results after the calibration by using the AI model are compared through the indexes such as CSI and FSS.

[0103] The results show that, by using the CNN model for calibration, compared with directly using the WRF for improvement, the improvement is achieved on each index. Specifically, in this embodiment, the comprehensive improvement of the method in this embodiment on the FSS score is 10.1% compared with the traditional method, the comprehensive improvement on the CSI score is 9.8%, and the improvement on the overall RMSE is 11.2%.

[0104] In addition, with reference to Figure 3 , the Figure 1Corresponding to the method, the embodiment of the application further provides a rainfall prediction system. The system can include a first processing unit 1001, a second processing unit 1002, a third processing unit 1003, a fourth processing unit 1004, and a fifth processing unit 1005. The first processing unit 1001 can be used to obtain a first set of historical rainfall data of a ground observation station and a second set of historical rainfall data of a radar observation station. The first set of historical rainfall data is a set of historical rainfall data observed by the ground observation station before the current day. The second set of historical rainfall data is a set of rainfall data of the first set of historical rainfall data and the current day. The second processing unit 1002 can be used to determine first predicted rainfall data of the current day under a background field based on the background field data and the first set of historical rainfall data, and determine second predicted rainfall data after the current day under the background field based on the background field data and the second set of historical rainfall data. The third processing unit 1003 can be used to input the first predicted rainfall data and the second set of historical rainfall data into a convolutional neural network to obtain third predicted rainfall data after the current day. The fourth processing unit 1004 can be used to adjust the parameters of the convolutional neural network based on the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network. The fifth processing unit 1005 can be used to input rainfall data before a prediction day into the trained convolutional neural network to obtain target rainfall data of the prediction day.

[0105] It should be noted that the first processing unit can be any integrated circuit unit or microprocessor unit obtained by integrating a chip with processing function and its peripheral circuit through existing integrated technology. The first processing unit and the second processing unit can also be any integrated circuit module or microprocessor module obtained by integrating a chip with processing function and its peripheral circuit through existing integrated technology. The first processing unit and the second processing unit can further include one or more memories.

[0106] It should be noted that the contents in the above rainfall prediction method embodiments are applicable to the rainfall prediction system embodiments. The rainfall prediction system embodiments specifically realize the same functions as the above rainfall prediction method embodiments, and achieve the same beneficial effects as the above rainfall prediction method embodiments.

[0107] Corresponding to the method, Figure 1 The embodiment of the application further provides a rainfall prediction device. The specific structure can refer to Figure 4 , comprising:

[0108] at least one processor 1011;

[0109] at least one memory 1012 for storing at least one program;

[0110] When the at least one program is executed by the at least one processor, the at least one processor implements the rainfall prediction method.

[0111] The contents in the method embodiments are applicable to the device embodiments, the device embodiments specifically implement the functions same as the method embodiments, and achieve the beneficial effects same as the method embodiments.

[0112] Corresponding to the method of Figure 1 Corresponding to the method of

[0113] The contents in the method embodiments are applicable to the storage medium embodiments, the storage medium embodiments specifically implement the functions same as the method embodiments, and achieve the beneficial effects same as the method embodiments.

[0114] In some alternative embodiments, the functions / operations mentioned in the block diagram can not occur in the order mentioned in the operation diagram. For example, depending on the functions / operations involved, two blocks shown in succession can actually be executed substantially concurrently or the blocks can sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts are provided by way of example only. The disclosed methods are not limited to the operations and logical flows presented in the diagrams. Alternative embodiments are possible where the order of various operations is changed and where sub operations described as part of a larger operation are executed independently.

[0115] Furthermore, although the present application is described in the context of functional modules, it is to be understood that one or more of the functions and / or features can be integrated in a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It is also to be understood that detailed discussion of the actual implementation of each module is unnecessary to an understanding of the present application. Rather, the actual implementation is within the routine of an engineer given the nature of the functional modules, their properties, and their interrelationships as described in the disclosure herein. Therefore, the present application is not limited to the specific embodiments disclosed herein, but rather, covers all modifications and variations that fall within the scope of the claims and their equivalents.

[0116] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts of the prior art that make contributions or parts of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes several programs for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0117] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of ordered steps for implementing logical functions, which can be embodied in any computer readable medium for use by a program execution system, device or equipment (such as a computer-based system, a system including a processor, or other system that can take programs from a program execution system, device or equipment and execute them), or in conjunction with these program execution systems, devices or equipment. For the purpose of this specification, "computer readable medium" can be any device that can contain, store, communicate, propagate or transport programs for use by or in conjunction with program execution systems, devices or equipment.

[0118] More specific examples (a non-exhaustive list) of the computer readable medium include the following: an electrical connection having one or more wires (electrical devices), a portable computer diskette (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium can even be paper or other suitable medium on which the program can be printed, because the program can be obtained electronically, for example, by optical scanning of the paper or other medium, followed by editing, interpreting or otherwise processing, if necessary, in other suitable ways, to be electronically obtained and then stored in the computer memory.

[0119] It should be understood that various parts of the present application can be implemented in hardware, software, firmware or a combination thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable

[0120] In the above description of the present specification, the description referring to the terms "one embodiment", "another embodiment" or "certain embodiments" or the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Also, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in one or more embodiments or examples.

[0121] Although the embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, alternatives and variations can be made thereto without departing from the principles and spirit of the present application, the scope of which is defined by the appended claims and their equivalents.

[0122] The above is a specific description of the preferred embodiments of the present application, but the present application is not limited to the described embodiments, and those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present application, and these equivalent modifications or substitutions are included in the scope defined by the claims of the present application.

Claims

1. A rainfall prediction method, characterized in that, Includes the following steps: Acquire a first set of historical rainfall data from ground observation stations and a second set of historical rainfall data from radar observation stations; wherein the first set of historical rainfall data is a set of historical rainfall data observed by ground observation stations up to the current day; and the second set of historical rainfall data is a set of the first set of historical rainfall data and rainfall data for the current day. Based on the background field data and the first set of historical rainfall data, the first predicted rainfall data for the current day under the background field is determined, and based on the background field data and the second set of historical rainfall data, the second predicted rainfall data for the days after the current day under the background field is determined. The determination of the first predicted rainfall data for the current day based on the background field data and the first set of historical rainfall data includes: Configure the configuration parameters of the weather research and forecasting model and send the background field data and physical boundary conditions to the weather research and forecasting model to obtain the simulation model; Run the simulation model and input the first set of historical rainfall data into the simulation model to obtain the first predicted rainfall data; The determination of the second predicted rainfall data for the current day and beyond based on background field data and a second set of historical rainfall data includes: Configure the configuration parameters of the weather research and forecasting model and send the background field data and physical boundary conditions to the weather research and forecasting model to obtain the simulation model; The simulation model is run and a second set of historical rainfall data is input into the simulation model to obtain the second predicted rainfall data; the background field data includes the corresponding background field obtained from the database provided by the European Centre for Medium-Range Weather Forecasting. The first set of predicted rainfall data and the second set of historical rainfall data are input into a convolutional neural network to obtain the third set of predicted rainfall data for the current day and beyond. Based on the second and third predicted rainfall data, the parameters of the convolutional neural network are adjusted to obtain a trained convolutional neural network. The rainfall data prior to the predicted date is input into the trained convolutional neural network to obtain the target rainfall data for the predicted date.

2. The rainfall prediction method according to claim 1, characterized in that, The step of inputting the first predicted rainfall data and the second historical rainfall data set into a convolutional neural network to obtain the third predicted rainfall data for the current day and beyond specifically includes: The first predicted rainfall data and the second historical rainfall data set are preprocessed to obtain training data; The training data is input into the convolutional neural network to obtain the third predicted rainfall data.

3. The rainfall prediction method according to claim 2, characterized in that, The step of preprocessing the first predicted rainfall data and the second historical rainfall data set to obtain training data specifically includes: The first predicted rainfall data and the second historical rainfall data set are aligned, and the data format of the first predicted rainfall data and the second historical rainfall data set is adjusted to obtain training data.

4. The rainfall prediction method according to claim 1, characterized in that, The step of adjusting the parameters of the convolutional neural network based on the second and third predicted rainfall data to obtain a trained convolutional neural network specifically includes: The training error is determined based on the second and third predicted rainfall data. When the training error is greater than the preset error, the parameters of the convolutional neural network are adjusted and the convolutional neural network is retrained until the training error is less than or equal to the preset error, thus obtaining the trained convolutional neural network.

5. The rainfall prediction method according to claim 1, characterized in that, The step of determining the training error based on the second predicted rainfall data and the third predicted rainfall data specifically includes: The first difference is obtained by subtracting the second predicted rainfall data from the third predicted rainfall data; The first difference is taken as the training error.

6. A rainfall prediction system, characterized in that, include: The first processing unit is used to acquire the first set of historical rainfall data from the ground observation station and the second set of historical rainfall data from the radar observation station. The first historical rainfall data set is a collection of historical rainfall data up to the current day observed by ground observation stations; the second historical rainfall data set is a collection of the first historical rainfall data set and the rainfall data for the current day. The second processing unit is used to determine the first predicted rainfall data for the current day under the background field based on the background field data and the first historical rainfall data set, and to determine the second predicted rainfall data for the days after the current day under the background field based on the background field data and the second historical rainfall data set. The determination of the first predicted rainfall data for the current day based on the background field data and the first set of historical rainfall data includes: Configure the configuration parameters of the weather research and forecasting model and send the background field data and physical boundary conditions to the weather research and forecasting model to obtain the simulation model; Run the simulation model and input the first set of historical rainfall data into the simulation model to obtain the first predicted rainfall data; The determination of the second predicted rainfall data for the current day and beyond based on background field data and a second set of historical rainfall data includes: Configure the configuration parameters of the weather research and forecasting model and send the background field data and physical boundary conditions to the weather research and forecasting model to obtain the simulation model; The simulation model is run and a second set of historical rainfall data is input into the simulation model to obtain the second predicted rainfall data; the background field data includes the corresponding background field obtained from the database provided by the European Centre for Medium-Range Weather Forecasting. The third processing unit is used to input the first predicted rainfall data and the second historical rainfall data set into a convolutional neural network to obtain the third predicted rainfall data after the current day. The fourth processing unit is used to adjust the parameters of the convolutional neural network based on the second predicted rainfall data and the third predicted rainfall data to obtain a trained convolutional neural network. The fifth processing unit is used to input the rainfall data before the prediction date into the trained convolutional neural network to obtain the target rainfall data for the prediction date.

7. A rainfall prediction device, characterized in that... include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements a rainfall prediction method as described in any one of claims 1-5.

8. A computer-readable storage medium storing processor-executable instructions, characterized in that, The processor-executable instructions, when executed by the processor, are used to perform a rainfall prediction method as described in any one of claims 1-5.

Images