Adjacent rainfall forecasting method based on radar extrapolation of evolutionary network and Unet network and Z-R relation

Abstract

The invention discloses an approaching rainfall forecasting method based on radar extrapolation of an evolutionary network and a Unet network and a Z-R relation. The method comprises the following steps: acquiring historical live rainfall data in an hourly period and using the historical live rainfall data to construct a radar echo sample data set; constructing a radar echo short-time proximity extrapolation forecasting model, wherein the radar echo short-time proximity extrapolation forecasting model comprises an evolution network module and a Unet variant module; constructing a model training loss function, and training a radar echo short-time proximity extrapolation forecasting model in combination with the radar echo sample data set; and obtaining radar jigsaw echoes one hour before the prediction time, inputting the radar jigsaw echoes into the trained model to obtain predicted radar echoes, and obtaining a rainfall grid point prediction field based on the Z-R relational expression and the predicted radar echoes. By designing the evolutionary network module, the Unet variant module and the loss function for the generator and the discriminator, the forecasting precision of the close heavy rainfall is improved, the method has a good application prospect, and a more accurate forecasting service is provided for disaster prevention and reduction.

Description

Technical Field

[0001] The invention belongs to the technical field of weather forecasting, and in particular relates to a nowcasting method of precipitation based on radar extrapolation and ZR relationship of an evolutionary network and a Unet network. Background Art

[0002] As one of the important elements of nowcast weather forecast, precipitation forecast is of great significance to the management of modern cities. It is directly related to land and air transportation management, large-scale event support and flood control. High-precision nowcast precipitation forecast is an indispensable part of future smart city construction. However, so far, the forecast information obtained by existing meteorological forecasters is not targeted enough. Among them, there is still a large gap between the TS score of the objective forecast of near-term heavy precipitation and the forecast demand for disaster prevention and mitigation of actual major meteorological disasters, which affects meteorological disaster prevention and mitigation decision-making. The meteorological department urgently needs to improve the accuracy of weather forecasts.

[0003] Nowcasting based on dense observation data such as weather radar is one of the main means of heavy precipitation weather monitoring and warning services. Radar nowcasting is essentially a spatiotemporal sequence prediction task, which can also be regarded as a video prediction task. Taking the past sequence radar echo map as input to predict the future radar echo map, the traditional extrapolation method based on the assumption of linear evolution of radar echo cannot describe nonlinear processes such as the evolution of convective system, so it has great limitations in practical applications. The current deep learning technology has been found to be able to better handle the spatiotemporal changes of convective systems and their internal correlations. Therefore, it is a good idea to solve the precipitation nowcasting problem through deep learning technology, which has also become a new means to improve the nowcasting of heavy precipitation weather. At present, the refined grid forecast of precipitation nowcasting based on radar extrapolation still has the problem of increased accumulated errors, which eventually leads to increased forecast errors; with the extension of the forecast time, the forecast results will become "fuzzy" and the current nowcasting skill score is low. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a precipitation forecasting method based on radar extrapolation and ZR relationship of evolutionary network and Unet network to solve the problems existing in the above prior art.

[0005] To achieve the above object, the present invention provides a precipitation forecasting method based on radar extrapolation and ZR relationship of evolutionary network and Unet network, comprising:

[0006] Acquire historical real-time rainfall data for each hour, and construct a radar echo sample data set based on the historical real-time rainfall data;

[0007] Constructing a radar echo short-time nowcasting extrapolation prediction model; the generator of the radar echo short-time nowcasting extrapolation prediction model includes an evolutionary network module and a Unet variant module;

[0008] Constructing a model training loss function, and training the radar echo short-time nowcasting extrapolation prediction model based on the radar echo sample data set and the model training loss function;

[0009] The radar jigsaw echo one hour before the prediction time is obtained, and the trained radar echo short-term nowcasting prediction model is input to obtain the predicted radar echo, and the precipitation grid forecast field is obtained based on the ZR relationship and the predicted radar echo.

[0010] Optionally, the real-time rainfall data includes real-time rainfall observation data from ground meteorological stations and real-time observation data from weather radars; the real-time observation data from the weather radars is a radar combined reflectivity puzzle.

[0011] Optionally, the process of constructing a radar echo sample dataset includes:

[0012] Each hour within the time range corresponding to the historical real-time rainfall data is used as the starting time of precipitation in the next hour, and the radar combined reflectivity puzzle and precipitation of each 6-minute frame one hour before the starting time are obtained; the radar combined reflectivity puzzles with the top six digits sorted by the time interval between the acquisition time and the starting time are extracted, and overlapped segmentation is performed, and the segmented grids are filtered according to the proportion of echo intensity. After all the screening is completed, the radar echo sample data set is obtained by combining it with the precipitation.

[0013] Optionally, the radar echo sample data set is decomposed into two components by the evolutionary network module, and the evolutionary operator is calculated on the two components to obtain preliminary radar echo extrapolation data; the radar echo sample data set and the preliminary radar echo extrapolation data are used as inputs of the Unet variant module.

[0014] Optionally, on the basis of Unet, an attention mechanism, a hole convolution module and a residual module are introduced to obtain the Unet variant module.

[0015] Optionally, the model training loss function includes a weighted loss function and a power spectrum loss function; the generator training of the model adopts the weighted loss function; and the discriminator training of the model adopts the power spectrum loss function.

[0016] Optionally, the weighted loss function is:

[0017]

[0018] Among them, WeightedLoss is the total loss of the weighted loss function; N is the total number of samples; ωi is the weight of the i-th sample, usually used to adjust the loss of the minority class; y i The true label of the i-th sample; The predicted label of the i-th sample.

[0019] Optionally, the power spectrum loss function is:

[0020] L PSD =||P model (f)-P target (f)|| 2

[0021] Among them, L PSD is the power spectral density loss; P model (f) is the power spectral density of the model output signal; P target (f) is the power spectral density of the target signal, and ||·|| represents the norm.

[0022] Compared with the prior art, the present invention has the following advantages and technical effects:

[0023] The present invention uses the evolutionary network module to calculate the input radar echo sample data set on the input model to obtain a preliminary echo extrapolation data set; the obtained echo extrapolation data set and the initial radar echo sample data set are used as the input of the Unet variant module, and finally the output of the entire generator is obtained. The forecasting ability of the forecasting model for heavy rainfall above 20 mm is improved by more than 15% compared with the current level of imminent rainstorm forecasting, which can effectively improve the forecast accuracy of imminent heavy rainfall, has good application prospects, and provides more accurate forecasting services for disaster prevention and mitigation.

[0024] The present invention designs a power spectrum loss function for the discriminator of the forecast model, which can better characterize the characteristics of radar echo data, more directly evaluate the energy distribution difference between the radar echo data generated by the generator and the real data, and drive the generator to generate a power spectrum distribution that is closer to the real data in the frequency domain. For the generator of the forecast model, a weighted loss function is designed, which is the sum of the weighted losses of the weighted mean absolute error and the weighted mean square error. The weighting coefficient here is assigned different weights according to the actual radar echo intensity, that is, the stronger the radar echo intensity, the greater the weight assigned. The information feedback capability of the radar strong echo area is increased through the above-mentioned weight assignment rule, so that it will not be averaged out by a large area of ​​weak echo area, and a high-precision forecast of precipitation echoes above 30dBZ is achieved, which has a higher forecast accuracy in the early warning of disaster prevention and mitigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0026] Figure 1 It is a flow chart of an objective forecasting method for near-falling precipitation based on radar extrapolation and ZR relationship based on an evolutionary network module and a Unet-Attention-Aspp network structure (EAAUnet) according to an embodiment of the present invention;

[0027] Figure 2 It is the EAAUnet radar extrapolation technology processing model of the embodiment of the present invention;

[0028] Figure 3 A schematic diagram of actual precipitation according to an embodiment of the present invention;

[0029] Figure 4 It is a schematic diagram of precipitation forecast according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0031] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0032] Embodiment 1

[0033] like Figure 1-4 As shown, in this embodiment, a method for nowcasting precipitation based on radar extrapolation and ZR relationship of evolutionary network and Unet network is provided, including:

[0034] Step S1, obtaining the real-time rainfall observation data of the ground meteorological station and the real-time observation data of the weather radar.

[0035] Specifically, the real-time rainfall observation data of ground meteorological stations are obtained from the TianQing database, which is the real-time rainfall in multiple time periods with 1 hour as the node within the historical time period of each station; the radar real-time observation data are obtained from the SWAN system issued by the China Meteorological Administration, and the radar real-time observation data are radar combined reflectivity puzzle data.

[0036] Step S2, constructing a radar echo extrapolation sample data set.

[0037] Specifically, the data used is the radar puzzle of Swan, and the sampling period is from July 2016 to June 2022. During this period, for each hour period when the number of stations with a height of more than 10 mm reaches more than 20, 30 consecutive frames of radar echo are extracted every 12 minutes as a preliminary sample. The preliminary radar echo samples are further segmented with overlap, and the original 900×1200 grid size puzzle is divided into 20 256×256 grids. When the proportion of those with a height greater than 30 dbz in these 20 grids reaches more than 15%, the radar echo grid sample is selected into the final forecast sample data set, and finally 33,244 samples are constructed.

[0038] Step S3, obtaining preliminary radar echo extrapolation data for the next hour through extrapolation of the evolutionary network module;

[0039] Specifically, in order to reduce the pressure of model training, the past 6 consecutive frames of echo are used as model input, and the radar echo map of the next 10 frames is output, that is, the radar echo map of the next hour. This is achieved through the evolutionary network module of the extrapolation model. The input of the model is mainly decomposed into the speed V and intensity S components of the movement through evolutionary coding. These two components are calculated through the operation of the evolutionary operator to obtain preliminary radar echo extrapolation data.

[0040] Step S4, constructing a Unet variant module to obtain the input of the EAAUnet radar extrapolation model discriminator;

[0041] Specifically, the radar data is first input into the evolutionary network module, and then the output of this evolutionary network module is combined with the previous model input as the input of the Unet variant module, and finally the output of the entire generator is obtained.

[0042] Step S5, constructing the EAAUnet radar extrapolation model;

[0043] The main structure of the model is generally an adversarial network structure, and the generator has two parts: one is the evolutionary network module, and the other is the Unet variant module. As for the discriminator of this adversarial network, this embodiment adopts a simple Unet network. The specific operation process of the model is as follows: the radar data input is first input into the evolutionary network module, and then the output of this evolutionary network module is combined with the previous model input as the input of the Unet variant module, and finally the output of the entire generator is obtained. When the model is trained, these three modules use different loss functions to update the model, including a weighted loss function and a power spectrum (PSD) loss function.

[0044] Step S6, constructing a loss function for model training;

[0045] Specifically, in order to better train the model, this embodiment transforms the loss function. In the training of the model, the role of the loss function is to evaluate the fit of the current training of the model and to modify its weight according to the value of the loss function.

[0046] A weighted loss function is designed for the generator, which is the sum of the weighted losses of the weighted mean absolute error and the weighted mean square error. Its general form can be expressed as:

[0047]

[0048] Where: WeightedLoss is the total loss of the weighted loss function; N is the total number of samples; ω i is the weight of the i-th sample, usually used to adjust the loss of the minority class; y i The true label of the i-th sample; The predicted label of the i-th sample.

[0049] The weighting coefficient is assigned different weights according to the actual radar echo strength, that is, the stronger the radar echo strength, the greater the weight assigned. The purpose of this is to increase the information feedback capability of the strong radar echo area so that it is not averaged by a large area of ​​weak echo area. Set the actual radar echo strength to 9 levels:

[0050] First-level echo intensity: 0~15dBZ, second-level echo intensity: 15~25dBZ, third-level echo intensity: 25~35dBZ, fourth-level echo intensity: 35~40dBZ, fifth-level echo intensity: 40~45dBZ, sixth-level echo intensity: 45~50dBZ, seventh-level echo intensity: 50~55dBZ, eighth-level echo intensity: 55~65dBZ, ninth-level echo intensity: ≥65dBZ.

[0051] A power spectrum loss function (PSDLoss) is designed for the discriminator. It is a loss function used in the field of signal processing and system identification. Especially in the model training process, when the frequency characteristics of the signal need to be optimized, this loss function measures the difference between the power spectrum density (PSD) of the model output signal and the PSD of the target signal. Power spectrum density is the distribution of signal energy in the frequency domain and is usually used to describe the frequency characteristics of the signal. PSDLoss calculates the loss by comparing the PSD of the model output signal with the PSD of the real signal or the expected signal, thereby prompting the model to learn the correct frequency response. Its basic expression is:

[0052] L PSD =||P model (f)-P target(f)|| 2

[0053] Among them, L PSD is the power spectral density loss; P model (f) is the power spectral density of the model output signal; P target (f) is the power spectral density of the target signal; ||·|| represents a certain norm, usually the L2 norm (Euclidean norm), but other norms can also be used.

[0054] The use of the power spectrum loss function has three characteristics: first, it can better characterize the characteristics of the echo data; second, it can more directly evaluate the energy distribution difference between the echo data generated by the generator and the real data; third, it drives the generator to generate a power spectrum distribution that is closer to the real echo data in the frequency domain.

[0055] Step S7, training and obtaining a prediction model;

[0056] Step S8, using the ZR relationship to perform calculation and conversion processing from radar echo to precipitation, and generate a precipitation grid forecast field.

[0057] Specifically, the ZR (Zawadzki-Renard relation) is used to invert precipitation from radar echoes, and the grid point forecast value of precipitation is obtained by calculation. The calculation formula is as follows:

[0058] Z=aR b

[0059] Where Z is the radar reflectivity factor, usually in mm^6 / m^3; R is the precipitation rate, in mm / h; a and b are constants whose values ​​depend on the precipitation type and radar wavelength. Here we set a=300 and b=1.4.

[0060] This embodiment mainly constructs an adversarial network deep learning model including an evolutionary network module and a Unet variant module. In the selection of samples, an overlapping segmentation technique is used; the input of the model is constructed to include two different parts of radar data input; the main network structure is based on Unet, and an attention mechanism, a hole convolution module, and a residual module are added to improve the feature extraction and extrapolation capabilities of the model; in order to better train this model, the loss function is further adaptively modified. These are the biggest differences between this method and the current method of precipitation forecasting based on radar echo extrapolation.

[0061] The following example uses the precipitation report for the next hour starting at 08:00 on July 29, 2023 as an example:

[0062] (1) Collect sample data. The statistical period is from June 2016 to June 2022. The samples from June 2016 to December 2021 are used to train the forecast model; the samples from January 2022 to June 2022 are used to test the forecast model. The hourly time of the hour in this period is used as the starting time for precipitation in the next hour. The collected data are: the actual observation data of ground rainfall at the ground meteorological station every hour and the swan radar combined reflectivity puzzle framed every 6 minutes. The actual observation data of ground rainfall is the precipitation in the past hour.

[0063] (2) Radar sample data processing. The swan radar combined reflectivity puzzle is a 900*1200 puzzle grid data. The last 6 frames of radar combined reflectivity puzzles one hour before the start time are cropped into 256×256 grid areas with partial overlap. If the proportion of more than 15% of the 20 256×256 grids is greater than 30dbz, this grid will be selected into the final data set, totaling 33244 samples. The data set is divided into: 1) Training sample set: samples from July 2016 to December 2021 (29,000); 2) Test sample set: samples from January 2022 to June 2022 (4244).

[0064] (3) Construct the input of the forecast model. Use the past 6 consecutive frames of radar echo matrix (s1_data) as the input of the ENetwork model, output the next 10 frames of radar echo (1 hour), and decompose the input of the model into the speed V and intensity S components of the movement through evolutionary coding. These two components are calculated by the evolutionary operator to obtain the preliminary echo extrapolation and the preliminary echo extrapolation moment s2_data. s1_data and s2_data are used together to construct the final forecast model input matrix s3_data with a total of 16 channel data. At this time, the dimension of the s3_data tensor is (batchsize, 16, 256, 256), where batchsize is the batch size of the training input, 16 is 16 channel data, and 256 is the width and height of the data.

[0065] (4) Construct the EAAUnet radar extrapolation prediction model. The 16-channel input matrix s3_data is used as the input of the EAAUnet model, which is further trained and the optimal model is saved. During the model training, different loss functions are used to train the generator and discriminator of the model:

[0066] (a) Generator loss function design: The generator loss function adopts a weighted loss function. That is, according to the intensity of radar echo, the first-level echo intensity is 0-15dBZ, the second-level echo intensity is 15-25dBZ, the third-level echo intensity is 25-35dBZ, the fourth-level echo intensity is 35-40dBZ, the fifth-level echo intensity is 40-45dBZ, the sixth-level echo intensity is 45-50dBZ, the seventh-level echo intensity is 50-55dBZ, the eighth-level echo intensity is 55-65dBZ, and the ninth-level echo intensity is greater than or equal to 65dBZ. The weights assigned to each level are: first-level echo intensity: 1, second-level echo intensity: 2, third-level echo intensity: 5, fourth-level echo intensity: 10, fifth-level echo intensity: 20, sixth-level echo intensity: 50, seventh-level echo intensity: 60, eighth-level echo intensity: 90, and ninth-level echo intensity: 100.

[0067] The final loss function is:

[0068]

[0069] (b) Discriminator loss function design: The discriminator loss function adopts a power spectrum loss function (PSDLoss), which is a loss function used in the field of signal processing and system identification. Especially in the model training process, when the frequency characteristics of the signal need to be optimized, this loss function measures the difference between the power spectrum density (PSD) of the model output signal and the PSD of the target signal. Power spectrum density is the distribution of signal energy in the frequency domain and is usually used to describe the frequency characteristics of the signal. PSDLoss calculates the loss by comparing the PSD of the model output signal with the PSD of the real signal or the expected signal, thereby prompting the model to learn the correct frequency response. Its basic expression is:

[0070] L PSD =||P model (f)-P target (f)|| 2

[0071] Among them, L PSD is the power spectral density loss; P model (f) is the power spectral density P of the model output signal target (f) is the power spectral density of the target signal; ||·|| represents a certain norm, and L2 is the Euclidean norm.

[0072] (5) Calculate the latest forecast sample. For the precipitation forecast for the next hour at 08:00 on July 29, 2023, 08:00 on July 29, 2023 is taken as the starting time, and the past 6 frames of radar echo puzzles at that time are obtained from the SWAN system (i.e., 07:30 on July 29, 2023, 07:36 on July 29, 2023, 07:42 on July 29, 2023, 07:48 on July 29, 2023, 07:54 on July 29, 2023, and 08:00 on July 29, 2023, a total of 6 frames of radar echo). Since the grid data size of the SWAN puzzle is 900×1200, the shape of the model input X matrix at this time is (6, 900, 1200), where 6 is the number of frames of the past 6 consecutive echo frames, and 900 and 1200 are the height and width of the radar echo.

[0073] (6) Extrapolation using the trained model. First, the model input X in (6) is divided into 20 (6, 256, 256) tensors according to the overlapping segmentation method in (2), and each of them is input into the trained model to obtain the corresponding 20 (10, 256, 256) radar echo extrapolations. The 20 (10, 256, 256) radar echo extrapolations are further reversely spliced, and the average value of the overlapping grid area is calculated. Finally, the radar echo extrapolation of the entire area for the next 10 frames (10, 900, 1200) is obtained. That is, the radar echo for the next hour (each frame of radar echo is 6 minutes apart, and 10 frames is 1 hour).

[0074] (7) Radar echo inversion precipitation calculation. The ZR relationship is used to invert radar echo precipitation, calculate the precipitation at the grid point in the next hour, and complete the forecast of precipitation at the grid point in the next hour. Figure 3 and Figure 4 This is a comparison chart between the forecast using this method and the actual situation.

[0075] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A precipitation forecasting method based on radar extrapolation and ZR relationship of evolutionary network and Unet network, characterized in that: The following steps are involved: Acquire historical real-time rainfall data for each hour, and construct a radar echo sample data set based on the historical real-time rainfall data; Constructing a radar echo short-time nowcasting extrapolation prediction model; the generator of the radar echo short-time nowcasting extrapolation prediction model includes an evolutionary network module and a Unet variant module; Constructing a model training loss function, and training the radar echo short-time nowcasting extrapolation prediction model based on the radar echo sample data set and the model training loss function; The radar jigsaw echo one hour before the prediction time is obtained, and the trained radar echo short-term nowcasting prediction model is input to obtain the predicted radar echo, and the precipitation grid forecast field is obtained based on the ZR relationship and the predicted radar echo.

2. The method for nowcasting precipitation based on radar extrapolation and ZR relationship of evolutionary network and Unet network according to claim 1 is characterized in that: The real-time rainfall data includes the real-time rainfall observation data of the ground meteorological station and the real-time observation data of the weather radar; the real-time observation data of the weather radar is a radar combined reflectivity puzzle.

3. The method for nowcasting precipitation based on radar extrapolation and ZR relationship of evolutionary network and Unet network according to claim 2 is characterized in that: The process of constructing a radar echo sample dataset includes: Each hour within the time range corresponding to the historical real-time rainfall data is used as the starting time of precipitation in the next hour, and the radar combined reflectivity puzzle and precipitation of each 6-minute frame one hour before the starting time are obtained; the radar combined reflectivity puzzles with the top six digits sorted by the time interval between the acquisition time and the starting time are extracted, and overlapped segmentation is performed, and the segmented grids are filtered according to the proportion of echo intensity. After all the screening is completed, the radar echo sample data set is obtained by combining it with the precipitation.

4. The method for nowcasting precipitation based on radar extrapolation and ZR relationship of evolutionary network and Unet network according to claim 1, characterized in that: The radar echo sample data set is decomposed into two components by the evolutionary network module, and the evolutionary operator is calculated on the two components to obtain preliminary radar echo extrapolation data; the radar echo sample data set and the preliminary radar echo extrapolation data are used as inputs of the Unet variant module.

5. The method for nowcasting precipitation based on radar extrapolation and ZR relationship of evolutionary network and Unet network according to claim 4 is characterized in that: On the basis of Unet, an attention mechanism, a hole convolution module and a residual module are introduced to obtain the Unet variant module.

6. The method for nowcasting precipitation based on radar extrapolation and ZR relationship of evolutionary network and Unet network according to claim 1, characterized in that: The model training loss function includes a weighted loss function and a power spectrum loss function; the model generator training adopts the weighted loss function; the model discriminator training adopts the power spectrum loss function.

7. The method for nowcasting precipitation based on radar extrapolation and ZR relationship of evolutionary network and Unet network according to claim 6, characterized in that: The weighted loss function is: Among them, WeightedLoss is the total loss of the weighted loss function; N is the total number of samples; ω i is the weight of the i-th sample, usually used to adjust the loss of the minority class; y i The true label of the i-th sample; The predicted label of the i-th sample.

8. The method for nowcasting precipitation based on radar extrapolation and ZR relationship of evolutionary network and Unet network according to claim 6, characterized in that: The power spectrum loss function is: L PSD =||P model (f)-P target (f)|| 2 Among them, L PSD is the power spectral density loss; P model (f) is the power spectral density of the model output signal; P target (f) is the power spectral density of the target signal, and ||·|| represents the norm.

Images