Method and device for simulating combined reflectivity of ground-based radar, electronic equipment and medium

Abstract

The invention discloses a method and device for simulating the combined reflectivity of a ground-based radar, electronic equipment and a medium, and relates to the technical field of meteorological monitoring. The method comprises the following steps: acquiring satellite full-disk data, first longitude and latitude data and global atmosphere reanalysis data, wherein the global atmosphere reanalysis data comprises original surface temperature data and second longitude and latitude data; the satellite full-disc data, the first longitude and latitude data, the original surface temperature data and the second longitude and latitude data are processed, splicing data of a preset dimension are obtained, and the splicing data comprise target surface temperature data, target infrared channel data in the satellite full-disc data and bright temperature difference channel data; the brightness temperature difference channel data is calculated based on the target infrared channel data; and the spliced data is standardized and then input into a pre-trained deep learning model to obtain simulated ground-based radar combined reflectivity data, and the simulated ground-based radar combined reflectivity data is restored to an original value, so that the radar data quality is improved, and the usability is improved.

Description

Technical Field

[0001] The present application relates to the field of meteorological monitoring technology, and in particular to a method, device, electronic equipment and medium for simulating ground-based radar combined reflectivity. Background Art

[0002] There are many types of disastrous weather, such as thunderstorms, squall lines, tornadoes and hail, which are generated by atmospheric convection and have highly nonlinear characteristics. Such weather has a short life cycle, strong suddenness and is difficult to predict, which can easily cause various meteorological disasters. Therefore, accurately monitoring and identifying the formation and development of disastrous weather is crucial to improving the ability to defend against meteorological disasters and reducing loss of life and economy.

[0003] In order to effectively respond to sudden disastrous weather, my country has started to build a new generation of Doppler weather radar networks since 1998. Currently, about 240 weather radar stations have been deployed across the country. However, due to terrain and economic constraints, radar stations in remote areas are sparsely distributed, and electromagnetic waves will attenuate when propagating in the atmosphere, which affects the spatiotemporal continuity of radar data and reduces its reference value. Summary of the invention

[0004] In view of the above-mentioned defects or deficiencies in the related art, it is desirable to provide a method, device, electronic device and medium for simulating ground-based radar combined reflectivity, which can improve the quality of radar data and enhance the usability.

[0005] In a first aspect, the present application provides a method for simulating ground-based radar combined reflectivity, the method comprising:

[0006] Acquire satellite full disk data and first longitude and latitude data and global atmospheric reanalysis data, wherein the global atmospheric reanalysis data includes original surface temperature data and second longitude and latitude data, wherein the satellite full disk data is mapped one-to-one with the first longitude and latitude data, and the original surface temperature data is mapped one-to-one with the second longitude and latitude data;

[0007] Processing the satellite full disk data and the first longitude and latitude data as well as the original surface temperature data and the second longitude and latitude data to obtain spliced ​​data of a preset dimension, wherein the spliced ​​data includes target surface temperature data, target infrared channel data and bright temperature difference channel data in the satellite full disk data, wherein the bright temperature difference channel data is calculated based on the target infrared channel data;

[0008] The spliced ​​data is standardized and then input into a pre-trained deep learning model to obtain simulated ground-based radar combined reflectivity data, and the simulated ground-based radar combined reflectivity data is restored to the original value, and the deep learning model includes a CREFUNet model.

[0009] Optionally, in some embodiments of the present application, the processing of the satellite full disk data and the first longitude and latitude data and the original surface temperature data and the second longitude and latitude data to obtain spliced ​​data of a preset dimension includes:

[0010] The satellite full disk data and the first longitude and latitude data are cropped to a preset size, and the target infrared channel data in the satellite full disk data and the third longitude and latitude data corresponding to the target infrared channel data are extracted;

[0011] Taking the third longitude and latitude data as a reference, the original surface temperature data and the second longitude and latitude data are screened to obtain the target surface temperature data matching the third longitude and latitude data, and the brightness temperature difference channel data is calculated using the target infrared channel data;

[0012] The target surface temperature data, the target infrared channel data and the brightness temperature difference channel data are spliced ​​into spliced ​​data of the preset dimension.

[0013] Optionally, in some embodiments of the present application, the method of screening the original surface temperature data and the second longitude and latitude data based on the third longitude and latitude data to obtain the target surface temperature data matching the third longitude and latitude data includes:

[0014] Traversing the longitudes and latitudes of each first longitude and latitude grid corresponding to the third longitude and latitude data in the satellite full disk data and the first longitude and latitude data, and searching for longitudes and latitudes in a preset range in each second longitude and longitude grid of the original surface temperature data and the second longitude and longitude data according to the longitudes and latitudes of the first longitude and longitude grid;

[0015] The distance between the longitude and latitude of the first longitude and longitude grid and the longitude and latitude in the preset range is calculated, and the second longitude and longitude grid corresponding to the longitude and longitude matching group with the smallest distance is selected to obtain the target surface temperature data.

[0016] Optionally, in some embodiments of the present application, the obtaining of satellite full disk data and first latitude and longitude data and global atmospheric reanalysis data further includes:

[0017] Obtain the latitude and longitude data of ground-based radar network;

[0018] The satellite full disk data and the first longitude and latitude data are matched according to the longitude and latitude data of the ground-based radar network.

[0019] Optionally, in some embodiments of the present application, the deep learning model is trained by the following steps:

[0020] Constructing model input data and model label data, and dividing the model input data and model label data into a training set and a validation set, the model input data including sample surface temperature data, sample infrared channel data and brightness temperature difference channel data, the model label data including sample ground-based radar combined reflectivity data, wherein the model input data matches the model label data;

[0021] The training set is used to train the model parameters, and the verification set is used to verify the model performance to obtain the CREFUNet model.

[0022] Optionally, in some embodiments of the present application, constructing model input data and model label data, and dividing the model input data and model label data into a training set and a validation set, includes:

[0023] Performing temporal resolution matching and spatial resolution matching on the model input data and the model label data;

[0024] The training set and validation set are generated by replacing outliers in the model input data and the model label data and performing data standardization and dimension correction operations.

[0025] Optionally, in some embodiments of the present application, the performing time resolution matching on the model input data and the model label data includes:

[0026] Based on the time of the sample ground-based radar combined reflectivity data, the starting time corresponding to the sample infrared channel data and the brightness temperature difference channel data with the smallest time difference and the time corresponding to the sample surface temperature data are matched.

[0027] In a second aspect, the present application provides a device for simulating ground-based radar combined reflectivity, the device comprising:

[0028] an acquisition module configured to acquire satellite full disk data and first longitude and latitude data and global atmospheric reanalysis data, wherein the global atmospheric reanalysis data includes original surface temperature data and second longitude and latitude data, wherein the satellite full disk data is mapped one-to-one with the first longitude and latitude data, and the original surface temperature data is mapped one-to-one with the second longitude and latitude data;

[0029] a processing module configured to process the satellite full disk data and the first longitude and latitude data and the original surface temperature data and the second longitude and latitude data to obtain spliced ​​data of a preset dimension, wherein the spliced ​​data includes target surface temperature data, target infrared channel data and bright temperature difference channel data in the satellite full disk data, and the bright temperature difference channel data is calculated based on the target infrared channel data;

[0030] The simulation module is configured to standardize the spliced ​​data and input the data into a pre-trained deep learning model to obtain simulated ground-based radar combined reflectivity data, and restore the simulated ground-based radar combined reflectivity data to the original value, wherein the deep learning model includes a CREFUNet model.

[0031] In a third aspect, the present application provides an electronic device, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the instruction, the program, the code set or the instruction set is loaded and executed by the processor to implement the steps of any one of the methods described in the first aspect.

[0032] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any one of the methods described in the first aspect.

[0033] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0034] The embodiments of the present application provide a method, device, electronic device and medium for simulating the combined reflectivity of a ground-based radar. By integrating the target surface temperature data, the target infrared channel data and the brightness temperature difference channel data in the satellite full disk data, and inputting these data into a pre-trained deep learning model, the simulated ground-based radar combined reflectivity data can be automatically obtained. Since the full disk data is obtained by direct remote sensing using a satellite in the high altitude, the influence of signal attenuation and noise can be reduced. At the same time, the surface temperature data and the brightness temperature difference channel data with cloud top glacier characteristics are added, which can avoid the misjudgment caused by the brightness temperature value in the low temperature area being similar to the brightness temperature value in the echo or precipitation area, thereby effectively improving the radar data quality and greatly improving the availability. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 A schematic flow chart of a method for simulating ground-based radar combined reflectivity provided in an embodiment of the present application;

[0037] Figure 2 A schematic diagram of a deep learning network structure provided in an embodiment of the present application;

[0038] Figure 3A structural block diagram of a device for simulating ground-based radar combined reflectivity provided in an embodiment of the present application;

[0039] Figure 4 A structural block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on 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.

[0041] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0042] 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. Figures 1 to 4 The method, device, electronic device and medium for simulating ground-based radar combined reflectivity provided by the embodiments of the present application are described in detail.

[0043] Terminology explanation:

[0044] ① Ground-based radar combined reflectivity data: It comes from the MOC radar mosaic network combined reflectivity data in the TianQing meteorological big data cloud platform. Similar products also include basic reflectivity, vertical cumulative liquid water content, and 1-hour cumulative precipitation data. Among them, the radar mosaic network combined reflectivity data covers the entire China region, covering 3.87°N~53.55°N and 73.66°E~135.03°E, with equal longitude and latitude projection, time resolution of 6min, and spatial resolution of 1KM.

[0045] ② Satellite full disk data: It comes from the Geostationary Radiation Imager (AGRI) of Fengyun-4B (FY4B). The Geostationary Radiation Imager has 13 observation bands, the first 6 bands are visible light bands, and the last 9 bands are infrared bands. Among them, the FY4B / AGRI L1 full disk data has five spatial resolutions of 250m, 500m, 1KM, 2KM and 4KM, while the infrared channel data has only 4KM spatial resolution.

[0046] ③ Global Atmospheric Reanalysis Data (ERA5): The fifth generation of atmospheric reanalysis data sets from the European Centre for Medium-Range Weather Forecasts (ECMWF) for global climate observations from January 1950 to the present. It is produced by the Copernicus Climate Change Service (C3S) of the European Centre for Medium-Range Weather Forecasts and provides hourly estimates of a large number of atmospheric, land and ocean climate variables. These data cover the Earth on a 30-kilometer grid and use 137 data from the surface to 80 kilometers to resolve the atmosphere, including uncertainty information for all variables when reducing spatial and temporal resolution. Global atmospheric reanalysis data combines model data with observational data from around the world to form a globally complete and consistent data set.

[0047] Please refer to Figure 1 , which is a flow chart of a method for simulating ground-based radar combined reflectivity provided in an embodiment of the present application, the method specifically comprises the following steps:

[0048] S101, obtaining satellite full disk data, first longitude and latitude data, and global atmospheric reanalysis data, where the global atmospheric reanalysis data includes original surface temperature data and second longitude and latitude data.

[0049] Among them, the satellite full disk data is mapped one-to-one with the first longitude and latitude data, that is, the satellite full disk data and the first longitude and latitude data have a point-to-point correspondence. Similarly, the original surface temperature data and the second longitude and latitude data are mapped one-to-one, that is, the original surface temperature data and the second longitude and latitude data also have a point-to-point correspondence.

[0050] S102, processing the satellite full disk data and the first longitude and latitude data as well as the original surface temperature data and the second longitude and latitude data to obtain stitching data of a preset dimension, wherein the stitching data includes target surface temperature data, target infrared channel data and bright temperature difference channel data in the satellite full disk data, and the bright temperature difference channel data is calculated based on the target infrared channel data.

[0051] Exemplarily, after determining the range of longitude and latitude to be simulated, fine-tuning is performed to meet the requirements of the deep learning model for the input data dimension. Therefore, the embodiment of the present application can first (i) crop the satellite full disk data and the first longitude and latitude data to a preset size, and extract the target infrared channel data in the satellite full disk data and the third longitude and latitude data corresponding to the target infrared channel data, for example, the preset size is 880*1280. Then (ii), based on the third longitude and latitude data, the original surface temperature data and the second longitude and latitude data are screened to obtain the target surface temperature data matching the third longitude and longitude data, and the target infrared channel data is used to calculate the brightness temperature difference channel data. For example, when obtaining the target surface temperature data, the longitude and latitude of each first longitude and latitude grid corresponding to the third longitude and longitude data is first traversed in the satellite full disk data and the first longitude and longitude data, and according to the longitude and latitude of the first longitude and longitude grid, the second longitude and longitude grids of the original surface temperature data and the second longitude and longitude data are searched for the target surface temperature data that is in the preset range. Assume the longitude and latitude of the range, where the preset range corresponding to the surface temperature data is the absolute value of the difference of 0.26°, and then calculate the distance between the longitude and latitude of the first longitude and latitude grid and the longitude and latitude in the preset range, and select the second longitude and latitude grid corresponding to the longitude and latitude matching group with the smallest distance to obtain the target surface temperature data. For the longitude and latitude that are not in the preset range, skip it. The advantage of this setting is that it has the advantages of no interpolation, no change to the original data, and no need to consider different projection methods between different data, and has strong adaptability. The longitude and latitude are converted into distances by the following formula (1), that is,

[0052] lat delta = Rad(lat1)-Rad(lat2);

[0053] lon delta = Rad(lon1)-Rad(lon2);

[0054]

[0055] In formula (1), D represents the distance between two points, lon1 represents the longitude of the first longitude and latitude grid, lat1 represents the latitude of the first longitude and latitude grid, lon2 represents the longitude in the preset range, lat2 represents the latitude in the preset range, Rad represents the angle-to-radian operation, lon delta Indicates the difference in longitude in radians, lat delta It represents the difference between radians and latitudes, temp represents the haversine operation, asin represents the inverse sine function, and ER represents the radius of the earth, and its value is 6371 km.

[0056] For example, when obtaining the brightness temperature difference channel data, the following formula (2) is used for calculation:

[0057] BTD1 = BT7-BT3;

[0058] BTD2 = BT8 + BT6 - 2 * BT7; (2)

[0060] In formula (2), BTD1 represents the first brightness temperature difference channel, BTD2 represents the second brightness temperature difference channel, BT3 represents the brightness temperature at the center wavelength of 6.25 μm, BT6 represents the brightness temperature at the center wavelength of 8.55 μm, BT7 represents the brightness temperature at the center wavelength of 10.8 μm, and BT8 represents the brightness temperature at the center wavelength of 12 μm.

[0061] It should be noted that the attributes of 12 channel data, including 1 surface temperature data, 9 infrared channel data and 2 brightness temperature difference channel data, are shown in Table 1 below.

[0062] Table 1 Attributes of channel data

[0063] Channel Name Channel Description Main range of values Channel characteristics BT1 Central wavelength 3.75 (high) μm brightness temperature 200～350K High albedo targets such as clouds BT2 Central wavelength 3.75 (low) μm brightness temperature 200～350K Low albedo target, ground surface BT3 Brightness temperature at central wavelength 6.25μm 200～350K High-level water vapor BT4 Brightness temperature at central wavelength 6.95μm 200～350K Middle layer water vapor BT5 Central wavelength 7.92μm brightness temperature 200～350K Low-level water vapor BT6 Brightness temperature at central wavelength 8.55μm 200～350K Total water vapor, clouds BT7 Brightness temperature at central wavelength 10.8μm 200～350K Clouds, surface temperature, etc. BT8 Brightness temperature at central wavelength 12μm 200～350K Clouds, total water vapor, surface temperature BT9 Brightness temperature at central wavelength 13.3μm 200～350K Clouds, water vapor BTD1 Brightness temperature difference associated with BT3 and BT7 -10～70K Tropospheric cloud top height BTD2 Brightness temperature difference associated with BT6, BT7, and BT8 -10～10K Yunding Glacier SKT ERA5 surface temperature data 240～300K ---

[0064] Finally (iii), the target surface temperature data, the target infrared channel data and the brightness temperature difference channel data are spliced ​​into spliced ​​data of a preset dimension, for example, the 12 channels recorded in Table 1 are sequentially spliced ​​into spliced ​​data of 880*1280*12 dimensions.

[0065] For example, when obtaining the satellite full disk data and the first longitude and latitude data and the global atmospheric reanalysis data, the embodiment of the present application can also obtain the longitude and latitude data of the ground-based radar network, that is, determine the longitude and latitude range to be simulated based on the longitude and latitude data of the ground-based radar network, and match the satellite full disk data and the first longitude and longitude data according to the longitude and longitude data of the ground-based radar network, wherein the preset range corresponding to the longitude and latitude data of the ground-based radar network is the absolute value of the difference of 0.01°, that is, all points with longitude and latitude differences greater than 0.01° are removed during matching. After this matching, a piece of data consistent with the longitude and latitude range of the ground-based radar network can be obtained, and then the matched longitude and latitude are used to match the longitude and latitude corresponding to the original surface temperature data, and the target surface temperature data is obtained. For the matching principle, please refer to the above records, which will not be repeated here. The advantage of such a setting is that the subsequently obtained simulated ground-based radar combined reflectivity data can correspond one-to-one with the actual ground-based radar combined reflectivity data, which is convenient for statistical errors.

[0066] S103, standardize the spliced ​​data and input the data into a pre-trained deep learning model to obtain simulated ground-based radar combined reflectivity data, and restore the simulated ground-based radar combined reflectivity data to the original value, where the deep learning model includes a CREFUNet model.

[0067] It should be noted that when training a deep learning model, the embodiment of the present application first (I) constructs model input data and model label data, and divides the model input data and model label data into a training set and a validation set, wherein the model input data includes sample surface temperature data, sample infrared channel data, and bright temperature difference channel data, and the model label data includes sample ground-based radar combined reflectivity data, wherein the model input data matches the model label data. For example, the model input data and the model label data are matched with time resolution and spatial resolution, and then the abnormal values ​​in the model input data and the model label data are replaced, and after data standardization and dimension correction operations, a training set and a validation set are generated. Further, in the process of time resolution matching, in combination with Table 2, based on the principle of the nearest time matching, the time of the sample ground-based radar combined reflectivity data is used as a reference, and the starting time corresponding to the sample infrared channel data and the bright temperature difference channel data with the smallest time difference and the time corresponding to the sample surface temperature data are matched. If the time difference is the same, the data of the previous time is selected. Since the surface temperature data will not change much within 1 hour, the maximum time error caused by matching data with a time resolution of 6 minutes will not exceed 40 minutes. Therefore, the time matching of the sample ground-based radar combined reflectivity data and the surface temperature data is reasonable, and the spatial resolution matching principle is similar to the matching process of the first longitude and latitude grid and the second longitude and latitude grid mentioned above, which will not be repeated here. In this way, point-to-point matching of the model input data and the model label data is achieved.

[0068] Table 2 Data resolution

[0069] Data name Temporal resolution Spatial resolution Ground-based radar combined reflectivity data 6min 1KM FY4B / AGRI L1 full disc data 15min 4KM ERA5 surface temperature data 60min 30KM

[0070] Furthermore, since the model label data contains abnormal values ​​such as undetected values ​​and no-echo values, where the undetected value represents a fill value that is not within the detection range of the ground-based radar (Nan in the sample ground-based radar combined reflectivity data), and the no-echo value represents that it is within the detection range, but no echo is received at this location (-128 in the sample ground-based radar combined reflectivity data), and if the model training data contains Nan values, it will affect the model training, making the simulation results all Nan values, therefore, in the embodiment of the present application, the fill value and no-echo value of the model label data are both set to -9999, and at the same time, the fill value position of the model label data is obtained, and the model input data at the same position is set to -9999, so that this part of the data does not participate in the model training.

[0071] However, since the numerical distribution ranges of the input data of the model are different and the numerical values ​​are too large, it will make the model difficult to converge. Therefore, the embodiment of the present application shrinks the numerical range of the data to the interval of 0 to 1 through the following formula (3), that is,

[0072]

[0073] In formula (3), y represents the normalized value, x represents the original value, and x min Indicates the minimum value of the main numerical range (see Table 1 for the main numerical range), x max Indicates the maximum value of the main numerical range (see Table 1 for the main numerical range). This range is set according to the range of most data in the data, and does not rule out the existence of valid data outside the range. Finally, -9999 in the model input data and model label data are changed to 0, avoiding the disadvantage that most model prediction values ​​are negative.

[0074] Also, since the UNet network has requirements for the input data dimension, the input data dimension cannot change and must be a multiple of 10, the embodiment of the present application cuts the 886*1283 data dimension into 880*1280, that is, cuts off the last 6 rows and 3 columns. Finally, the training data uses 1468 hours of data from March 2023 to January 2024, including the 3rd, 9th, 15th, 19th, 23rd, and 27th of each month, a total of 6 days. The training set has 1200 hours of data, and the validation set has 268 hours of data.

[0075] Then (II), the model parameters are trained using the training set, and the model performance is verified using the validation set to obtain the CREFUNet model. Since the CREFUnet model uses the UNet network framework, in order to ensure that the complete combined reflectivity data can be simulated without a sense of cutting, the CREFUnet model in the embodiment of the present application uses the complete Chinese region data as input, and the dimension is 880*1280*12, and the network structure is as follows Figure 2 As shown in Figure 1, the CREFUnet model includes a 3*3 2D convolution + RELU activation function, a 2*2 max pooling layer, a 2*2 2D deconvolution layer, a connection layer, and a full connection layer. For example, the model parameters of the CREFUNet network are initialized, and the data of one time is taken as a batch, and is sequentially input into the CREFUNet network for calculation to obtain the simulation results. Secondly, the error between the simulation result and the corresponding time model label data is calculated by the mean square error loss function shown in the following formula (4), that is,

[0076]

[0077] In formula (4), MSE represents the mean square error loss function, Represents the simulation results, and y represents the model label data at the corresponding time.

[0078] Again, the calculated loss is input to the Adam optimizer to optimize the model parameters of the CREFUNet network. It should be noted that the Adam optimizer is an adaptive learning rate optimizer based on the gradient descent algorithm, which adaptively adjusts the learning rate of each parameter according to the historical gradient and update of each parameter during the training process, thereby accelerating the neural network training. In the embodiment of the present application, the learning rate is set to 0.0001. If the learning rate is set too high, the model will overfit, and if it is set too low, the convergence speed of the model will slow down. When a training round is completed, the model parameters of the CREFUNet network are saved as the model trained in this round, and the simulation loss is verified with the data of the validation set to evaluate the model performance. If the simulation loss no longer decreases, it indicates that the model has converged, the training is completed and the final model is saved. The training rounds in the embodiment of the present application are 100 rounds.

[0079] Exemplarily, the embodiment of the present application can obtain the original value of the simulated ground-based radar combined reflectivity data through the inverse transformation formula (5) of formula (3), that is,

[0080] x=y*(x max -x min )+x min (5)

[0081] Accordingly, x max The corresponding combined reflectivity maximum threshold is 70, x min The corresponding minimum threshold of combined reflectivity is 0.

[0082] The method for simulating ground-based radar combined reflectivity provided in the embodiment of the present application can automatically obtain simulated ground-based radar combined reflectivity data by integrating target surface temperature data, target infrared channel data and brightness temperature difference channel data in satellite full disk data, and inputting these data into a pre-trained deep learning model. Since the full disk data is obtained by direct remote sensing using satellites in the high altitude, the influence of signal attenuation and noise can be reduced. At the same time, the addition of surface temperature data and brightness temperature difference channel data with cloud top glacier characteristics can avoid misjudgment caused by brightness temperature values ​​in low temperature areas being similar to brightness temperature values ​​in areas with echoes or precipitation, thereby effectively improving the quality of radar data and greatly improving availability.

[0083] Based on the above embodiments, the present application provides a device for simulating ground-based radar combined reflectivity. The device 100 for simulating ground-based radar combined reflectivity can be applied to Figure 1-2 In the method for simulating the combined reflectivity of ground-based radar in the corresponding embodiment. Figure 3 The device 100 for simulating ground-based radar combined reflectivity comprises:

[0084] An acquisition module 101 is configured to acquire satellite full disk data and first longitude and latitude data and global atmospheric reanalysis data, wherein the global atmospheric reanalysis data includes original surface temperature data and second longitude and latitude data, wherein the satellite full disk data is mapped one-to-one with the first longitude and latitude data, and the original surface temperature data is mapped one-to-one with the second longitude and latitude data;

[0085] The processing module 102 is configured to process the satellite full disk data, the first longitude and latitude data, the original surface temperature data, and the second longitude and latitude data to obtain spliced ​​data of a preset dimension, wherein the spliced ​​data includes target surface temperature data, target infrared channel data and bright temperature difference channel data in the satellite full disk data, and the bright temperature difference channel data is calculated based on the target infrared channel data;

[0086] The simulation module 103 is configured to standardize the spliced ​​data and input it into a pre-trained deep learning model to obtain simulated ground-based radar combined reflectivity data, and restore the simulated ground-based radar combined reflectivity data to the original value. The deep learning model includes a CREFUNet model.

[0087] Optionally, in some embodiments of the present application, the processing module 102 is specifically used to crop the satellite full disk data and the first longitude and latitude data to a preset size, and extract the target infrared channel data in the satellite full disk data and the third longitude and latitude data corresponding to the target infrared channel data;

[0088] Taking the third longitude and latitude data as a reference, the original surface temperature data and the second longitude and latitude data are screened to obtain target surface temperature data matching the third longitude and latitude data, and the bright temperature difference channel data is calculated using the target infrared channel data;

[0089] The target surface temperature data, target infrared channel data and brightness temperature difference channel data are stitched into stitched data of preset dimensions.

[0090] Optionally, in some embodiments of the present application, the processing module 102 is further specifically configured to traverse the longitudes and latitudes of each first longitude and latitude grid corresponding to the third longitude and latitude data in the satellite full disk data and the first longitude and latitude data, and search for longitudes and latitudes within a preset range in each second longitude and longitude grid of the original surface temperature data and the second longitude and longitude data according to the longitudes and latitudes of the first longitude and longitude grid;

[0091] The distance between the longitude and latitude of the first longitude and longitude grid and the longitude and latitude in a preset range is calculated, and the second longitude and longitude grid corresponding to the longitude and longitude matching group with the smallest distance is selected to obtain the target surface temperature data.

[0092] Optionally, in some embodiments of the present application, the acquisition module 101 is also specifically used to acquire the longitude and latitude data of the ground-based radar network; and match the satellite full disk data and the first longitude and latitude data according to the longitude and latitude data of the ground-based radar network.

[0093] Optionally, in some embodiments of the present application, the simulation module 103 is further specifically used to construct model input data and model label data, and divide the model input data and model label data into a training set and a validation set, the model input data includes sample surface temperature data, sample infrared channel data and brightness temperature difference channel data, and the model label data includes sample ground-based radar combined reflectivity data;

[0094] The training set is used to train the model parameters, and the validation set is used to verify the model performance to obtain the CREFUNet model.

[0095] Optionally, in some embodiments of the present application, the simulation module 103 is also specifically used to perform temporal resolution matching and spatial resolution matching on the model input data and the model label data; replace outliers in the model input data and the model label data, and generate a training set and a validation set after data standardization and dimensionality correction operations.

[0096] Optionally, in some embodiments of the present application, the simulation module 103 is also specifically used to match the starting time corresponding to the sample infrared channel data and the brightness temperature difference channel data with the smallest time difference and the time corresponding to the sample surface temperature data, based on the time of the sample ground-based radar combined reflectivity data when matching the time resolution of the model input data and the model label data.

[0097] It should be noted that, for the description of the same steps and the same contents in this embodiment as those in other embodiments, reference can be made to the description in other embodiments and will not be repeated here.

[0098] The device for simulating the combined reflectivity of ground-based radar provided in the embodiment of the present application can automatically obtain the simulated ground-based radar combined reflectivity data by integrating the target surface temperature data, the target infrared channel data and the brightness temperature difference channel data in the satellite full disk data, and inputting these data into a pre-trained deep learning model. Since the full disk data is obtained by direct remote sensing using a satellite in the high altitude, the signal attenuation and noise effects can be reduced. At the same time, the addition of surface temperature data and brightness temperature difference channel data with cloud top glacier characteristics can avoid misjudgment caused by the brightness temperature values ​​in low temperature areas being similar to the brightness temperature values ​​in areas with echoes or precipitation, thereby effectively improving the quality of radar data and greatly improving availability.

[0099] Based on the above embodiments, the present application provides an electronic device. Figure 4The electronic device 200 may include a processor 201 and a memory 202. The memory 202 stores at least one instruction, at least one program, code set or instruction set, which is loaded and executed by the processor 201 to implement Figure 1-2 The corresponding embodiment simulates the method steps of the ground-based radar combined reflectivity.

[0100] As another aspect, the present application provides a computer-readable storage medium for storing program code, the program code for executing the aforementioned Figure 1-2 Any implementation manner of the method for simulating ground-based radar combined reflectivity in the corresponding embodiment.

[0101] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0102] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0103] In addition, each functional module in each embodiment of the present application may be integrated into a processing unit, or each module may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium.

[0104] Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the whole or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling 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 method for simulating the combined reflectivity of ground-based radar in various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0105] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for simulating the combined reflectivity of ground-based radar, It is characterized in that The method comprises: Acquire satellite full disk data and first longitude and latitude data and global atmospheric reanalysis data, wherein the global atmospheric reanalysis data includes original surface temperature data and second longitude and latitude data, wherein the satellite full disk data is mapped one-to-one with the first longitude and latitude data, and the original surface temperature data is mapped one-to-one with the second longitude and latitude data; Processing the satellite full disk data and the first longitude and latitude data as well as the original surface temperature data and the second longitude and latitude data to obtain spliced ​​data of a preset dimension, wherein the spliced ​​data includes target surface temperature data, target infrared channel data and bright temperature difference channel data in the satellite full disk data, wherein the bright temperature difference channel data is calculated based on the target infrared channel data; The spliced ​​data is standardized and then input into a pre-trained deep learning model to obtain simulated ground-based radar combined reflectivity data, and the simulated ground-based radar combined reflectivity data is restored to the original value, and the deep learning model includes a CREFUNet model.

2. The method according to claim 1, It is characterized in that The processing of the satellite full disk data and the first longitude and latitude data and the original surface temperature data and the second longitude and latitude data to obtain spliced ​​data of a preset dimension includes: The satellite full disk data and the first longitude and latitude data are cropped to a preset size, and the target infrared channel data in the satellite full disk data and the third longitude and latitude data corresponding to the target infrared channel data are extracted; Taking the third longitude and latitude data as a reference, the original surface temperature data and the second longitude and latitude data are screened to obtain the target surface temperature data matching the third longitude and latitude data, and the brightness temperature difference channel data is calculated using the target infrared channel data; The target surface temperature data, the target infrared channel data and the brightness temperature difference channel data are spliced ​​into spliced ​​data of the preset dimension.

3. The method according to claim 2, It is characterized in that The method of screening the original surface temperature data and the second longitude and latitude data based on the third longitude and latitude data to obtain the target surface temperature data matching the third longitude and latitude data includes: Traversing the longitudes and latitudes of each first longitude and latitude grid corresponding to the third longitude and latitude data in the satellite full disk data and the first longitude and latitude data, and searching for longitudes and latitudes in a preset range in each second longitude and longitude grid of the original surface temperature data and the second longitude and longitude data according to the longitudes and latitudes of the first longitude and longitude grid; The distance between the longitude and latitude of the first longitude and longitude grid and the longitude and latitude in the preset range is calculated, and the second longitude and longitude grid corresponding to the longitude and longitude matching group with the smallest distance is selected to obtain the target surface temperature data.

4. The method according to any one of claims 1 to 3, It is characterized in that The obtaining of satellite full disk data, first latitude and longitude data and global atmospheric reanalysis data also includes: Obtain the latitude and longitude data of ground-based radar network; The satellite full disk data and the first longitude and latitude data are matched according to the longitude and latitude data of the ground-based radar network.

5. The method according to any one of claims 1 to 3, It is characterized in that The deep learning model is trained through the following steps: Constructing model input data and model label data, and dividing the model input data and model label data into a training set and a validation set, the model input data including sample surface temperature data, sample infrared channel data and brightness temperature difference channel data, the model label data including sample ground-based radar combined reflectivity data, wherein the model input data matches the model label data; The training set is used to train the model parameters, and the verification set is used to verify the model performance to obtain the CREFUNet model.

6. The method according to claim 5, It is characterized in that The constructing of model input data and model label data, and dividing the model input data and model label data into a training set and a validation set, includes: Performing temporal resolution matching and spatial resolution matching on the model input data and the model label data; The training set and validation set are generated by replacing outliers in the model input data and the model label data and performing data standardization and dimension correction operations.

7. The method according to claim 6, It is characterized in that The performing time resolution matching on the model input data and the model label data comprises: Based on the time of the sample ground-based radar combined reflectivity data, the starting time corresponding to the sample infrared channel data and the brightness temperature difference channel data with the smallest time difference and the time corresponding to the sample surface temperature data are matched.

8. A device for simulating the combined reflectivity of ground-based radar, It is characterized in that The device comprises: an acquisition module configured to acquire satellite full disk data and first longitude and latitude data and global atmospheric reanalysis data, wherein the global atmospheric reanalysis data includes original surface temperature data and second longitude and latitude data, wherein the satellite full disk data is mapped one-to-one with the first longitude and latitude data, and the original surface temperature data is mapped one-to-one with the second longitude and latitude data; a processing module configured to process the satellite full disk data and the first longitude and latitude data and the original surface temperature data and the second longitude and latitude data to obtain spliced ​​data of a preset dimension, wherein the spliced ​​data includes target surface temperature data, target infrared channel data and bright temperature difference channel data in the satellite full disk data, and the bright temperature difference channel data is calculated based on the target infrared channel data; The simulation module is configured to standardize the spliced ​​data and input the data into a pre-trained deep learning model to obtain simulated ground-based radar combined reflectivity data, and restore the simulated ground-based radar combined reflectivity data to the original value, wherein the deep learning model includes a CREFUNet model.

9. An electronic device, It is characterized in that The electronic device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the instruction, the program, the code set or the instruction set is loaded and executed by the processor to implement the steps of any one of the methods described in claims 1 to 7.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any one of claims 1 to 7.

Images