Meteorological data processing method and device, electronic equipment and storage medium

Abstract

The invention provides a meteorological data processing method and apparatus, an electronic device and a storage medium. The method comprises the steps of constructing a fusion feature map by fusing multi-source data features; low-level fusion features corresponding to the fusion feature map are extracted, down-sampling is carried out on the fusion feature map, then high-level fusion features corresponding to the fusion feature map are further extracted, and meanwhile the spatial resolution is kept; performing up-sampling on the fusion feature map after the high-level fusion features are extracted to recover spatial resolution, and splicing the fusion feature map after up-sampling and the fusion feature map after down-sampling to determine a spliced feature map; and carrying out convolution operation on the spliced feature map, and converting the spliced feature map into small-scale meteorological data after fusing the low-level fusion features and the high-level fusion features. According to the method, downscaling processing of the meteorological data can be realized by adopting fewer computing resources, and meanwhile, the meteorological data from different data sources are effectively fused, so that the problem of limitation in processing a nonlinear and complex meteorological relationship in the prior art is solved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of meteorological data analysis, and in particular, to a method, apparatus, electronic device, and storage medium for processing meteorological data. Background Art

[0002] The acquisition and processing of meteorological data play a crucial role in environmental science, meteorology, and the study of meteorological changes. Meteorological models usually provide meteorological information on a global or regional scale in a large-scale and low-resolution form. These models can capture macroscopic meteorological patterns and trends, but are insufficient in local details. To meet specific application requirements, such as agricultural planning, urban management, and extreme weather warning, it is necessary to convert these large-scale and low-resolution data into small-scale and high-resolution data, that is, to perform downscaling processing.

[0003] Traditional downscaling methods can be divided into two categories: dynamical downscaling and statistical downscaling. Dynamical downscaling uses regional climate models (RCMs) to perform higher-resolution simulations based on global climate models (GCMs). This method has a high computational cost and extremely high requirements for computing resources. Statistical downscaling uses historical observation data to establish statistical relationships between meteorological variables and applies these relationships to future meteorological scenarios. This method depends on high-quality historical data and has limited performance when facing non-linear and complex meteorological relationships. Multi-source meteorological data (such as temperature, precipitation, humidity, wind speed, etc.) provides rich information, and the effect of downscaling can be improved through data fusion and machine learning methods. However, how to effectively fuse and process these multi-source data remains an urgent problem to be solved. Summary of the Invention

[0004] The embodiments of the present disclosure at least provide a method, apparatus, electronic device, and storage medium for processing meteorological data, and the beneficial effects are as follows: Downscaling processing of meteorological data can be achieved with fewer computing resources, and at the same time, meteorological data from different data sources can be effectively fused, overcoming the limitations in dealing with non-linear and complex meteorological relationships in the prior art.

[0005] The embodiments of the present disclosure provide a method for processing meteorological data, including:

[0006] Obtain multiple types of large-scale meteorological data, extract features corresponding to each of the large-scale meteorological data to construct a preliminary feature map, and construct a fusion feature map according to the feature dependence relationship between the preliminary feature maps;

[0007] Extract low-level fusion features corresponding to the fusion feature map, perform downsampling on the fusion feature map to reduce the spatial resolution of the fusion feature map, and then further extract high-level fusion features corresponding to the fusion feature map while maintaining the spatial resolution;

[0008] Upsample the fused feature map after extracting the high-level fused features to restore the spatial resolution, and splice the upsampled fused feature map with the downsampled fused feature map to determine a spliced feature map;

[0009] Perform a convolution operation on the spliced feature map. After fusing the low-level fused features and the high-level fused features, convert the spliced feature map into small-scale meteorological data.

[0010] In an optional implementation manner, determine the feature dependency based on the following steps:

[0011] Align the preliminary feature maps to the same spatial scale, and use a linear transformation to adjust the preliminary feature maps to the same number of channels;

[0012] Divide the aligned preliminary feature maps into multiple feature blocks;

[0013] Perform a linear transformation on each of the feature blocks to determine corresponding query vectors, key vectors, and value vectors;

[0014] Determine an attention weight based on the dot product between the query vector and the key vector, and use the attention weight as the feature dependency.

[0015] In an optional implementation manner, construct the fused feature map based on the following steps:

[0016] Input the feature blocks into a multi-layer feature fusion network. In each layer, for each of the feature blocks, weight the value vector corresponding to this feature block and the value vectors corresponding to other feature blocks according to the attention weight to determine a hierarchical fused feature vector and generate a corresponding hierarchical fused feature map, where between each layer, the hierarchical fused feature vectors generated at the bottom layer are retained and passed to the upper layer through a residual connection;

[0017] Splice and fuse the hierarchical fused feature maps corresponding to each layer to generate the fused feature map.

[0018] In an optional implementation manner, extract the low-level fused features corresponding to the fused feature map, and downsample the fused feature map to reduce the spatial resolution of the fused feature map. Specifically, it includes:

[0019] Input the fused feature map into the encoder of a preset downscaling model, where the encoder includes a convolutional layer, a residual layer, and a max pooling layer;

[0020] Perform a convolution operation on the fused feature map through the convolutional layer to determine the low-level fused features, and transfer the low-level fused features to the residual layer;

[0021] In the residual layer, superimpose the fused feature map and the low-level fused features through residual connection and transfer them to the max pooling layer;

[0022] Perform downsampling through the max pooling layer, and store the feature map with reduced spatial resolution in the skip connection.

[0023] In an alternative embodiment, extract the high-level fused features corresponding to the fused feature map while maintaining the spatial resolution, specifically including:

[0024] Input the downsampled feature map into the bottleneck layer of the preset downscaling model, and extract the high-level fused features through convolution operation and residual connection;

[0025] Among them, no pooling operation is performed on the bottleneck layer to maintain the spatial resolution.

[0026] In an alternative embodiment, upsample the fused feature map after extracting the high-level fused features to restore the spatial resolution, and splice the upsampled fused feature map with the downsampled fused feature map to determine the spliced feature map, specifically including:

[0027] Input the feature map output by the encoder into the decoder of the preset downscaling model, where the decoder includes a transposed convolutional layer and a feature fusion layer;

[0028] Perform upsampling on the input feature map through the transposed convolutional layer to restore the spatial resolution, and input the upsampled feature map into the skip connection;

[0029] In the skip connection, splice the stored feature map with reduced spatial resolution with the upsampled feature map to determine the spliced feature map.

[0030] In an alternative embodiment, fuse the low-level fused features and the high-level fused features, specifically including:

[0031] Input the spliced feature map into the feature fusion layer, perform a convolution operation on the spliced feature map, and fuse the low-level fused features and the high-level fused features.

[0032] The embodiments of the present disclosure also provide a device for processing meteorological data, including:

[0033] A feature fusion module, configured to obtain various types of large-scale meteorological data, extract features corresponding to each type of the large-scale meteorological data to construct a preliminary feature map, and construct a fusion feature map according to the feature dependence relationship between the preliminary feature maps;

[0034] A downscaling module, configured to extract low-level fusion features corresponding to the fusion feature map, perform downsampling on the fusion feature map to reduce the spatial resolution of the fusion feature map, and then further extract high-level fusion features corresponding to the fusion feature map while maintaining the spatial resolution;

[0035] A resolution enhancement module, configured to upsample the fusion feature map after extracting the high-level fusion features to restore the spatial resolution, and splice the upsampled fusion feature map with the downsampled fusion feature map to determine a spliced feature map;

[0036] A data conversion module, configured to perform a convolution operation on the spliced feature map, and convert the spliced feature map into small-scale meteorological data after fusing the low-level fusion features and the high-level fusion features.

[0037] An embodiment of the present disclosure further provides an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device runs, the processor communicates with the memory through the bus. When the machine-readable instructions are executed by the processor, the above-mentioned method for processing meteorological data or the steps in any possible implementation manner of the above-mentioned method for processing meteorological data are executed.

[0038] An embodiment of the present disclosure further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, the above-mentioned method for processing meteorological data or the steps in any possible implementation manner of the above-mentioned method for processing meteorological data are executed.

[0039] An embodiment of the present disclosure further provides a computer program product, including a computer program / instructions. When the computer program and instructions are executed by a processor, the above-mentioned method for processing meteorological data or the steps in any possible implementation manner of the above-mentioned method for processing meteorological data are implemented.

[0040] A method, apparatus, electronic device, and storage medium for processing meteorological data provided by an embodiment of the present disclosure obtain various types of large-scale meteorological data, extract features corresponding to each of the large-scale meteorological data to construct a preliminary feature map, and construct a fusion feature map according to the feature dependence relationship between the preliminary feature maps; extract low-level fusion features corresponding to the fusion feature map, downsample the fusion feature map to reduce the spatial resolution of the fusion feature map, and then further extract high-level fusion features corresponding to the fusion feature map while maintaining the spatial resolution; upsample the fusion feature map after extracting the high-level fusion features to restore the spatial resolution, and splice the upsampled fusion feature map with the downsampled fusion feature map to determine a spliced feature map; perform a convolution operation on the spliced feature map, and after fusing the low-level fusion features and the high-level fusion features, convert the spliced feature map into small-scale meteorological data. It is possible to implement downscaling processing of meteorological data with fewer computing resources, and at the same time effectively fuse meteorological data from different data sources, overcoming the limitations in processing non-linear and complex meteorological relationships in the prior art.

[0041] To make the above objects, features, and advantages of the present disclosure more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, provides a detailed description as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] To more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the accompanying drawings required for the embodiments. The accompanying drawings herein are incorporated into the specification and constitute a part of this specification. These drawings illustrate embodiments that conform to the present disclosure and, together with the specification, are used to explain the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0043] Figure 1 Shows a flowchart of a method for processing meteorological data provided by an embodiment of the present disclosure;

[0044] Figure 2 Shows a flowchart of a method for constructing a fusion feature map provided by an embodiment of the present disclosure;

[0045] Figure 3 Shows a schematic diagram of an apparatus for processing meteorological data provided by an embodiment of the present disclosure;

[0046] Figure 4 Shows a schematic diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present disclosure. Apparently, the described embodiments are only some, but not all, of the embodiments of the present disclosure. The components of the embodiments of the present disclosure generally described and illustrated in the figures herein can be arranged and designed in a variety of different configurations. Therefore, the detailed description of the embodiments of the present disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed present disclosure, but merely represents selected embodiments of the present disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of the present disclosure without creative efforts fall within the scope of protection of the present disclosure.

[0048] It should be noted that like reference numerals and letters denote like items in the following figures, and thus, once an item is defined in one figure, further definition and explanation thereof are not required in subsequent figures.

[0049] As used herein, the term "and / or" merely describes an association relationship and indicates that three relationships may exist. For example, A and / or B may represent three cases: A alone, both A and B present simultaneously, and B alone. Additionally, the term "at least one" as used herein means any one of a plurality or any combination of at least two of a plurality. For example, including at least one of A, B, and C may mean including any one or more elements selected from the set composed of A, B, and C.

[0050] It has been found through research that traditional downscaling methods can be divided into two categories: dynamical downscaling and statistical downscaling. Dynamical downscaling uses regional climate models (RCMs) to perform higher-resolution simulations based on global climate models (GCMs). This method has a high computational cost and extremely high requirements for computing resources. Statistical downscaling uses historical observational data to establish statistical relationships between meteorological variables and applies these relationships to future meteorological scenarios. This method relies on high-quality historical data and has limited performance when faced with non-linear and complex meteorological relationships. Multi-source meteorological data (such as temperature, precipitation, humidity, wind speed, etc.) provides rich information, and the downscaling effect can be improved through data fusion and machine learning methods. However, how to effectively fuse and process this multi-source data remains an urgent problem to be solved.

[0051] Based on the above research, the present disclosure provides a method, apparatus, electronic device, and storage medium for processing meteorological data. By acquiring various types of large-scale meteorological data, extracting features corresponding to each of the large-scale meteorological data to construct a preliminary feature map, and constructing a fusion feature map according to the feature dependence relationship between the preliminary feature maps; extracting low-level fusion features corresponding to the fusion feature map, downsampling the fusion feature map to reduce the spatial resolution of the fusion feature map, and then further extracting high-level fusion features corresponding to the fusion feature map while maintaining the spatial resolution; upsampling the fusion feature map after extracting the high-level fusion features to restore the spatial resolution, and splicing the upsampled fusion feature map with the downsampled fusion feature map to determine a spliced feature map; performing a convolution operation on the spliced feature map, and after fusing the low-level fusion features and the high-level fusion features, converting the spliced feature map into small-scale meteorological data. It is possible to achieve downscaling processing of meteorological data with fewer computing resources, and at the same time effectively fuse meteorological data from different data sources, overcoming the limitations in processing non-linear and complex meteorological relationships in the prior art.

[0052] To facilitate the understanding of this embodiment, first, a method for processing meteorological data disclosed in the embodiments of the present disclosure will be introduced in detail. The execution subject of the method for processing meteorological data provided in the embodiments of the present disclosure is generally a computer device with certain computing capabilities. Such a computer device includes, for example: a terminal device, a server, or other processing devices. The terminal device may be a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. In some possible implementation manners, the method for processing meteorological data may be implemented by a processor invoking computer-readable instructions stored in a memory.

[0053] See Figure 1 As shown in the flowchart of a method for processing meteorological data provided in the embodiments of the present disclosure, the method includes steps S101 to S104, where:

[0054] S101. Acquire various types of large-scale meteorological data, extract features corresponding to each of the large-scale meteorological data to construct a preliminary feature map, and construct a fusion feature map according to the feature dependence relationship between the preliminary feature maps.

[0055] In a specific implementation, first, various types of large-scale meteorological data are obtained from different data sources, and feature extraction operations are performed on each type of large-scale meteorological data to respectively extract preliminary feature maps containing the corresponding features of the large-scale meteorological data. Then, feature fusion operations are performed on the preliminary feature maps corresponding to all types of large-scale meteorological data to fuse the meteorological data from different data sources and generate a fused feature map with rich semantics and details.

[0056] Here, the large-scale meteorological data is large-scale and low-resolution meteorological data. Large-scale refers to meteorological data with a wide spatial coverage, such as the output of global or continental-level meteorological models; low-resolution means less spatial details, usually each data point covers a large area, for example, each data point represents an area of 100 kilometers or more.

[0057] Exemplarily, the large-scale meteorological data may include terrain mask, FY4B cloud image, numerical forecast data (such as the air pressure of CMA-MESO3km), temperature, geopotential height, radial wind, zonal wind, vertical velocity, specific humidity, precipitation, relative humidity, etc.

[0058] It should be noted that each data source provides at least one type of large-scale meteorological data.

[0059] Specifically, preliminary feature extraction can be performed separately for each type of large-scale meteorological data. This feature extraction process can be implemented using a convolutional neural network (CNN) or a shallow Transformer module to obtain the preliminary feature maps corresponding to each type of large-scale meteorological data.

[0060] Further, as a possible implementation manner, the construction process of the fused feature map can be referred to Figure 2 As shown, it is a flowchart of a method for constructing a fused feature map provided by an embodiment of the present disclosure. The method includes steps S1011 to S1015, where:

[0061] S1011. Align the preliminary feature maps to the same spatial scale and adjust the preliminary feature maps to the same number of channels using a linear transformation.

[0062] S1012. Divide the aligned preliminary feature maps into multiple feature blocks; perform a linear transformation on each feature block to determine the corresponding query vector, key vector, and value vector.

[0063] S1013. Determine the attention weights according to the dot product between the query vector and the key vector, and use the attention weights as the feature dependence relationship.

[0064] S1014. Input the feature block into a multi-layer feature fusion network. For each feature block in each layer, weight the value vector corresponding to this feature block and the value vectors corresponding to other feature blocks according to the attention weights to determine the hierarchical fusion feature vector and generate the corresponding hierarchical fusion feature map.

[0065] S1015. Concatenate and fuse the hierarchical fusion feature maps corresponding to each layer to generate the fusion feature map.

[0066] In a specific implementation, the feature fusion process between multi-source meteorological data can be implemented using a multi-layer feature fusion network. Preferably, the Swin Transformer model can be used. During the processing, since the preliminary feature maps corresponding to the large-scale meteorological data of different data sources may have different spatial resolutions or perspectives, it is first necessary to align the feature maps to the same spatial scale through interpolation, translation, or other alignment methods. At the same time, the preliminary feature maps of different data sources may have different numbers of channels, and they are adjusted to the same number of channels through linear transformation or other techniques.

[0067] Here, the relationship and dependence between the large-scale meteorological data corresponding to different data sources can be captured by referring to the multi-head self-attention mechanism to establish connections between different feature maps and weight them according to their correlation.

[0068] Specifically, first perform a chunking operation to divide the aligned preliminary feature map into feature blocks of a fixed size. These feature blocks serve as the input to the multi-layer feature fusion network. Then, at each level, use the multi-head self-attention mechanism to perform a linear transformation on these feature blocks to obtain query vectors, key vectors, and value vectors. Calculate the dot product between the query vector and the key vector to obtain the attention weights, and then weight these weights to the corresponding value vectors to obtain the fused fusion feature vector.

[0069] As a possible implementation, to reduce the computational complexity, the preliminary feature map can be divided into multiple local windows. Perform self-attention calculation within each window to capture the feature relationships within the local area. To maintain the flow of global information, window sliding or other methods can also be used to enable the information between different windows to be transmitted and fused with each other.

[0070] It should be noted that since the multi-layer feature fusion network adopts a hierarchical architecture, self-attention calculation and fusion operations are performed on the feature map at each layer. After each layer is processed, the output feature map serves as the input to the next layer. At the same time, the hierarchical fusion feature vectors generated at the bottom layer are retained and transmitted to the upper layer through residual connections between layers.

[0071] Furthermore, through layer-by-layer processing and cross-layer connection, the feature maps of each layer are gradually fused, and the processed feature maps are upsampled to the target resolution to form the final fused feature map.

[0072] S102. Extract the low-level fused features corresponding to the fused feature map, downsample the fused feature map to reduce the spatial resolution of the fused feature map, and then further extract the high-level fused features corresponding to the fused feature map while maintaining the spatial resolution.

[0073] In a specific implementation, after fusing the meteorological data from different data sources, the fused feature map is input into a preset downscaling model, and a downsampling operation is performed in the preset downscaling model to reduce the spatial resolution of the fused feature map, that is, to reduce the scale of the meteorological data.

[0074] Here, the preset downscaling model can be implemented using the pix2pixGAN model. The pix2pixGAN model includes a generator and a discriminator. The generator is responsible for converting the input fused feature map into small-scale, high-resolution meteorological data, and the discriminator is used to make the result generated by the generator closer to higher-resolution data.

[0075] Among them, the generator can adopt the ResUNet architecture, including the structure of an encoder, a bottleneck layer, a decoder, and an output layer. The encoder extracts multi-level features and gradually reduces the spatial resolution of the feature map while retaining important information; the bottleneck layer further processes the feature map to extract high-level semantic features while maintaining the size of the feature map; the decoder restores the spatial resolution of the feature map step by step to generate high-resolution meteorological data, and at the same time uses skip connections to fuse low-level detail information; the output layer converts the feature map in the decoder part into the final small-scale, high-resolution meteorological data.

[0076] Specifically, the downsampling process can be implemented through the encoder in the generator, including the following steps 1-step 4:

[0077] Step 1. Input the fused feature map into the encoder of the preset downscaling model, where the encoder includes a convolutional layer, a residual layer, and a max pooling layer.

[0078] Step 2. Perform a convolution operation on the fused feature map through the convolutional layer to determine the low-level fused features, and transfer the low-level fused features to the residual layer.

[0079] Step 3. In the residual layer, superimpose the fused feature map and the low-level fused features through a residual connection and transfer them to the max pooling layer.

[0080] Step 4: Perform downsampling through the max pooling layer, and store the feature map with reduced spatial resolution into the skip connection.

[0081] In a specific implementation, first, perform a preliminary convolution operation on the fused feature map in the convolutional layer of the encoder to extract low-level fused features. Exemplarily, a 3x3 convolutional kernel can be used, with a stride of 1 and padding of 1, and the output feature map maintains the original size. Then, transfer the low-level fused features to the residual block. Each residual layer contains multiple convolutional blocks with residual connections. Preferably, each residual layer consists of two 3x3 convolutions, batch normalization, and ReLU activation. The residual connection ensures the superposition of the input features and the features after the convolution operation, enhancing the gradient flow. Further, perform max pooling operation on the feature map after the residual layer to downsample the feature map and reduce the spatial resolution. Exemplarily, 2x2 pooling can be used, with a stride of 2, to halve the size of the feature map.

[0082] In this way, through the downsampling operation of the max pooling layer, the spatial dimension of the feature map is continuously reduced to aggregate information of larger spatial regions by reducing the spatial resolution.

[0083] Exemplarily, assume that the size of the input fused feature map is 256x256xC (C is the number of channels). After the operation of the convolutional layer, the size of the feature map can be 256x256xC. After the residual layer, the size of the feature map can be 256x256xC. And after the downsampling of the max pooling layer, the size of the feature map can be 128x128xC.

[0084] It should be noted that after each downsampling step, the downsampled feature map is stored through the skip connection for subsequent use by the decoder.

[0085] Here, after downsampling to reduce the spatial resolution of the fused feature map, perform convolution operation and residual connection through the bottleneck layer to further process the feature map, extract high-level semantic features, and maintain the size of the feature map. Specifically, it can be implemented in the following way: input the downsampled feature map into the bottleneck layer of the preset downscaling model, and extract the high-level fused features through convolution operation and residual connection;

[0086] Among them, no pooling operation is performed in the bottleneck layer to maintain the spatial resolution. Exemplarily, a convolution operation of 128x128xC can continue to be used.

[0087] In this way, the high-resolution fused feature map output after feature fusion of large-scale, low-resolution meteorological data is downsampled through convolution and max pooling operations, multi-level features are extracted, downscaling is achieved, and further processed by the bottleneck layer to extract high-level semantic features, so as to extract multi-level features, gradually reduce the spatial resolution of the feature map, and retain important information.

[0088] S103. Upsample the fused feature map after extracting the high-level fused features to restore the spatial resolution, and splice the upsampled fused feature map with the downsampled fused feature map to determine a spliced feature map.

[0089] In a specific implementation, after the encoder downsamples the fused feature map to reduce the spatial resolution, it is input into the decoder of a preset downscaling model to gradually restore the spatial resolution of the feature map, generate high-resolution meteorological data, and at the same time use skip connections to fuse low-level detailed information.

[0090] Here, the upsampling process may include the following steps 1-step 3:

[0091] Step 1. Input the feature map output by the encoder into the decoder of the preset downscaling model, where the decoder includes a transposed convolution layer and a feature fusion layer.

[0092] Step 2. Upsample the input feature map through the transposed convolution layer to restore the spatial resolution, and input the upsampled feature map into the skip connection.

[0093] Step 3. In the skip connection, splice the stored feature map with reduced spatial resolution with the upsampled feature map to determine the spliced feature map.

[0094] In a specific implementation, the feature map is gradually upsampled through a transposed convolution or an upsampling layer to restore the spatial resolution. For example, use a 2x2 transposed convolution (or upsampling) with a stride of 2 to double the size of the feature map. Then, a skip connection is used to splice the feature map stored in the encoder part with the upsampled feature map to determine the spliced feature map and retain the detailed information.

[0095] Exemplarily, assume that the size of the upsampled feature map is 256x256xC, and after upsampling, the size of the feature map is 256x256xC. Then, splice the corresponding layer feature map of the encoder (obtained from the skip connection storage) with a size of 256x256x2C.

[0096] It should be noted that although the decoder part gradually restores the spatial resolution of the feature map through upsampling (transposed convolution or other upsampling operations), this process is different from the input data and generates small-scale, high-resolution data. This upsampling process is actually to restore higher-resolution output data from the downscaled abstract features.

[0097] S104. Perform a convolution operation on the spliced feature map. After fusing the low-level fused features and the high-level fused features, convert the spliced feature map into small-scale meteorological data.

[0098] In a specific implementation, the decoder performs a convolution operation on the spliced feature map through a convolutional layer and a residual block to fuse high-level semantic and low-level detail information, and then converts the spliced feature map output by the decoder into the final small-scale, high-resolution meteorological data through an output layer.

[0099] Here, small-scale refers to meteorological data with a small spatial coverage range (such as the output of a regional-level meteorological model); high-resolution means rich spatial details, and each data point covers a small area (for example, each data point represents an area of 1 kilometer or less).

[0100] Specifically, the spliced feature map is input into the feature fusion layer of the decoder, and a convolution operation is performed on the spliced feature map to fuse low-level fusion features and high-level fusion features.

[0101] Furthermore, the output layer of the generator finally converts the feature map output by the decoder into small-scale, high-resolution meteorological data through a convolutional layer. Exemplarily, a 1x1 convolutional kernel is used, and the number of output channels is the number of channels of the small-scale, high-resolution meteorological data.

[0102] As a possible implementation, the loss function of the preset downscaling model includes a generator loss and a discriminator loss. The generator loss can include an L1 loss (or an L2 loss) and an adversarial loss. The L1 loss is used to measure the pixel-level difference between the generated meteorological data and the real high-resolution meteorological data, and the adversarial loss is used to improve the authenticity of the generated data. The discriminator loss is used to optimize the discriminator so that it can effectively distinguish between the generated high-resolution meteorological data and the real data.

[0103] Here, during the training process, the generator and the discriminator can be alternately trained: the generator generates high-resolution meteorological data, and the discriminator judges its authenticity, and the two networks are continuously optimized through mutual confrontation.

[0104] A method for processing meteorological data provided by an embodiment of the present disclosure obtains various types of large-scale meteorological data, extracts features corresponding to each type of the large-scale meteorological data to construct a preliminary feature map, and constructs a fusion feature map according to the feature dependence relationship between the preliminary feature maps; extracts low-level fusion features corresponding to the fusion feature map, downsamples the fusion feature map to reduce the spatial resolution of the fusion feature map, and then further extracts high-level fusion features corresponding to the fusion feature map while maintaining the spatial resolution; upsamples the fusion feature map after extracting the high-level fusion features to restore the spatial resolution, and splices the upsampled fusion feature map with the downsampled fusion feature map to determine a spliced feature map; performs a convolution operation on the spliced feature map, and after fusing the low-level fusion features and the high-level fusion features, converts the spliced feature map into small-scale meteorological data. It can achieve the downscaling processing of meteorological data with less computing resources, and effectively fuse meteorological data from different data sources, overcoming the limitations in processing non-linear and complex meteorological relationships in the prior art.

[0105] Those skilled in the art can understand that in the above method of the specific embodiment, the writing order of each step does not mean a strict execution order and does not constitute any limitation to the implementation process. The specific execution order of each step should be determined according to its function and possible internal logic.

[0106] Based on the same inventive concept, an embodiment of the present disclosure also provides a processing device for meteorological data corresponding to the method for processing meteorological data. Since the principle of solving problems by the device in the embodiment of the present disclosure is similar to the above method for processing meteorological data in the embodiment of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0107] Please refer to Figure 3 , Figure 3 which is a schematic diagram of a processing device for meteorological data provided by an embodiment of the present disclosure. As Figure 3 shown in, the processing device 300 for meteorological data provided by an embodiment of the present disclosure includes:

[0108] A feature fusion module 310, configured to obtain various types of large-scale meteorological data, extract features corresponding to each type of the large-scale meteorological data to construct a preliminary feature map, and construct a fusion feature map according to the feature dependence relationship between the preliminary feature maps.

[0109] A downscaling module 320, configured to extract low-level fusion features corresponding to the fusion feature map, downsample the fusion feature map to reduce the spatial resolution of the fusion feature map, and then further extract high-level fusion features corresponding to the fusion feature map while maintaining the spatial resolution.

[0110] The resolution improvement module 330 is configured to upsample the feature map after extracting the high-level fused features to restore the spatial resolution, and splice the upsampled feature map with the downsampled feature map to determine a spliced feature map.

[0111] The data conversion module 340 is configured to perform a convolution operation on the spliced feature map, and after fusing the low-level fused features and the high-level fused features, convert the spliced feature map into small-scale meteorological data.

[0112] For the processing flow of each module in the device and the interaction flow between modules, reference can be made to the relevant descriptions in the above method embodiments, which will not be elaborated here.

[0113] A meteorological data processing device provided in an embodiment of the present disclosure obtains various types of large-scale meteorological data, extracts features corresponding to each large-scale meteorological data to construct a preliminary feature map, and constructs a fused feature map according to the feature dependency relationship between the preliminary feature maps; extracts low-level fused features corresponding to the fused feature map, downsamples the fused feature map to reduce the spatial resolution of the fused feature map, and then further extracts high-level fused features corresponding to the fused feature map while maintaining the spatial resolution; upsamples the fused feature map after extracting the high-level fused features to restore the spatial resolution, and splices the upsampled feature map with the downsampled feature map to determine a spliced feature map; performs a convolution operation on the spliced feature map, and after fusing the low-level fused features and the high-level fused features, converts the spliced feature map into small-scale meteorological data. It can achieve downscaling processing of meteorological data with less computing resources, and at the same time effectively fuse meteorological data from different data sources, overcoming the limitations in dealing with non-linear and complex meteorological relationships in the prior art.

[0114] Corresponding to Figure 1 the meteorological data processing method in Figure 4 as shown in

[0115] a processor 41, a memory 42, and a bus 43; the memory 42 is used to store executable instructions, including an internal memory 421 and an external memory 422; here, the internal memory 421 is also called the main memory, which is used to temporarily store the operation data in the processor 41 and the data exchanged with the external memory 422 such as a hard disk. The processor 41 exchanges data with the external memory 422 through the internal memory 421. When the electronic device 400 runs, the processor 41 communicates with the memory 42 through the bus 43, so that the processor 41 executes Figure 1 the steps of the method for processing meteorological data in

[0116] Embodiments of the present disclosure also provide a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, it executes the steps of the method for processing meteorological data described in the foregoing method embodiments. Wherein, the storage medium may be a volatile or non-volatile computer-readable storage medium.

[0117] Embodiments of the present disclosure also provide a computer program product, which includes computer instructions. When the computer instructions are executed by a processor, they can execute the steps of the method for processing meteorological data described in the foregoing method embodiments. For details, please refer to the foregoing method embodiments and will not be elaborated herein.

[0118] Among them, the above computer program product can be specifically implemented in the form of hardware, software, or a combination thereof. In an alternative embodiment, the computer program product is specifically embodied as a computer storage medium. In another alternative embodiment, the computer program product is specifically embodied as a software product, such as a Software Development Kit (SDK), etc.

[0119] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the above-described device can refer to the corresponding process in the foregoing method embodiments and will not be elaborated herein. In several embodiments provided by the present disclosure, it should be understood that the disclosed device and method can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection may be through some communication interfaces, and the indirect coupling or communication connection of the device or unit may be in an electrical, mechanical or other form.

[0120] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed across multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0121] In addition, in each embodiment of the present disclosure, each functional unit may be integrated into a processing unit, may exist physically separately as individual units, or two or more units may be integrated into one unit.

[0122] If the described function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such an understanding, the technical solution of the present disclosure, in essence or the part that contributes to the prior art or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present disclosure. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs that can store program codes.

[0123] Finally, it should be noted that: the above-described embodiments are only specific implementation manners of the present disclosure, used to illustrate the technical solutions of the present disclosure, rather than limiting them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: any person skilled in the art within the technical scope disclosed by the present disclosure can still modify the technical solutions described in the foregoing embodiments or can easily think of changes, or perform equivalent replacements for some of the technical features; and these modifications, changes, or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should all be covered within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A method for processing meteorological data, characterized in that, Including: Obtain various types of large-scale meteorological data, extract the corresponding features of each piece of the large-scale meteorological data to construct a preliminary feature map, and construct a fused feature map according to the feature dependence relationship between the preliminary feature maps; Extract the low-level fused features corresponding to the fused feature map, downsample the fused feature map to reduce the spatial resolution of the fused feature map, and then further extract the high-level fused features corresponding to the fused feature map while maintaining the spatial resolution; Upsample the fused feature map after extracting the high-level fused features to restore the spatial resolution, and splice the upsampled fused feature map with the downsampled fused feature map to determine a spliced feature map; Perform a convolution operation on the spliced feature map. After fusing the low-level fused features and the high-level fused features, convert the spliced feature map into small-scale meteorological data.

2. The method according to claim 1, characterized in that Determine the feature dependence relationship based on the following steps: Align the preliminary feature maps to the same spatial scale, and use a linear transformation to adjust the preliminary feature maps to the same number of channels; Divide the aligned preliminary feature maps into multiple feature blocks; Perform a linear transformation on each feature block to determine the corresponding query vector, key vector, and value vector; Determine the attention weight according to the dot product between the query vector and the key vector, and use the attention weight as the feature dependence relationship.

3. The method according to claim 2, characterized in that, Construct the fused feature map based on the following steps: Input the feature blocks into a multi-layer feature fusion network. In each layer, for each feature block, weight the value vector corresponding to this feature block and the value vectors corresponding to other feature blocks according to the attention weight to determine a hierarchical fused feature vector and generate a corresponding hierarchical fused feature map, where between each layer, the hierarchical fused feature vector generated at the bottom layer is retained and passed to the upper layer through a residual connection; Splice and fuse the hierarchical fused feature maps corresponding to each layer to generate the fused feature map.

4. The method according to claim 1, characterized in that, Extract the low-level fused features corresponding to the fused feature map, and downsample the fused feature map to reduce the spatial resolution of the fused feature map. Specifically including: Input the fused feature map into the encoder of a preset downscaling model, where the encoder includes a convolutional layer, a residual layer, and a max pooling layer; Perform a convolution operation on the fused feature map through the convolutional layer to determine the low-level fused features, and pass the low-level fused features to the residual layer; In the residual layer, superimpose the fused feature map and the low-level fused features through a residual connection and pass them to the max pooling layer; Perform downsampling through the max pooling layer, and store the feature map with reduced spatial resolution in a skip connection.

5. The method according to claim 4, characterized in that Extract the high-level fused features corresponding to the fused feature map while maintaining the spatial resolution. Specifically including: Input the downsampled feature map into the bottleneck layer of the preset downscaling model, and extract the high-level fused features through convolution operations and residual connections; Among them, no pooling operation is performed on the bottleneck layer to maintain the spatial resolution.

6. The method according to claim 4, characterized in that, The fused feature map after extracting the high-level fused features is upsampled to restore the spatial resolution, and the upsampled fused feature map is concatenated with the downsampled fused feature map. Determining the concatenated feature map specifically includes: Input the feature map output by the encoder into the decoder of the preset downscaling model, where the decoder includes a transposed convolution layer and a feature fusion layer; The transposed convolution layer upsamples the input feature map to restore the spatial resolution, and inputs the upsampled feature map into the skip connection; In the skip connection, the feature map with reduced spatial resolution stored is concatenated with the upsampled feature map to determine the concatenated feature map.

7. The method according to claim 6, wherein Fusing the low-level fused features and the high-level fused features specifically includes: Input the concatenated feature map into the feature fusion layer, perform a convolution operation on the concatenated feature map, and fuse the low-level fused features and the high-level fused features.

8. A processing device for meteorological data, characterized in that, Includes: A feature fusion module for obtaining various types of large-scale meteorological data, extracting the corresponding features of each large-scale meteorological data to construct a preliminary feature map, and constructing a fused feature map according to the feature dependence relationship between the preliminary feature maps; A downscaling module for extracting the low-level fused features corresponding to the fused feature map, downsampling the fused feature map to reduce the spatial resolution of the fused feature map, and then further extracting the high-level fused features corresponding to the fused feature map while maintaining the spatial resolution; A resolution enhancement module for upsampling the fused feature map after extracting the high-level fused features to restore the spatial resolution, and concatenating the upsampled fused feature map with the downsampled fused feature map to determine the concatenated feature map; A data conversion module for performing a convolution operation on the concatenated feature map, and after fusing the low-level fused features and the high-level fused features, converting the concatenated feature map into small-scale meteorological data.

9. An electronic device, characterized in that, Includes: A processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device runs, the processor communicates with the memory through the bus. When the machine-readable instructions are executed by the processor, the steps of the method for processing meteorological data according to any one of claims 1 to 7 are executed.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium. When the computer program is run by the processor, the steps of the method for processing meteorological data according to any one of claims 1 to 7 are executed.

Images