Method and apparatus for processing meteorological radar image data

Abstract

The present disclosure provides a meteorological radar image data processing method and device, belonging to the technical field of meteorological monitoring, the method comprising: obtaining a set of meteorological radar image data to be processed, the set of meteorological radar image data comprising a plurality of meteorological radar images to be processed; processing the meteorological radar images in the set of meteorological radar image data using a pre-trained meteorological radar image processing model to generate a set of processing result images, the processing result images in the set of processing result images identifying corresponding clutter distribution areas; performing interpolation processing on the processing result images in the set of processing result images to generate a set of standard images; and using a short-term extrapolation model to predict the short-term change trend of convective weather according to the set of standard images. In this way, the clutter interference in the radar echo image data can be filtered, thereby improving the accuracy of extrapolation prediction.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure generally relate to the field of meteorological monitoring, and more particularly, to a method and apparatus for processing meteorological radar image data. BACKGROUND

[0002] Severe convective weather has the characteristics of small spatial scale, short life history, strong burst, rapid development and evolution, and great destructive power. If it cannot be timely warned, it may cause great harm. The short-term nowcasting for the next 0-2 hours is also a big difficulty in meteorological forecasting work. The traditional numerical model prediction method models the physical model of atmospheric motion, and then gives the prediction by simulating the changes of the physical model in the future time period. The advantage of this method is that it is more accurate for long-term weather prediction, but it is not ideal for weather conditions within 0-2 hours. The method of imageology extrapolation through radar echo image can better capture the short-term trend of severe convective weather, so as to give more accurate prediction.

[0003] The method of using radar echo for short-term extrapolation is a data-driven method. If the quality of the radar echo image data used is not good, it will greatly affect the accuracy of the final extrapolation prediction. The original data of the radar echo image contains a large amount of noise interference, which mainly comes from the ground and buildings, oceans, rain and snow weather, bird flocks and insects, etc., greatly affecting the accuracy of the extrapolation prediction. SUMMARY

[0004] According to embodiments of the present disclosure, a meteorological radar image data processing scheme is provided for filtering noise interference in radar echo image data, thereby improving the accuracy of extrapolation prediction.

[0005] In a first aspect of the present disclosure, a method for processing meteorological radar image data is provided, comprising:

[0006] obtaining a set of meteorological radar images to be processed, the set of meteorological radar images to be processed including a plurality of meteorological radar images to be processed;

[0007] processing the meteorological radar images in the set of meteorological radar images using a pre-trained meteorological radar image processing model to generate a set of processed images, the processed images in the set of processed images identifying corresponding noise distribution regions;

[0008] performing interpolation processing on the processed images in the set of processed images to generate a set of standard images;

[0009] using a short-term extrapolation model to predict the short-term trend of convective weather according to the set of standard images.

[0010] In some embodiments, the weather radar image processing model is trained by:

[0011] a preset number of weather radar image data and weather radar image data corresponding to the clutter region distribution information are taken as a training sample set, weather radar image data in the training sample set are taken as input of the model, weather radar image data corresponding to the identified clutter distribution region are taken as output, and the SE-ResUnet network is trained;

[0012] in response to the similarity between the identified clutter distribution region in the output weather radar image data and the corresponding clutter region distribution information being less than a preset threshold, parameters in the SE-ResUnet network are adjusted, and the SE-ResUnet network after the parameter adjustment is continuously trained;

[0013] the above process is repeated until the similarity between the identified clutter distribution region in the output weather radar image data and the corresponding clutter region distribution information is greater than the preset threshold, the training of the SE-ResUnet network is completed, and the weather radar image processing model is generated.

[0014] In some embodiments, the training sample set is generated by:

[0015] indexing radar echo data according to a fixed format time sequence;

[0016] dividing the index into a plurality of continuous frames, each continuous frame including a radar echo image;

[0017] if the number of radar echo images in which the area of the high reflectivity region with reflectivity greater than c is not less than b% of the whole radar echo image is greater than d, the corresponding continuous frame is determined as a strong convective weather continuous frame, wherein a, c, and c are positive integers, a is greater than d, and b is a number between 0 and 1;

[0018] identifying the clutter distribution region in the radar echo image in the strong convective weather continuous frame to generate the training sample set.

[0019] In some embodiments, the SE-ResUnet network includes an SE module, a ResNet module, and a Unet module.

[0020] In some embodiments, the SE module is used to compress features along the spatial dimension, convert each two-dimensional feature channel into a real number, match the output dimension and the input feature channel number, generate a weight for each feature channel, and re-label the original features in the channel latitude.

[0021] In some embodiments, the ResNet module includes a plurality of residual blocks connected in series, and the width and depth of the network are adjusted by adjusting the number of channels and the number of residual blocks to obtain networks with different expression capabilities.

[0022] In some embodiments, the Unet module adopts an encoding and decoding structure, performs multiple down-sampling on the image, then performs up-sampling, and fuses low-level features and high-level features.

[0023] In a second aspect of the present disclosure, a processing apparatus for meteorological radar image data is provided, including:

[0024] An image acquisition module is configured to acquire a set of meteorological radar image data to be processed, the set of meteorological radar image data including a plurality of meteorological radar images to be processed;

[0025] A clutter distribution area determination module is configured to process the meteorological radar images in the set of meteorological radar image data using a pre-trained meteorological radar image processing model to generate a set of processing result images, the processing result images in the set of processing result images identifying corresponding clutter distribution areas;

[0026] A standard image generation module is configured to perform interpolation processing on the processing result images in the set of processing result images to generate a set of standard images;

[0027] A weather prediction module is configured to use a short-term extrapolation model to predict the short-term trend of convective weather according to the set of standard images.

[0028] In a third aspect of the present disclosure, an electronic device is provided, including a memory and a processor, the memory having a computer program stored thereon, and the processor implementing the method described above when executing the program.

[0029] In a fourth aspect of the present disclosure, a computer readable storage medium is provided, having a computer program stored thereon, and the program implementing the method described above when executed by a processor.

[0030] The processing method for meteorological radar image data of the present disclosure can filter the clutter interference in the radar echo image data, thereby improving the accuracy of extrapolation prediction.

[0031] The content described in the summary section is not intended to limit the key or important features of the embodiments of the present disclosure, nor to limit the scope of the present disclosure. Other features of the present disclosure will become apparent through the following description. BRIEF DESCRIPTION OF DRAWINGS

[0032] The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings. In the drawings:

[0033] Figure 1 A flow chart of a method for processing weather radar image data according to an embodiment of the present disclosure is shown;

[0034] Figure 2 A structural schematic diagram of a device for processing weather radar image data according to an embodiment of the present disclosure is shown;

[0035] Figure 3 A structural schematic diagram of an apparatus for processing weather radar image data according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0036] 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 described clearly and completely below with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are some but not all of the embodiments of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present disclosure.

[0037] In addition, the term "and / or" herein merely describes an association relationship of associated objects, which means that there can be three relationships, for example, A and / or B can mean that there are three cases of A alone, A and B together, and B alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0038] The present application proposes a method for processing weather radar image data to filter the clutter interference in the radar echo image data, generate high-quality radar echo image data, and further improve the data accuracy of short-term extrapolation.

[0039] Specifically, as shown in Figure 1 A flow chart of a method for processing weather radar image data according to an embodiment of the present disclosure is shown. As an optional embodiment of the present disclosure, the method for processing weather radar image data can include the following steps:

[0040] S101: Obtain a set of weather radar image data to be processed, which includes a plurality of weather radar images to be processed.

[0041] Embodiments of the present application can be applied to the field of meteorological monitoring to predict severe convective weather. The method of image extrapolation by radar echo image can better capture the short-term trend of severe convective weather, thereby enabling more accurate prediction. However, the method of short-term extrapolation using radar echoes is a data-driven method, and if the quality of the radar echo image data used is poor, it will also greatly affect the accuracy of the final extrapolation prediction. Based on this, the method of the embodiments of the present application is used to improve the quality of image data.

[0042] Specifically, when filtering the clutter interference in the radar echo image data, first, a set of meteorological radar image data to be processed is needed. The set of meteorological radar image data includes a plurality of meteorological radar images to be processed.

[0043] In this embodiment, the meteorological radar image to be processed is image data that may contain clutter interference, and when the meteorological radar image to be processed is image data containing clutter interference, the clutter interference needs to be filtered.

[0044] S102: Process the meteorological radar images in the set of meteorological radar image data using the pre-trained meteorological radar image processing model to generate a set of processing result images, and the processing result images in the set of processing result images identify the corresponding clutter distribution areas.

[0045] Specifically, when the set of meteorological radar image data to be processed is obtained, the pre-trained meteorological radar image processing model is used to process the meteorological radar images in the set of meteorological radar image data to be processed to generate a set of processing result images, and the processing result images in the set of processing result images identify the corresponding clutter distribution areas.

[0046] The meteorological radar image processing model in this embodiment is a pre-trained SE-ResUnet network model, which is trained in the following way:

[0047] A preset number of meteorological radar image data and meteorological radar image data corresponding clutter area distribution information are used as a training sample set, the meteorological radar image data in the training sample set is used as the input of the model, and the meteorological radar image data corresponding to the identified clutter distribution area is used as the output, and the SE-ResUnet network is trained;

[0048] In response to the similarity between the identified clutter distribution area in the output meteorological radar image data and the corresponding clutter area distribution information being less than a preset threshold, the parameters in the SE-ResUnet network are adjusted, and the SE-ResUnet network with the adjusted parameters is continuously trained;

[0049] The above process is repeated until the similarity between the identified clutter distribution area in the output meteorological radar image data and the corresponding clutter area distribution information is greater than a preset threshold, the training of the SE-ResUnet network is completed, and a meteorological radar image processing model is generated.

[0050] The SE-ResUnet network in this embodiment includes an SE module, a ResNet module, and a Unet module. The SE module is used to perform feature compression along the spatial dimension on the input data, convert each two-dimensional feature channel into a real number, match the output dimension with the input feature channel number, and generate a weight for each feature channel to re-label the original features in the channel latitude. In this way, the mutual dependence between model channels can be taken into account.

[0051] The ResNet module includes a plurality of series of residual blocks. The structure of the ResNet module is very easy to modify and extend. The width and depth of the network can be adjusted by adjusting the number of channels and the number of residual blocks to obtain a network with different expression capabilities without worrying about the "degradation" problem caused by too many network layers. As long as the training data is sufficient, gradually deepening the network can achieve better performance.

[0052] The Unet module adopts an encoding and decoding structure, which first performs multiple down-sampling on the image, then performs up-sampling, and fuses low-level features and high-level features.

[0053] By combining the SE module, the ResNet module, and the Unet module, a more excellent fitting effect can be achieved.

[0054] In addition, the training sample set of the meteorological radar image processing model in this embodiment is generated in the following way:

[0055] An index is constructed for the radar echo data according to the fixed format time sequence. For example, the radar echo images can be sorted according to the time points corresponding to the radar echo images. The time point of the radar echo image can be in the format of yyyymmdd / hh / mm, i.e., year, month, day, hour, and minute. An index is established according to the time point corresponding to the radar echo image. Then, the index is divided into a plurality of continuous frames, each continuous frame including a radar echo image; if the area of the high-reflection area with reflectivity greater than c in each continuous frame is not less than b% of the number of radar echo images of the entire radar echo image, and the number of radar echo images is greater than d, the corresponding continuous frame is determined as a strong convective weather continuous frame, where a, c, and c are positive integers, a is greater than d, and b is a number between 0 and 1.

[0056] For example, the indexes can be divided into multiple continuous frames, each continuous frame including 10 radar echo images; for each continuous frame, if the number of radar echo images in which the area of high reflection region with reflectivity greater than 0.75 is not less than 55% of the whole radar echo image is greater than 5, all radar echo images in the continuous frame are taken as available radar echo images. Then, the clutter distribution region in the radar echo images in the strong convective weather continuous frame is identified to generate a training sample set. For example, the clutter distribution range in the available radar echo images can be manually labeled, for example, a matrix in which each pixel is represented by 0 and 1 can be labeled. After the training sample set is generated, the SE-ResUnet network is trained by using the generated training sample set to generate a meteorological radar image processing model.

[0057] The meteorological radar image processing model is used to process the meteorological radar images in the meteorological radar image data set to generate a processing result image set, and a matrix image represented by 0 and 1 is generated, in which 1 represents that there is clutter interference in the corresponding region.

[0058] S103: Interpolation processing is performed on the processing result images in the processing result image set to generate a standard image set.

[0059] In this embodiment, after the meteorological radar image processing model is used to process the meteorological radar images in the meteorological radar image data set to generate a processing result image set, interpolation processing is performed on each image in the processing result image set to filter the clutter interference. For example, mean filtering and Gaussian filtering can be performed.

[0060] S104: The short-term change trend of the convective weather is predicted according to the standard image set by using a short-term extrapolation model.

[0061] The short-term change trend of the convective weather is predicted according to the filtered standard image set by using a short-term extrapolation model.

[0062] Table 1 is a comparison of the radar short-term extrapolation performance improvement effect after using the noise reduction processing on the radar echo image sample:

[0063] Table 1: Comparison table of radar short-term extrapolation performance improvement effect

[0064]

[0065] As can be seen from the table, the use of conventional noise reduction methods can significantly improve the effect of radar short-term extrapolation; and the use of the method of the present application can further improve the effect.

[0066] The meteorological radar image data processing method of the present application can filter the clutter interference in the radar echo image data, thereby improving the accuracy of extrapolation prediction.

[0067] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a combination of a series of actions, but those skilled in the art should know that the disclosure is not limited by the order of the described actions, because according to the disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the disclosure.

[0068] The above is the introduction of the method embodiment, and the scheme described in the disclosure will be further described through the device embodiment.

[0069] As shown in Figure 2 Fig. 2 is a structural schematic diagram of a weather radar image data processing device according to an embodiment of the disclosure. The weather radar image data processing device according to the embodiment includes:

[0070] An image acquisition module 201 is configured to acquire a weather radar image data set to be processed, wherein the weather radar image data set includes a plurality of weather radar images to be processed.

[0071] A clutter distribution area determination module 202 is configured to process the weather radar images in the weather radar image data set by using a pre-trained weather radar image processing model, to generate a processing result image set, wherein a corresponding clutter distribution area is identified in a processing result image in the processing result image set.

[0072] A standard image generation module 203 is configured to perform interpolation processing on the processing result images in the processing result image set, to generate a standard image set.

[0073] A weather prediction module 204 is configured to predict a short-term change trend of convective weather according to the standard image set by using a short-term extrapolation model.

[0074] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described modules can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0075] Figure 3A schematic block diagram of an electronic device 300 that can be used to implement embodiments of the present disclosure is shown. As shown, the device 300 includes a central processing unit (CPU) 301 that can perform various suitable actions and processes in accordance with computer program instructions stored in a read-only memory (ROM) 302 or computer program instructions loaded from a storage unit 308 into a random access memory (RAM) 303. Various programs and data required by the device 300 to operate can also be stored within the RAM 303. The CPU 301, ROM 302, and RAM 303 are connected to each other by a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0076] A number of components in the device 300 are connected to the I / O interface 305, including an input unit 306, such as a keyboard, a mouse, etc.; an output unit 307, such as various types of displays, speakers, etc.; a storage unit 308, such as a magnetic disk, a magneto-optical disk, etc.; and a communication unit 309, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 309 allows the device 300 to exchange information / data with other devices over a computer network, such as the Internet, and / or various telecommunication networks.

[0077] The processing unit 301 performs various methods and processes described above, which are tangibly embodied in a machine-readable medium, such as the storage unit 308. In some embodiments, part or all of a computer program can be loaded onto the device 300 via the ROM 302 and / or the communication unit 309. When the computer program is loaded onto the RAM 303 and executed by the CPU 301, one or more steps of the methods described above can be performed. Alternatively, in other embodiments, the CPU 301 can be configured to perform the above methods by any other suitable means, such as by means of firmware.

[0078] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that can be used include field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip systems (SOCs), complex programmable logic devices (CPLDs), etc.

[0079] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, causes the machine to perform the functions / acts specified in the flowcharts and / or block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine and partially on a remote machine or entirely on a remote machine or server.

[0080] In the context of the present disclosure, a machine-readable medium can be a tangible medium that contains or stores program code for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium would include one or more lines of a system, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0081] Further, while operations are depicted in a particular order, this should not be understood as requiring such an order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing can be advantageous. Likewise, while several specific implementation details have been discussed, such details should not be construed as limiting the scope of the disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in subcombination or as separate embodiments, apart from the other features described.

[0082] Although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

1. A method for processing weather radar image data, characterized in that, include: Obtain a dataset of weather radar images to be processed, which includes multiple weather radar images to be processed; A pre-trained meteorological radar image processing model is used to process meteorological radar images in the meteorological radar image dataset, generating a processed image set. The processed image set identifies corresponding clutter distribution regions. The meteorological radar image processing model is trained as follows: a preset number of meteorological radar image data and corresponding clutter region distribution information are used as a training sample set. The meteorological radar image data in the training sample set is used as the model input, and the meteorological radar image data with identified clutter distribution regions is used as the output. The SE-ResUnet network is trained. In response to the similarity between the identified clutter distribution regions in the output meteorological radar image data and the corresponding clutter region distribution information being less than a preset threshold, the parameters in the SE-ResUnet network are adjusted, and the SE-ResUnet network with adjusted parameters is trained again. This process is repeated until the similarity between the identified clutter distribution regions in the output meteorological radar image data and the corresponding clutter region distribution information is greater than a preset threshold, completing the training of the SE-ResUnet network and generating the meteorological radar image processing model. Interpolation processing is performed on the processed images in the image set to generate a standard image set; The short-term trend of convective weather is predicted based on the standard image set using a short-term extrapolation model.

2. The processing method according to claim 1, characterized in that, The training sample set is generated in the following way: An index is built for radar echo data based on a time series in a fixed format; The index is divided into multiple consecutive frames, each consecutive frame including a radar echo images; If the area of ​​the high reflectivity region greater than c in each consecutive frame is not less than b% of the entire radar echo image and the number of radar echo images is greater than d, the corresponding consecutive frame is determined as a strong convective weather consecutive frame, where a and d are positive integers, a is greater than d, and c is a number between 0 and 1. The clutter distribution areas in radar echo images of continuous frames of severe convective weather are identified to generate a training sample set.

3. The processing method according to claim 2, characterized in that, The SE-ResUnet network includes an SE module, a ResNet module, and a Unet module.

4. The processing method according to claim 3, characterized in that, The SE module is used to compress the input data along the spatial dimension, turning each two-dimensional feature channel into a real number. The output dimension matches the number of input feature channels, and weights are generated for each feature channel to recalibrate the original features in the channel dimension.

5. The processing method according to claim 3, characterized in that, The ResNet module includes multiple cascaded residual blocks. By adjusting the number of channels and the number of residual blocks, the width and depth of the network can be adjusted to obtain networks with different expressive capabilities.

6. The processing method according to claim 3, characterized in that, The Unet module employs an encoding and decoding structure, first downsampling the image multiple times, then upsampling it, and finally fusing low-level and high-level features.

7. A device for processing meteorological radar image data, characterized in that, include: The image acquisition module is used to acquire a dataset of weather radar images to be processed, which includes multiple weather radar images to be processed. The clutter distribution region determination module is used to process meteorological radar images in the meteorological radar image dataset using a pre-trained meteorological radar image processing model, generating a processing result image set in which the corresponding clutter distribution regions are identified. The meteorological radar image processing model is trained as follows: a preset number of meteorological radar image data and corresponding clutter region distribution information are used as a training sample set; the meteorological radar image data in the training sample set is used as the model input, and the meteorological radar image data with identified clutter distribution regions is used as the output, thus training the SE-ResUnet network; in response to the similarity between the identified clutter distribution regions in the output meteorological radar image data and the corresponding clutter region distribution information being less than a preset threshold, the parameters in the SE-ResUnet network are adjusted, and the SE-ResUnet network with adjusted parameters is trained again; the above process is repeated until the similarity between the identified clutter distribution regions in the output meteorological radar image data and the corresponding clutter region distribution information is greater than a preset threshold, thus completing the training of the SE-ResUnet network and generating the meteorological radar image processing model. A standard image generation module is used to perform interpolation processing on the processed result images in the processed result image set to generate a standard image set; The weather forecasting module is used to predict the short-term changing trends of convective weather based on the standard image set using a short-term extrapolation model.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 6.

Images