A method for estimating rainfall intensity of a geostationary meteorological satellite and related equipment

Abstract

This application belongs to the technical field of rainfall prediction, and relates to a method for estimating rainfall intensity of a geostationary meteorological satellite and related equipment. The method includes: First, preprocess the infrared channel information of the Fengyun-4B satellite and the dual-frequency precipitation radar products, and perform spatio-temporal matching on them due to their spatio-temporal resolutions. Second, establish a rainfall detection model based on the Attention-Unet model by introducing auxiliary information such as longitude, latitude, and elevation to predict whether there is rainfall at each pixel point. Then, extract the spatial structure information of the rainfall cloud clusters according to the rainy pixel points. Finally, introduce the infrared channel information, longitude and latitude information, elevation data, and the spatial structure information of the rain clusters into the Attention-Unet model to construct a rainfall intensity estimation model based on the spatial structure of the rain clusters for rainfall estimation. This application can enhance the ability to extract rainfall areas; at the same time, enhance the ability to capture rainfall details and improve the model efficiency.

Description

Technical Field

[0001] This application relates to the technical field of rainfall prediction, applicable to the agricultural or power fields, and particularly relates to a method for estimating rainfall intensity from geostationary meteorological satellites and related equipment. Background Art

[0002] Rainfall is the most important driving factor in the terrestrial hydrological cycle and also the main factor contributing to the uncertainty in the prediction of hydrological fluxes and states. Therefore, accurate rainfall observations are crucial for water level prediction, flood forecasting, drought monitoring, and water resource management, etc. Traditional rainfall observations mainly rely on ground-based observations, and the data obtained has problems such as low spatio-temporal resolution. Among them, the tipping bucket rain gauge has extremely high measurement accuracy at the point scale, but due to the sparse distribution of stations and large network density, it cannot accurately reflect the rainfall at the areal scale. Ground-based radar observations can provide the spatio-temporal distribution of rainfall within a radius of 300 kilometers, but their coverage cannot be extended to the global scale, especially in the ocean, polar regions, mountainous areas, and undeveloped areas. Compared with ground-based observations, satellite observations can, to a certain extent, make up for these deficiencies by providing high spatio-temporal resolution image information with uniform global coverage, especially in uninhabited areas and sea areas lacking ground-based observations.

[0003] The new generation of geostationary meteorological satellites are equipped with visible light, infrared, and water vapor sensors, having high spatio-temporal resolution and being able to provide infrared information on various cloud layers and rainfall activities, and can well capture rapidly changing weather phenomena within a short period of time. However, when using the infrared spectrum to obtain cloud information, the satellite can only detect the top information of the cloud and cannot penetrate the cloud body itself, resulting in the cloud becoming a "black body". Thus, the accuracy of traditional rainfall observations based on geostationary meteorological satellites is relatively low.

[0004] As a classic convolutional neural network structure, Unet is widely used in the field of image segmentation in the task of estimating rainfall intensity due to its efficient feature extraction ability and multi-scale representation ability. However, its structural design is relatively simple, mainly based on an interleaved upsampling and downsampling process, resulting in certain limitations in processing complex textures and details. Specifically, the resolution bottleneck of Unet limits its performance in capturing fine rainfall features, and at the same time, the compression of the encoder module may cause edge information to be gradually lost during the decoding process, thus affecting the prediction accuracy. Summary of the Invention

[0005] The purpose of the embodiments of this application is to propose a method for estimating rainfall intensity from geostationary meteorological satellites and related equipment to solve the problem of relatively low accuracy of traditional rainfall observations based on geostationary meteorological satellites.

[0006] To solve the above technical problems, the embodiments of this application provide a method for estimating rainfall intensity from geostationary meteorological satellites, adopting the following technical solutions:

[0007] Obtain the geostationary satellite observation data, dual-frequency rain radar products, and elevation data of the target area, and preprocess the geostationary satellite observation data to obtain infrared brightness temperature data; the target area is the full disk area of the geostationary satellite;

[0008] Use the spatio-temporal matching technology to unify the time resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rain radar products;

[0009] Input the infrared brightness temperature data, longitude and latitude information, and the elevation data into a rainfall detection model to detect whether rainfall occurs, and obtain a rainfall detection result indicating whether there is rainfall at the pixel points. Among them, the rainfall detection model is an Attention-Unet model;

[0010] According to the rainfall detection result, perform a spatial structure information extraction operation on the pixel points determined to have rain to obtain the spatial structure information of the rainfall cloud cluster;

[0011] Input the infrared brightness temperature data, the longitude and latitude information, the elevation data, and the spatial structure information of the rainfall cloud cluster into a rainfall intensity estimation model to perform a rainfall intensity estimation operation, and obtain a rainfall amount prediction result. Among them, the rainfall intensity estimation model is an Attention-Unet model;

[0012] Output the rainfall amount prediction result.

[0013] Further, the preprocessing of the geostationary satellite observation data includes:

[0014] Perform a radiation calibration operation on the geostationary satellite to convert the DN value of the infrared channel of the geostationary satellite into a meaningful radiation brightness temperature value;

[0015] Through the longitude and latitude lookup correction method, convert the nominal projection of the geostationary satellite into vector data to obtain the infrared brightness temperature data.

[0016] Further, the use of the spatio-temporal matching technology to unify the time resolution and spatial resolution of the obtained geostationary satellite infrared brightness temperature map and the dual-frequency rain radar products includes:

[0017] According to the scanning time of the global rain radar, perform time matching on the infrared brightness temperature data of the geostationary satellite and the dual-frequency rain radar products within the same time range;

[0018] Linearly interpolate the infrared brightness temperature data of the geostationary satellite and the dual-frequency rain radar products onto a 0.05°*0.05° grid to achieve spatial matching.

[0019] Further, after the step of inputting the infrared brightness temperature data, longitude and latitude information, and elevation data into the rainfall detection model to detect whether rainfall occurs and obtaining the detection result of whether rainfall occurs at the pixel points, the following steps are further included:

[0020] Evaluate the rainfall detection result according to the probability of detection (POD) index, false alarm rate (FAR) index, and critical success index (CSI) to obtain the evaluation result of the rainfall detection ability of the rainfall detection model.

[0021] Further, the probability of detection index Is expressed as:

[0022]

[0023] The false alarm rate index Is expressed as:

[0024]

[0025] The critical success index Is expressed as:

[0026]

[0027] Wherein, Indicates that both the predicted data and the observed data are positive classes; Indicates that the predicted data is a negative class and the observed data is a positive class; Indicates that the predicted data is a positive class and the observed data is a negative class.

[0028] Further, after the step of inputting the infrared brightness temperature data, the longitude and latitude information, the elevation data, and the spatial structure information of the rainfall cloud cluster into the rainfall intensity estimation model for rainfall intensity estimation operation to obtain the rainfall amount prediction result, the following steps are further included:

[0029] Evaluate the rainfall amount prediction result according to the root mean square error and the correlation coefficient to obtain the evaluation result of the rainfall amount prediction ability of the rainfall intensity estimation model, wherein the root mean square error Is expressed as:

[0030]

[0031] Wherein, Represents the true rainfall amount of the th pixel point, Represents the predicted rainfall amount of the th pixel point, and N represents the number of pixel points;

[0032] The correlation coefficient Is expressed as:

[0033]

[0034] Among them, represents the true rainfall of the th pixel point, represents the predicted rainfall of the th pixel point, represents the average value of the true rainfall, represents the average value of the predicted rainfall, represents the standard deviation of the predicted rainfall, represents the standard deviation of the true rainfall, and N represents the number of pixel points.

[0035] To solve the above technical problems, an embodiment of the present application further provides a rainfall intensity estimation device for geostationary meteorological satellites, which adopts the following technical solutions:

[0036] A data acquisition module, configured to acquire geostationary satellite observation data, dual-frequency rainfall radar products, and elevation data of a target area, and preprocess the geostationary satellite observation data to obtain infrared brightness temperature data; the target area is the full disk area of the geostationary satellite;

[0037] A data matching module, configured to unify the time resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rainfall radar products by using a spatio-temporal matching technique;

[0038] A rainfall detection module, configured to input the infrared brightness temperature data, longitude and latitude information, and the elevation data into a rainfall detection model to detect whether rainfall occurs, and obtain a rainfall detection result indicating whether rainfall occurs at a pixel point, where the rainfall detection model is an Attention-Unet model;

[0039] A rainfall cloud cluster spatial structure information extraction module, configured to perform a spatial structure information extraction operation on the pixel points determined to have rain according to the rainfall detection result to obtain the spatial structure information of the rainfall cloud cluster;

[0040] A rainfall intensity estimation module, configured to input the infrared brightness temperature data, the longitude and latitude information, the elevation data, and the spatial structure information of the rainfall cloud cluster into a rainfall intensity estimation model to perform a rainfall intensity estimation operation, and obtain a rainfall prediction result, where the rainfall intensity estimation model is an Attention-Unet model;

[0041] A result output module, configured to output the rainfall prediction result.

[0042] Further, the data acquisition module includes:

[0043] A radiometric calibration sub-module is used to perform radiometric calibration operations on geostationary satellites, converting the digital number (DN) values of the infrared channels of geostationary satellites into meaningful radiance temperature values.

[0044] A projection transformation sub-module is used to convert the nominal projection of a geostationary satellite into vector data through a longitude-latitude lookup and correction method to obtain the infrared brightness temperature data.

[0045] To solve the above technical problems, an embodiment of the present application also provides a computer device, adopting the following technical solutions:

[0046] It includes a memory and a processor. Computer-readable instructions are stored in the memory, and when the processor executes the computer-readable instructions, the steps of the above-mentioned method for estimating the rainfall intensity of geostationary meteorological satellites are implemented.

[0047] To solve the above technical problems, an embodiment of the present application also provides a computer-readable storage medium, adopting the following technical solutions:

[0048] Computer-readable instructions are stored on the computer-readable storage medium, and when the computer-readable instructions are executed by a processor, the steps of the above-mentioned method for estimating the rainfall intensity of geostationary meteorological satellites are implemented.

[0049] The present application provides a method for estimating the rainfall intensity of geostationary meteorological satellites, including: obtaining the observation data of geostationary satellites, dual-frequency rain gauging radar products, and elevation data in a target area, and preprocessing the observation data of geostationary satellites to obtain infrared brightness temperature data; the target area is the full disk area of the geostationary satellite; using spatio-temporal matching technology to unify the time resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rain gauging radar products; inputting the infrared brightness temperature data, longitude-latitude information, and elevation data into a rainfall detection model to detect whether rainfall occurs, obtaining a rainfall detection result indicating whether rainfall occurs at pixel points, where the rainfall detection model is an Attention-Unet model; performing spatial structure information extraction operations on the pixel points determined to have rain according to the rainfall detection result to obtain the spatial structure information of the rain cloud; inputting the infrared brightness temperature data, longitude-latitude information, elevation data, and spatial structure information of the rain cloud into a rainfall intensity estimation model for rainfall intensity estimation operations to obtain a rainfall amount prediction result, where the rainfall intensity estimation model is an Attention-Unet model; outputting the rainfall amount prediction result. Compared with the prior art, the present application effectively introduces physical characteristics related to the rainfall occurrence intensity into the deep learning model by constructing the spatial structure information of rain clusters. By combining physical information and the deep learning model, while strengthening the ability to capture small rainfall, it also strengthens the prediction ability of the rainfall falling area, improving the accuracy and physical interpretability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] To more clearly illustrate the solutions in this application, the following will give a brief introduction to the drawings required for the description of the embodiments of this application. Obviously, the drawings in the following description are some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0051] Figure 1 is an exemplary system architecture diagram to which this application can be applied;

[0052] Figure 2 is the implementation flowchart of the method for estimating rainfall intensity of geostationary meteorological satellites provided by the embodiments of this application;

[0053] Figure 3 is the near-surface rainfall rate map of the dual-frequency rain radar product and the infrared brightness temperature map of FY-4B satellite obtained by applying the spatio-temporal matching method provided by the embodiments of this application;

[0054] Figure 4 is the structural schematic diagram of the Attention-Unet network provided by the embodiments of this application;

[0055] Figure 5 is the spatial structure information diagram of the rainfall cloud map provided by the embodiments of this application;

[0056] Figure 6 is the structural schematic diagram of the device for estimating rainfall intensity of geostationary meteorological satellites provided by the embodiments of this application;

[0057] Figure 7 is the structural schematic diagram of an embodiment of a computer device according to this application. Detailed implementation manners

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs; the terms used in the description of the embodiments of this application in this specification are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion. The terms "first", "second", etc. in the specification and claims of this application or the above drawings are used to distinguish different objects and are not used to describe a specific order.

[0059] References to "embodiments" in this specification mean that specific features, structures, or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0060] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0061] As Figure 1 shown, the system architecture 100 may include a terminal device 101, a network 102, and a server 103. The terminal device 101 may be a laptop computer 1011, a tablet computer 1012, or a mobile phone 1013. The network 102 is a medium for providing a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0062] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications may be installed on the terminal device 101, such as a web browser application, a shopping application, a search application, an instant messaging tool, an email client, a social platform software, etc.

[0063] The terminal device 101 may be various electronic devices with a display screen and supporting web browsing. In addition to the laptop computer 1011, the tablet computer 1012, or the mobile phone 1013, the terminal device 101 may also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop portable computer, a desktop computer, etc.

[0064] The server 103 may be a server providing various services, such as a background server supporting the pages displayed on the terminal device 101.

[0065] It should be noted that the method for estimating the rainfall intensity of a geostationary meteorological satellite provided in the embodiments of the present application is generally executed by a server / terminal device. Correspondingly, the device for estimating the rainfall intensity of a geostationary meteorological satellite is generally provided in the server / terminal device.

[0066] It should be understood that Figure 1 the numbers of the terminal devices, networks, and servers in [[ ]] are merely illustrative. According to the implementation requirements, there can be any number of terminal devices, networks, and servers.

[0067] Continuing to refer to Figure 2 , a flowchart of an embodiment of the method for estimating the rainfall intensity of a geostationary meteorological satellite according to the present application is shown. The method for estimating the rainfall intensity of a geostationary meteorological satellite includes: step S201, step S202, step S203, step S204, step S205, and step S206.

[0068] In step S201, obtain the geostationary satellite observation data, dual-frequency rain radar products, and elevation data of the target area, and preprocess the geostationary satellite observation data to obtain infrared brightness temperature data.

[0069] In the embodiment of the present application, obtaining data of a specific area and time includes the observation data of the orbital radiation imager of Fengyun-4B satellite, dual-frequency rain radar products, and elevation data.

[0070] In the embodiment of the present application, the Advanced Geostationary Radiation Imager (AGRI) is one of the main payloads of the Fengyun-4B satellite, with a total of 15 observation channels, including 3 visible light channels, 4 near-infrared channels, and 8 infrared channels. The spatial resolution is 4 km, and the lowest can reach 0.5 - 2 km.

[0071] In the embodiment of the present application, the dual-frequency rain radar product is the second-generation spaceborne rain radar carried by the Global Precipitation Measurement (GPM) satellite, which is a dual-frequency rainfall radar combining a 13.6 GHz Ku-band radar (KuPR) and a 35.5 GHz Ka-band radar (KaPR). The width of each scan of this data is 245 km, the detection height is 22 km upward from the ground surface, the horizontal resolution is about 5 km, and the vertical resolution is about 250 m.

[0072] In the embodiment of the present application, the elevation data is the data including land elevation and ocean depth covering the global scope in 2023 provided by the National Earth System Science Data Center.

[0073] In the embodiment of the present application, the preprocessing of the geostationary satellite observation data includes: performing radiometric calibration operations on the geostationary satellite to convert the DN value of the infrared channel into a radiance value; through the longitude and latitude lookup correction method, converting the nominal projection of the geostationary satellite into vector data, that is, converting the row and column numbers of the observation data into longitude and latitude.

[0074] In the embodiment of the present application, the target area is the full disk area of the geostationary satellite.

[0075] In step S202, the spatio-temporal matching technology is used to unify the temporal resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rainfall radar product;

[0076] In the embodiment of the present application, according to the scanning time of the global rainfall radar, the observation image of Fengyun-4B satellite is temporally matched with the dual-frequency rainfall radar product of the global rainfall radar in the same time range. According to the result of the temporal matching, the longitude and latitude range of the target area is determined, and the infrared brightness temperature data of Fengyun-4B satellite and the dual-frequency rainfall radar product are both linearly interpolated onto a 0.05° * 0.05° grid to achieve spatial matching. An example map of the matching result is as Figure 3 shown.

[0077] In step S203, the infrared brightness temperature data, longitude and latitude information, and elevation data are input into the rainfall detection model to detect whether rainfall occurs, and a rainfall detection result indicating whether rainfall occurs at the pixel points is obtained. Among them, the rainfall detection model is the Attention-Unet model.

[0078] In the embodiment of the present application, the Attention-Unet model effectively enhances the model's attention to the rainfall area and significantly improves the ability to extract detailed features and the prediction speed of the model by introducing an attention gate mechanism. Based on the Unet network, an attention sub-module is added to each decoder layer. This sub-module uses an attention gating network to automatically model the foreground and background and calculate the weights that should be assigned to pixels at different positions, so that the model can focus more on and pay attention to the rainfall area, while sensitively capturing details such as cloud edges and small rainfall distributions. The specific detailed operations are as Figure 4 shown, where the feature map of the same layer in the encoder downsampling , with a size of , after convolution, it gets . The feature map of the previous layer in the upsampling is , with a size of , after convolution, it gets . Adding and and then processing with the ReLU activation function gets . Then, after convolution calculation, it gets . Finally, performing sigmoid activation processing on can obtain the final attention score . Multiplying the attention coefficient calculated from and with again can obtain .

[0079] In step S204, according to the rainfall detection result, a spatial structure information extraction operation is performed on the pixel points determined to have rain, and the spatial structure information of the rainfall cloud cluster is obtained.

[0080] In the embodiment of the present application, to construct the spatial structure information of the rain cluster, the spatial structure information can be extracted from the pixel points determined to have rain according to the rainfall detection result, so as to improve the ability to estimate the rainfall intensity.

[0081] In the embodiment of the present application, the spatial structure information extraction operation is to determine the occurrence intensity mechanism of the rainfall cloud cluster according to the rainfall detection result in combination with the infrared brightness temperature change condition, and extract the information including the central position of the rainfall cloud cluster (the minimum value of the infrared brightness temperature in the 10.8μm band), the distance from the pixel point with rain to the cloud cluster boundary, the distance from the pixel point with rain to the cloud cluster center, and the change direction of the rainfall cloud cluster, so as to comprehensively describe the internal spatial structure and possible rainfall intensity of the rain cluster. The specific spatial structure information diagram of the rainfall cloud map is as Figure 5 shown.

[0082] In step S205, the infrared brightness temperature data, longitude and latitude information, elevation data, and the spatial structure information of the rainfall cloud cluster are input into the rainfall intensity estimation model for rainfall intensity estimation operation, and a rainfall amount prediction result is obtained. Among them, the rainfall intensity estimation model is the Attention-Unet model.

[0083] In the embodiment of the present application, the occurrence of rainfall is not only related to topographic conditions and cloud microphysical parameters. The spatial structure of the rainfall cloud cluster not only affects the distribution of rainfall, but also reveals the physical process and its spatio-temporal characteristics of rainfall. In order to comprehensively consider the influence of multiple factors on the occurrence of rainfall, by introducing the infrared brightness temperature data, longitude and latitude information, elevation data, and the spatial structure information of the rain cluster into the Attention-Unet model, a rainfall intensity estimation model based on the spatial structure of the rain cluster is constructed for rainfall intensity estimation. This model takes into account the new idea of integrating rainfall events and rainfall amounts, constructs rainfall estimation supported by the spatial structure of the rainfall cloud cluster, enhances the ability to extract rainfall areas, and enhances the interpretability of the model to a certain extent; at the same time, by introducing an attention mechanism module on the basis of the Unet model, it improves the attention of the model to rainfall-highly relevant features, plays a good supplementary role in the rainfall cloud cluster boundary, enhances the ability to capture rainfall details, and improves the model efficiency.

[0084] In step S206, the rainfall amount prediction result is output.

[0085] In the embodiment of the present application, first, the observation data of Fengyun-4B satellite is preprocessed, and the infrared brightness temperature data of Fengyun-4B satellite is spatio-temporally matched with the dual-frequency rain radar product by using the spatio-temporal matching technology to construct a sample data set; secondly, an Attention-Unet model is constructed by introducing an attention mechanism module, and the infrared brightness temperature data, auxiliary information such as longitude and latitude information, and elevation data are introduced into the model to construct a rainfall detection model; then, according to the rainfall detection result, the spatial structure information of the rainfall cloud cluster is extracted based on the rainy pixel points; finally, the infrared brightness temperature data, longitude and latitude information, elevation data and the spatial structure information of the rainfall cloud cluster of the Fengyun satellite are input into the Attention-Unet network to construct a rainfall intensity estimation model to accurately estimate the rainfall intensity.

[0086] In the embodiment of the present application, a method for estimating the rainfall intensity of a geostationary meteorological satellite is provided, including: obtaining the geostationary satellite observation data, dual-frequency rain radar product, and elevation data of the target area, and preprocessing the geostationary satellite observation data to obtain infrared brightness temperature data; using the spatio-temporal matching technology to unify the temporal resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rain radar product; inputting the infrared brightness temperature data, longitude and latitude information, and elevation data into the rainfall detection model to detect whether rainfall occurs, and obtaining the rainfall detection result of whether there is rainfall on the pixel points, where the rainfall detection model is an Attention-Unet model; performing a spatial structure information extraction operation on the pixel points determined to be rainy according to the rainfall detection result to obtain the spatial structure information of the rainfall cloud cluster; inputting the infrared brightness temperature data, longitude and latitude information, elevation data, and the spatial structure information of the rainfall cloud cluster into the rainfall intensity estimation model to perform a rainfall intensity estimation operation to obtain the rainfall prediction result, where the rainfall intensity estimation model is an Attention-Unet model; outputting the rainfall prediction result. Compared with the prior art, the present application effectively introduces the physical characteristics related to the rainfall occurrence intensity into the deep learning model by constructing the spatial structure information of the rain cluster. By combining physical information and the deep learning model, while strengthening the ability to capture small rainfall, the ability to predict the rainfall area is strengthened, and the accuracy and physical interpretability of the model are improved.

[0087] In some optional implementation manners of the embodiment of the present application, the preprocessing of the geostationary satellite observation data described above includes:

[0088] Performing a radiometric calibration operation on the geostationary satellite to convert the DN value of the infrared channel of the geostationary satellite into a meaningful radiance temperature value;

[0089] By using the longitude and latitude lookup correction method, the nominal projection of the geostationary satellite is converted into vector data to obtain the infrared brightness temperature data.

[0090] In some alternative implementation manners of the embodiments of the present application, the above-mentioned unification of the time resolution and the space resolution of the obtained infrared brightness temperature map of the geostationary satellite and the dual-frequency rain radar product by using the spatio-temporal matching technology includes:

[0091] Performing time matching on the infrared brightness temperature data of the geostationary satellite and the dual-frequency rain radar product within the same time range according to the scanning time of the global rain radar;

[0092] Linearly interpolating both the infrared brightness temperature data of the geostationary satellite and the dual-frequency rain radar product onto a grid of 0.05° * 0.05° to achieve spatial matching.

[0093] In some alternative implementation manners of the embodiments of the present application, after the step of inputting the infrared brightness temperature data, the longitude and latitude information, and the elevation data into the rainfall detection model to perform a rainfall detection operation and obtaining a rainfall detection result indicating whether rainfall occurs at a pixel point, the following steps are further included:

[0094] Evaluating the rainfall detection result according to the probability of detection (POD) index, the false alarm rate (FAR) index, and the critical success index (CSI) to obtain an evaluation result of the rainfall detection ability of the rainfall detection model.

[0095] In some alternative implementation manners of the embodiments of the present application, the above-mentioned probability of detection Is expressed as:

[0096]

[0097] The false alarm rate Is expressed as:

[0098]

[0099] The success rate Is expressed as:

[0100]

[0101] Wherein, Indicates that both the predicted data and the observed data are positive classes; Indicates that the predicted data is a negative class and the observed data is a positive class; Indicates that the predicted data is a positive class and the observed data is a negative class.

[0102] In the embodiments of the present application, the spatio-temporal matching technology refers to spatio-temporal matching of the observation images of Fengyun-4B satellite with the dual-frequency precipitation radar products carried by the Global Precipitation Measurement (GPM) satellite. Fengyun-4B satellite performs a full-disk scan every 15 minutes, and the GPM satellite performs a polar orbit scan observation of the Earth approximately every 92.6 minutes. The time range when the GPM satellite passes over the Fengyun-4B satellite can be roughly inferred from the start time and the number of scan lines of the GPM satellite scan, so as to achieve time matching between the observation images of Fengyun-4B satellite and the dual-frequency precipitation radar products. According to the result of time matching, the longitude and latitude range of the target area is determined, and the infrared brightness temperature data of Fengyun-4B satellite and the dual-frequency precipitation radar products are both linearly interpolated onto a 0.05° * 0.05° grid to achieve spatial matching.

[0103] In the embodiments of the present application, the present invention uses three scoring indices to evaluate the rainfall detection ability, namely the Probability of Detection (POD), the False Alarm Ratio (FAR), and the Critical Success Index (CSI), and their optimal values are 1, 0, and 1 respectively. These indices are all calculated based on a binary classification confusion matrix, as shown in Table 1. Their calculation formulas are as follows:

[0104]

[0105]

[0106]

[0107] Predicted as positive class Predicted as negative class Observed as positive class True Positive (TP) False Negative (FN) Observed as negative class False Positive (FP) True Negative (TN)

[0108] Table 1 Binary classification confusion matrix

[0109] In some optional implementation manners of the embodiments of the present application, after the step of inputting the infrared brightness temperature data, longitude and latitude information, elevation data, and the spatial structure information of the rainfall cloud cluster into the rainfall intensity estimation model to perform the rainfall intensity estimation operation and obtain the rainfall prediction result, the following steps are further included:

[0110] Evaluating the rainfall prediction result according to the root mean square error and the correlation coefficient to obtain the evaluation result of the rainfall prediction ability of the rainfall intensity estimation model, where the root mean square error algorithm is expressed as:

[0111]

[0112] where represents the true rainfall of the th pixel point, represents the predicted rainfall of the th pixel point, and N represents the number of pixel points;

[0113] The correlation coefficient CC is expressed as:

[0114]

[0115] where represents the actual rainfall of the th pixel point, represents the predicted rainfall of the th pixel point, represents the average value of the actual rainfall, represents the average value of the predicted rainfall, represents the standard deviation of the predicted rainfall, represents the standard deviation of the actual rainfall, and N represents the number of pixel points.

[0116] The root mean square error evaluates the error between the actual rainfall and the predicted rainfall, and its value range is greater than 0. The smaller the value, the better the performance of the model. The correlation coefficient is used to reflect the correlation degree between the actual rainfall and the predicted rainfall, and its value range is [-1, 1]. The larger the absolute value, the stronger the correlation.

[0117] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is to use a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results of theory, methods, technologies, and application systems.

[0118] Artificial intelligence basic technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. Artificial intelligence software technologies mainly include several major directions such as computer vision technology, robotics, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0119] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through computer-readable instructions. The computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0120] It should be understood that although the steps in the flowchart of the accompanying drawings are shown sequentially according to the indication of the arrows, these steps are not necessarily executed sequentially in the order indicated by the arrows. Unless otherwise clearly stated in this document, there is no strict order restriction for the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0121] Further referring to Figure 6 , as an implementation of the method shown above Figure 2 , an embodiment of a rainfall intensity estimation device for a geostationary meteorological satellite is provided in this application. This device embodiment corresponds to the method embodiment shown in Figure 2 , and this device can be specifically applied to various electronic devices.

[0122] As shown in Figure 6 , the geostationary meteorological satellite rainfall intensity estimation device 200 in the embodiment of this application includes:

[0123] A data acquisition module 210, configured to acquire geostationary satellite observation data, dual-frequency rain gauge radar products, and elevation data for a target area, and preprocess the geostationary satellite observation data to obtain infrared brightness temperature data;

[0124] A data matching module 220, configured to unify the time resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rain gauge radar products by using spatio-temporal matching technology;

[0125] A rainfall detection module 230, configured to input the infrared brightness temperature data, longitude and latitude information, and elevation data into a rainfall detection model to detect whether rainfall occurs, and obtain a rainfall detection result indicating whether rainfall occurs at a pixel point. Among them, the rainfall detection model is an Attention-Unet model;

[0126] A rainfall cloud cluster spatial structure information extraction module 240, configured to perform a spatial structure information extraction operation on the pixel points determined to have rain according to the rainfall detection result, and obtain the spatial structure information of the rainfall cloud cluster;

[0127] A rainfall intensity estimation module 250, configured to input the infrared brightness temperature data, longitude and latitude information, elevation data, and the spatial structure information of the rainfall cloud cluster into a rainfall intensity estimation model to perform a rainfall intensity estimation operation, and obtain a rainfall amount prediction result. Among them, the rainfall intensity estimation model is an Attention-Unet model;

[0128] A result output module 260, configured to output the rainfall prediction result.

[0129] In an embodiment of the present application, a geostationary meteorological satellite rainfall intensity estimation device 200 is provided, including: a data acquisition module 210, configured to acquire geostationary satellite observation data, dual-frequency rainfall radar products, and elevation data of a target area, and preprocess the geostationary satellite observation data to obtain infrared brightness temperature data; a data matching module 220, configured to use a spatio-temporal matching technique to unify the time resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rainfall radar products; a rainfall detection module 230, configured to input the infrared brightness temperature data, longitude and latitude information, and elevation data into a rainfall detection model to detect whether rainfall occurs, and obtain a rainfall detection result indicating whether rainfall occurs at a pixel point, where the rainfall detection model is an Attention-Unet model; a rainfall cloud cluster spatial structure information extraction module 240, configured to perform a spatial structure information extraction operation on the pixel points determined to have rain according to the rainfall detection result to obtain the spatial structure information of the rainfall cloud cluster; a rainfall intensity estimation module 250, configured to input the infrared brightness temperature data, longitude and latitude information, elevation data, and the spatial structure information of the rainfall cloud cluster into a rainfall intensity estimation model to perform a rainfall intensity estimation operation to obtain a rainfall prediction result, where the rainfall intensity estimation model is an Attention-Unet model; a result output module 260, configured to output the rainfall prediction result. Compared with the prior art, by constructing the spatial structure information of the rain cluster in the present application, physical features related to the rainfall occurrence intensity are effectively introduced into the deep learning model. By combining physical information and the deep learning model, while strengthening the ability to capture small rainfall, the ability to predict the rainfall falling area is strengthened, and the accuracy and physical interpretability of the model are improved.

[0130] In some optional implementation manners of the embodiment of the present application, the above data acquisition module 210 includes:

[0131] A radiometric calibration sub-module, configured to perform a radiometric calibration operation on the geostationary satellite, and convert the DN value of the infrared channel of the geostationary satellite into a meaningful radiance brightness temperature value;

[0132] A projection transformation sub-module, configured to convert the nominal projection of the geostationary satellite into vector data through a longitude and latitude lookup correction method to obtain the infrared brightness temperature data.

[0133] In some optional implementation manners of the embodiment of the present application, the above data matching module 220 includes:

[0134] A time matching sub-module, configured to perform time matching on the infrared brightness temperature data of the geostationary satellite and the dual-frequency rainfall radar products within the same time range according to the scanning time of the global rainfall radar;

[0135] A spatial matching sub-module, which is used to linearly interpolate the infrared brightness temperature data of geostationary satellites and the dual-frequency rain radar products onto a 0.05° * 0.05° grid to achieve spatial matching.

[0136] In some alternative implementation manners of the embodiment of the present application, the above-mentioned geostationary meteorological satellite rainfall intensity estimation device 200 further includes:

[0137] A rainfall detection ability evaluation module, which is used to perform a rainfall detection ability evaluation operation on the rainfall detection result according to the probability of detection (POD) index, the false alarm rate (FAR) index, and the critical success index (CSI), so as to obtain the rainfall detection ability evaluation result of the rainfall detection model. Among them, the probability of detection (POD) index is expressed as:

[0138]

[0139] The false alarm rate (FAR) index is expressed as:

[0140]

[0141] The critical success index (CSI) is expressed as:

[0142]

[0143] Among them, Indicates that both the predicted data and the observed data are positive classes; Indicates that the predicted data is a negative class and the observed data is a positive class; Indicates that the predicted data is a positive class and the observed data is a negative class.

[0144] In some alternative implementation manners of the embodiment of the present application, the above-mentioned geostationary meteorological satellite rainfall intensity estimation device 200 further includes:

[0145] A rainfall intensity prediction ability evaluation module, which is used to perform a rainfall amount prediction ability evaluation operation on the rainfall amount prediction result according to the root mean square error and the correlation coefficient, so as to obtain the rainfall amount prediction ability evaluation result of the rainfall intensity estimation model. Among them, the root mean square error Is expressed as:

[0146]

[0147] Among them, Represents the true rainfall amount of the th pixel point, Represents the predicted rainfall amount of the th pixel point, and N represents the number of pixel points;

[0148] The correlation coefficient Is expressed as:

[0149]

[0150] Among them, represents the true rainfall of the th pixel point, represents the predicted rainfall of the th pixel point, represents the average value of the true rainfall, represents the average value of the predicted rainfall, represents the standard deviation of the predicted rainfall, represents the standard deviation of the true rainfall, and N represents the number of pixel points.

[0151] To solve the above technical problems, the embodiments of the present application also provide a computer device. For details, please refer to Figure 7 , Figure 7 which is the basic structural block diagram of the computer device in the embodiments of the present application.

[0152] The computer device 300 includes a memory 310, a processor 320, and a network interface 330 that are communicatively connected to each other through a system bus. It should be noted that only the computer device 300 with components 310 - 330 is shown in the figure, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be alternatively implemented. Among them, those skilled in the art of the present technology can understand that a computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field - programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0153] The computer device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The computer device can perform human - computer interaction with users through means such as a keyboard, a mouse, a remote control, a touchpad, or a voice - controlled device.

[0154] The memory 310 includes at least one type of readable storage medium, which includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disc, etc. In some embodiments, the memory 310 may be an internal storage unit of the computer device 300, such as the hard disk or memory of the computer device 300. In other embodiments, the memory 310 may also be an external storage device of the computer device 300, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. equipped on the computer device 300. Of course, the memory 310 may also include both the internal storage unit and the external storage device of the computer device 300. In the embodiments of the present application, the memory 310 is generally used to store the operating system and various application software installed on the computer device 300, such as computer-readable instructions of the static meteorological satellite rainfall intensity estimation method. In addition, the memory 310 may also be used to temporarily store various data that have been output or will be output.

[0155] In some embodiments, the processor 320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor 320 is generally used to control the overall operation of the computer device 300. In the embodiments of the present application, the processor 320 is used to run the computer-readable instructions stored in the memory 310 or process data, such as running the computer-readable instructions of the static meteorological satellite rainfall intensity estimation method.

[0156] The network interface 330 may include a wireless network interface or a wired network interface, and this network interface 330 is generally used to establish a communication connection between the computer device 300 and other electronic devices.

[0157] For the computer device provided by the present application, the present application constructs the spatial structure information of rain clusters, effectively introduces the physical characteristics related to the rainfall occurrence intensity into the deep learning model, combines the physical information and the deep learning model, while strengthening the ability to capture small rainfall, strengthens the prediction ability of the rainfall area, and improves the accuracy and physical interpretability of the model.

[0158] The present application also provides another implementation manner, that is, to provide a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor, so that the at least one processor executes the steps of the static meteorological satellite rainfall intensity estimation method as described above.

[0159] For the computer-readable storage medium provided by the present application, the present application constructs the spatial structure information of rain clusters and effectively introduces the physical characteristics related to the rainfall occurrence intensity into the deep learning model. By combining physical information with the deep learning model, while strengthening the ability to capture small rainfall, it strengthens the ability to predict the rainfall area, and improves the accuracy and physical interpretability of the model.

[0160] Through the description of the above implementation manners, those skilled in the art can clearly understand that the method of the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation manner. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disc) and includes several instructions for causing a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in various embodiments of the present application.

[0161] Obviously, the above-described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The accompanying drawings show the preferred embodiments of the present application, but do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosed content of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions described in the foregoing specific embodiments, or perform equivalent replacements for some of the technical features. Any equivalent structure directly or indirectly using the content of the specification and drawings of the present application in other related technical fields shall be within the scope of the patent protection of the present application by the same token.

Claims

1. A method for estimating rainfall intensity of a geostationary meteorological satellite, characterized in that It includes the following steps: Obtain the geostationary satellite observation data, dual-frequency rainfall radar products, and elevation data of the target area, and preprocess the geostationary satellite observation data to obtain infrared brightness temperature data. Among them, the target area is the full disk area of the geostationary satellite, the geostationary satellite is Fengyun-4B satellite, and the elevation data is data covering the global range including land elevation and ocean depth; Use the spatio-temporal matching technology to unify the time resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rainfall radar products, and construct a sample data set; Input the sample data set, the infrared brightness temperature data, longitude and latitude information, and the elevation data into the rainfall detection model to detect whether rainfall occurs, and obtain the rainfall detection result of whether there is rainfall at the pixel points. Among them, the rainfall detection model is the Attention-Unet model introduced with an attention mechanism module, and the attention mechanism module focuses on and pays attention to the rainfall area, and at the same time sensitively captures the cloud edge and tiny rainfall distribution; Perform a spatial structure information extraction operation on the pixel points determined to have rain according to the rainfall detection result to obtain the spatial structure information of the rainfall cloud cluster. The spatial structure information includes the central position of the rainfall cloud cluster, the distance from the rainy pixel point to the cloud cluster boundary, the distance from the rainy pixel point to the cloud cluster center, and the change direction of the rainfall cloud cluster; Input the infrared brightness temperature data, the longitude and latitude information, the elevation data, and the spatial structure information of the rainfall cloud cluster into the rainfall intensity estimation model for rainfall intensity estimation operation to obtain the rainfall amount prediction result. Among them, the rainfall intensity estimation model is the Attention-Unet model; Output the rainfall intensity estimation result.

2. The rainfall intensity estimation method for a geostationary meteorological satellite according to claim 1, wherein The preprocessing of the geostationary satellite observation data includes: Perform a radiometric calibration operation on the geostationary satellite to convert the DN value of the infrared channel of the geostationary satellite into a meaningful radiance brightness temperature value; Through the longitude and latitude lookup correction method, convert the nominal projection of the geostationary satellite into vector data to obtain the infrared brightness temperature data.

3. The rainfall intensity estimation method for a geostationary meteorological satellite according to claim 1, characterized in that The use of the spatio-temporal matching technology to unify the time resolution and spatial resolution of the obtained geostationary satellite infrared brightness temperature data and the dual-frequency rainfall radar products includes: According to the scanning time of the global rain radar, perform time matching on the infrared brightness temperature data of the geostationary satellite and the dual-frequency rainfall radar products in the same time range; Linearly interpolate the infrared brightness temperature data of the geostationary satellite and the dual-frequency rainfall radar products onto a 0.05°*0.05° grid to achieve spatial matching.

4. The method for estimating rainfall intensity of a geostationary meteorological satellite according to claim 1, wherein After the step of inputting the infrared brightness temperature data, longitude and latitude information, and elevation data into the rainfall detection model to detect whether rainfall occurs and obtaining the detection result of whether there is rainfall at the pixel points, the following steps are further included: Evaluate the rainfall detection result according to the hit rate index POD, false alarm rate index FAR, and critical success index CSI to obtain the rainfall detection ability evaluation result of the rainfall detection model.

5. The rainfall intensity estimation method for a geostationary meteorological satellite according to claim 4, characterized in that, The hit rate index is expressed as: The false alarm rate index is expressed as: The key success indicators are expressed as: Among them, indicates that both the predicted data and the observed data are positive classes; indicates that the predicted data is a negative class and the observed data is a positive class; indicates that the predicted data is a positive class and the observed data is a negative class.

6. The rainfall intensity estimation method for a geostationary meteorological satellite according to claim 1, characterized in that After the step of inputting the infrared brightness temperature data, the longitude and latitude information, the elevation data and the spatial structure information of the rainfall cloud into the rainfall intensity estimation model to perform a rainfall intensity estimation operation to obtain a rainfall estimation result, the following steps are also included: Evaluate the rainfall estimation result according to the root mean square error and the correlation coefficient to obtain the evaluation result of the rainfall prediction ability of the rainfall intensity estimation model, where the root mean square error is expressed as: Among them, represents the true rainfall of the th pixel point, represents the predicted rainfall of the th pixel point, represents the number of pixel points; The correlation coefficient is expressed as: Among them, represents the true rainfall of the th pixel point, represents the predicted rainfall of the th pixel point, represents the average value of the true rainfall, represents the average value of the predicted rainfall, represents the standard deviation of the predicted rainfall, represents the standard deviation of the true rainfall, represents the number of pixel points.

7. A rainfall intensity estimation device for a geostationary meteorological satellite, characterized in that, include: A data acquisition module is used to acquire geostationary satellite observation data, dual-frequency rainfall radar products, and elevation data of the target area, and pre-process the geostationary satellite observation data to obtain infrared brightness temperature data, wherein the target area is the geostationary satellite full disk area, the geostationary satellite is Fengyun-4B, and the elevation data is data covering the global range including land elevation and ocean depth; A data matching module, used to unify the time resolution and spatial resolution of the infrared brightness temperature data and the dual-frequency rainfall radar product by using a time-space matching technology to construct a sample data set; A rainfall detection module is used to input the sample data set, the infrared brightness temperature data, the longitude and latitude information, and the elevation data into a rainfall detection model to detect whether rainfall has occurred, and obtain a rainfall detection result of whether rainfall has occurred at a pixel point. The rainfall detection model is an Attention-Unet model that introduces an attention mechanism module. The attention mechanism module focuses on and pays attention to the rainfall area, and at the same time keenly captures the edge of the cloud and the distribution of tiny rainfall; A rainfall cloud spatial structure information extraction module is used to perform a spatial structure information extraction operation on the pixel points determined to have rain according to the rainfall detection result to obtain the spatial structure information of the rainfall cloud, wherein the spatial structure information includes the center position of the rainfall cloud, the distance between the pixel points with rain and the cloud boundary, the distance between the pixel points with rain and the cloud center, and the change direction of the rainfall cloud; A rainfall intensity estimation module, used for inputting the infrared brightness temperature data, the longitude and latitude information, the elevation data and the spatial structure information of the rainfall cloud into a rainfall intensity estimation model to perform a rainfall intensity estimation operation to obtain a rainfall prediction result, wherein the rainfall intensity estimation model is an Attention-Unet model; The result output module is used to output the rainfall intensity estimation result.

8. The rainfall intensity estimation device for a geostationary meteorological satellite according to claim 7, characterized in that, The data acquisition module comprises: The radiation calibration submodule is used to perform radiation calibration operations on geostationary satellites and convert the DN value of the infrared channel of the geostationary satellite into a meaningful radiation brightness temperature value; The projection transformation submodule is used to transform the nominal projection of the geostationary satellite into vector data through the latitude and longitude lookup correction method to obtain the infrared brightness temperature data.

9. A computer device, comprising a memory and a processor, characterized in that, The memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for estimating rainfall intensity of a geostationary meteorological satellite according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the method for estimating rainfall intensity by a geostationary meteorological satellite according to any one of claims 1 to 6 are implemented.

Images