A method and system for integrating dry and wet distribution and heavy precipitation areas of star-ground precipitation.

Abstract

This invention relates to the field of data processing technology, and discloses a method and system for fusing satellite-to-ground precipitation data of dry and wet distributions and areas of heavy precipitation. The method includes: spatiotemporally gridding satellite remote sensing inversion precipitation data and ground station observation precipitation data of the target area to obtain a basic gridded satellite precipitation field and a basic gridded ground precipitation field; extracting multidimensional features from the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain a comprehensive precipitation feature dataset; constructing a dual-weight allocation benchmark; correcting the dry and wet boundaries of the basic gridded satellite precipitation field and performing difference analysis on the basic gridded ground precipitation field to obtain an objective residual field; and performing pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain a satellite-to-ground fused precipitation field. This invention can improve the efficiency of satellite-to-ground precipitation data fusion of dry and wet distributions and areas of heavy precipitation.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for fusing star-ground precipitation data with dry and wet distribution and heavy precipitation areas. Background Technology

[0002] Existing satellite-to-ground precipitation fusion technologies have significant shortcomings in the coordinated optimization of dry and wet precipitation distribution and heavy precipitation areas, making it difficult to simultaneously ensure the accuracy of characterization for both types of precipitation features. Most technologies do not fully consider the spatiotemporal heterogeneity of precipitation data, and the systematic bias correction between satellite remote sensing inversion data and ground station observation data is insufficient. This results in blurred dry and wet boundaries in the fusion results, making it impossible to accurately distinguish the boundaries between precipitation areas and areas without precipitation. Furthermore, the lack of targeted processing mechanisms for extracting information from heavy precipitation areas leads to insufficient accuracy in locating the centers of heavy precipitation and significant quantification errors in precipitation intensity, failing to meet the requirements for precise characterization of precipitation distribution details.

[0003] Existing technologies employ a relatively simplistic weighting mechanism, failing to construct a dual weighting system adapted to both dry / wet distribution and areas of heavy precipitation. This prevents dynamic adjustment of fusion weights based on precipitation characteristics across different regions. Furthermore, the multidimensional feature mining of the base gridded data is insufficient during data fusion, neglecting crucial information such as the spatial continuity and temporal evolution of precipitation. This results in significant discontinuities in the fusion results across dry / wet transition zones and the boundaries between heavy and light precipitation. Additionally, the residual analysis process lacks optimization through dry / wet boundary correction, failing to effectively eliminate the impact of local outliers. This further reduces the reliability of the fused data, making it unsuitable for applications requiring high-quality precipitation data, such as weather forecasting and hydrological simulation. Therefore, improving the efficiency of satellite-to-ground precipitation fusion that considers both dry / wet distribution and areas of heavy precipitation has become a pressing issue. Summary of the Invention

[0004] This invention provides a method and system for integrating star-ground precipitation data with dry and wet distribution and heavy precipitation areas to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for integrating dry and wet distribution and heavy precipitation areas, comprising:

[0006] S1. Acquire satellite remote sensing inversion precipitation data and ground station observation precipitation data for the target area;

[0007] S2. Spatiotemporally grid the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and the basic gridded ground precipitation field of the target area;

[0008] S3. Perform multi-dimensional feature extraction on the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain a comprehensive precipitation feature dataset of the target area;

[0009] S4. Based on the comprehensive precipitation feature dataset, construct a dual weight allocation benchmark for the dry-wet fusion weight and the heavy precipitation fusion weight in the target area;

[0010] S5. Correct the dry and wet boundaries of the basic gridded satellite precipitation field, and combine the corrected satellite precipitation field with the difference analysis of the basic gridded ground precipitation field to obtain the residual field of the target area.

[0011] S6. Based on the dual weight allocation benchmark and the residual field, perform pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain the satellite-ground fused precipitation field of the target area.

[0012] In a preferred embodiment, acquiring satellite remote sensing inversion precipitation data and ground station observation precipitation data for the target area includes:

[0013] From the remote sensing data of on-orbit meteorological satellites, infrared, microwave, and combined precipitation data covering the target area are selected as the multi-source satellite remote sensing precipitation data for the target area;

[0014] In the corresponding meteorological operational data center of the target area, the cumulative precipitation data recorded by effective surface meteorological observation stations within the same time period is extracted as the original surface station observation precipitation data of the target area.

[0015] Cross-validation is performed on the multi-source satellite remote sensing inversion precipitation data to eliminate systematic biases between different satellite data sources, thereby obtaining satellite remote sensing inversion precipitation data for the target area;

[0016] The original ground station precipitation data is subjected to quality control, invalid records are removed and obvious errors are corrected to obtain the ground station precipitation data for the target area.

[0017] In a preferred embodiment, the step of spatiotemporally gridding the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and the basic gridded ground precipitation field of the target area includes:

[0018] A standardized specification for the unified spatiotemporal grid of the target region is defined to obtain the target grid framework of the target region.

[0019] Based on the target grid framework, the satellite remote sensing inversion precipitation data is temporally resampled to obtain the satellite time-aligned data sequence of the target area;

[0020] Spatial reprojection and cropping are performed on the satellite time-aligned data sequence to obtain spatially aligned satellite data for the target region.

[0021] The spatially aligned satellite data is gridded to obtain the basic gridded satellite precipitation field of the target area;

[0022] Based on the target grid framework, spatial interpolation is performed on the precipitation data observed at the ground stations, and the estimated precipitation values ​​of the spatial grids in the target grid framework are statistically analyzed to obtain the basic gridded surface precipitation field of the target area.

[0023] In a preferred embodiment, the step of performing multidimensional feature extraction on the basic gridded satellite precipitation field and the basic gridded surface precipitation field to obtain a comprehensive precipitation feature dataset for the target region includes:

[0024] By statistically analyzing the precipitation distribution in the gridded satellite precipitation field and the gridded ground precipitation field, the statistical feature set of satellite precipitation and the statistical feature set of ground precipitation in the target area are obtained.

[0025] Spatial structure analysis is performed on the basic gridded satellite precipitation field to extract texture features that characterize the spatial continuity and gradient changes of precipitation, thereby obtaining the satellite precipitation spatial feature set of the basic gridded satellite precipitation field.

[0026] Time series analysis is performed on the basic gridded surface precipitation field to extract the time variation features that characterize the evolution of precipitation processes, thereby obtaining the surface precipitation time series feature set of the basic gridded surface precipitation field;

[0027] The satellite precipitation statistical feature set, the ground precipitation statistical feature set, the satellite precipitation spatial feature set, and the ground precipitation temporal feature set are fused and integrated to obtain the comprehensive precipitation feature set of the target area.

[0028] In a preferred embodiment, the step of constructing a dual weighting benchmark for the target region based on the integrated precipitation feature dataset and the combined dry and wet precipitation weights includes:

[0029] Based on the comprehensive precipitation feature data that centrally characterizes the spatial continuity of precipitation, the dry and wet distribution statistics of the target area are performed to obtain the background precipitation area of ​​the target area;

[0030] Based on the characteristics that centrally represent the reliability of ground observations in the comprehensive precipitation feature dataset, the precipitation intensity in the target area is calibrated at multiple scales to obtain the high-precipitation area in the target area.

[0031] The spatial continuity index of precipitation in the background precipitation area is correlated with the similarity of precipitation patterns in the surrounding grid to obtain the wet-dry fusion weight value of the target area.

[0032] The observation density and data quality level of adjacent ground stations in the high-precipitation area are quantified with confidence to obtain the heavy precipitation fusion weight value of the target area. The calculation formula of the heavy precipitation fusion weight value is as follows:

[0033] ;

[0034] In the formula, Spatial grid position in the target region The aforementioned heavy precipitation fusion weight value, To be at the spatial grid position Ground observation quality influencing factors at the location It is the hyperbolic tangent function. The preset intensity difference sensitivity coefficient, To be at the spatial grid position Standardized precipitation intensity differences at various locations;

[0035] The wet and dry fusion weight value and the heavy precipitation fusion weight value are spatially seamlessly fused, and the grid at the boundary of the fused weight region is smoothed to obtain the dual weight allocation benchmark of the target region.

[0036] In a preferred embodiment, the formula for calculating the wet-dry fusion weight value is as follows:

[0037] ;

[0038] In the formula, Spatial grid position in the target region The dry-wet fusion weight value at the location, The row and column index coordinates of the spatial grid within the target area. The spatial continuity index of precipitation, As the precipitation intensity gradient consistency factor, The preset gradient consistency adjustment coefficient, It is a natural exponential function.

[0039] In a preferred embodiment, the step of correcting the wet and dry boundaries of the basic gridded satellite precipitation field, and combining the corrected satellite precipitation field with the difference analysis of the basic gridded ground precipitation field to obtain the residual field of the target region, includes:

[0040] Based on the precipitation intensity of the grid in the basic gridded satellite precipitation field, the dry and wet boundary features characterizing the precipitation distribution are identified;

[0041] Based on the dry and wet boundary characteristics, the precipitation intensity of the corresponding grid in the basic gridded satellite precipitation field is adaptively adjusted to obtain the corrected satellite precipitation field of the target area;

[0042] The precipitation intensity values ​​of the grids in the corrected satellite precipitation field are compared with the precipitation intensity values ​​of the corresponding grids in the basic gridded ground precipitation field grid by grid-by-grid to obtain the preliminary residual field of the target area.

[0043] The preliminary residual field is spatially smoothed to eliminate local outliers, thus obtaining the residual field of the target region.

[0044] In a preferred embodiment, the step of subtracting the precipitation intensity values ​​of the grids in the corrected satellite precipitation field from the corresponding grids in the basic gridded ground precipitation field to obtain the preliminary residual field of the target area includes:

[0045] Spatial grid registration is performed between the corrected satellite precipitation field and the basic gridded ground precipitation field to obtain grid data pairs for the target area;

[0046] Based on the grid data pair, the precipitation intensity values ​​in the corrected satellite precipitation field and the basic gridded ground precipitation field are extracted synchronously according to the spatial grid order to obtain the corresponding precipitation intensity value set of the grid data pair;

[0047] Based on the corresponding precipitation intensity value set, the intensity values ​​of the basic gridded surface precipitation field are compared by difference to obtain the preliminary residual field of the target area.

[0048] In a preferred embodiment, the step of performing pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field based on the dual-weight allocation benchmark and the residual field to obtain the satellite-ground fused precipitation field of the target region includes:

[0049] Based on the fusion rule scheme in the dual weight allocation benchmark, the spatial grid of the target region is subjected to dominant rule filtering to obtain the adaptation fusion rule of the target region;

[0050] Based on the aforementioned adaptation and fusion rules, the basic gridded satellite precipitation field and the basic gridded ground precipitation field are weighted and fused to obtain the preliminary pixel-by-pixel fusion result of the target area;

[0051] The preliminary pixel-by-pixel fusion result is superimposed with the residual field after deviation compensation to obtain the deviation-corrected fused precipitation field of the target area;

[0052] The regional consistency of the deviation-corrected fused precipitation field is verified to obtain the intermediate fused precipitation field of the target region;

[0053] Based on the aforementioned dual-weight allocation benchmark, the intermediate fused precipitation field is subjected to adaptive spatial smoothing filtering in the dry-wet transition region and the boundary region between strong and weak precipitation to obtain the star-ground fused precipitation field of the target region.

[0054] To address the aforementioned problems, the present invention also provides a system for integrating dry and wet distribution and heavy precipitation areas, the system comprising:

[0055] The data acquisition module is used to acquire satellite remote sensing inversion precipitation data and ground station observation precipitation data for the target area;

[0056] The spatiotemporal gridding module is used to perform spatiotemporal gridding on the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and the basic gridded ground precipitation field of the target area.

[0057] The multidimensional feature extraction module is used to extract multidimensional features from the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain a comprehensive precipitation feature dataset of the target area.

[0058] The dual-weight construction module is used to construct a dual-weight allocation benchmark for the dry-wet fusion weight and the heavy precipitation fusion weight in the target area based on the comprehensive precipitation feature dataset.

[0059] The residual analysis module is used to correct the wet and dry boundaries of the basic gridded satellite precipitation field, and combine the corrected satellite precipitation field to perform difference analysis on the basic gridded ground precipitation field to obtain the residual field of the target area.

[0060] The weighted fusion module is used to perform pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field based on the dual weight allocation benchmark and the residual field, so as to obtain the satellite-ground fused precipitation field of the target area.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. This invention effectively eliminates systematic biases and erroneous records from data sources by cross-validating multi-source satellite remote sensing inversion data and controlling the quality of ground station observation data, ensuring the reliability of basic data. Spatiotemporal gridding processing achieves spatiotemporal alignment between satellite and ground data, providing a unified grid framework for fusion. Multidimensional feature extraction integrates statistical, spatial, and temporal features of precipitation, comprehensively depicting the distribution patterns and evolution characteristics of precipitation. Based on this, a dual allocation benchmark of dry / wet fusion weights and heavy precipitation fusion weights can accurately adapt to the characteristic requirements of different precipitation regions, improving the targeting and rationality of the fusion process and laying a solid foundation for high-quality fusion results.

[0063] 2. This invention clarifies the boundary between precipitation and non-precipitation areas by correcting the dry and wet boundaries of the basic gridded satellite precipitation field. Residual analysis using the corrected data effectively eliminates the influence of local outliers. During the pixel-by-pixel weighted fusion process, fusion bias is further calibrated through adaptive fusion rule selection, bias compensation overlay, and regional consistency verification. Adaptive spatial smoothing filtering in the dry-wet transition zone and the boundary between strong and weak precipitation ensures the spatial continuity of the fusion results. The resulting satellite-to-ground fused precipitation field accurately characterizes the dry and wet distribution features and precisely quantifies the areas and intensities of heavy precipitation, significantly improving the overall accuracy of satellite-to-ground precipitation fusion and providing high-quality precipitation data support for meteorological forecasting, hydrological simulation, and other related applications. Attached Figure Description

[0064] Figure 1 A schematic flowchart illustrating a method for integrating dry and wet distribution and heavy precipitation areas according to an embodiment of the present invention;

[0065] Figure 2 A functional block diagram of a ground-to-space precipitation fusion system for dry and wet distribution and heavy precipitation areas provided in an embodiment of the present invention;

[0066] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0067] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0068] This application provides a method for fusing satellite-to-ground precipitation data based on wet / dry distribution and heavy precipitation areas. The execution entity of this method includes, but is not limited to, at least one electronic device configured to execute the method provided in this application, such as a server or a terminal. In other words, the method can be executed by software or hardware installed on a terminal device or server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cluster of cloud servers. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0069] Reference Figure 1 The diagram shown is a flowchart illustrating a method for integrating dry / wet distribution and heavy precipitation areas according to an embodiment of the present invention. In this embodiment, the method for integrating dry / wet distribution and heavy precipitation areas includes:

[0070] S1. Acquire satellite remote sensing inversion precipitation data and ground station observation precipitation data for the target area;

[0071] In this embodiment of the invention, acquiring satellite remote sensing inversion precipitation data and ground station observation precipitation data of the target area includes:

[0072] From the remote sensing data of on-orbit meteorological satellites, infrared, microwave, and combined precipitation data covering the target area are selected as the multi-source satellite remote sensing precipitation data for the target area;

[0073] In the corresponding meteorological operational data center of the target area, the cumulative precipitation data recorded by effective surface meteorological observation stations within the same time period is extracted as the original surface station observation precipitation data of the target area.

[0074] Cross-validation is performed on the multi-source satellite remote sensing inversion precipitation data to eliminate systematic biases between different satellite data sources, thereby obtaining satellite remote sensing inversion precipitation data for the target area;

[0075] The original ground station precipitation data is subjected to quality control, invalid records are removed and obvious errors are corrected to obtain the ground station precipitation data for the target area.

[0076] From the raw remote sensing data acquired by meteorological satellites currently in normal on-orbit operation, the latitude and longitude boundary parameters of the target area are first analyzed by a computer program to generate a standardized regional description file. Then, satellite data metadata analysis tools are used to extract key information such as the observation swath width, coverage of latitude and longitude range, observation timestamp, sensor type, and data format of each satellite. By comparing the regional description file with the satellite observation coverage, satellite remote sensing data that fully or partially covers the target area and whose observation timestamps are consistent with the preset time period are selected. Subsequently, based on sensor type classification logic, infrared remote sensing inversion precipitation data, microwave remote sensing inversion precipitation data, and infrared-microwave joint inversion precipitation data are extracted from the selected satellite data. During the extraction process, the precipitation data from different satellite platforms and different formats are uniformly converted into a standardized raster data format by a computer program, and an index label containing metadata such as data source, sensor model, and inversion method is established. All three types of standardized raster data with index labels together constitute the multi-source satellite remote sensing inversion precipitation data of the target area, ensuring that the data can be directly called by subsequent processing steps and that the source is traceable.

[0077] Accessing the structured database of the official meteorological data center corresponding to the target area via a computer network, and using structured query language based on the latitude and longitude range of the target area, the system retrieves the basic information table of all ground meteorological observation stations within that area that have completed registration and are marked as "operating normally." This information table contains structured fields such as unique station identifiers, precise latitude and longitude coordinates, altitude, observation equipment parameters, and calibration records. Then, the program reads the timestamp range of precipitation data retrieved from the previous multi-source satellite remote sensing step, generates time query conditions for the same time period, and combines this with the "operating status" field in the station basic information table to filter out stations with no "fault reports" within that time period. A list of valid ground meteorological observation stations marked "Maintenance Shutdown" is generated. Based on this list, a database connection tool is used to extract the cumulative precipitation records of each valid station at each observation time within the corresponding time period from the ground observation data table of the meteorological business data center. During the extraction process, a computer program associates and binds each precipitation record with a unique station identifier and observation timestamp, forming a structured data set containing "station identifier - timestamp - cumulative precipitation". This data set is the original ground station observation precipitation data of the target area. At the same time, the program verifies that the timestamp format of each data is consistent with the time resolution of the satellite data to avoid data misalignment in the time dimension.

[0078] Three types of standardized raster data from multi-source satellite remote sensing precipitation data were read by a computer program. Based on the latitude and longitude range of the target area, a spatial grid division algorithm was used to generate a uniformly distributed target grid system. The grid division criteria were based on the principle of adapting to the spatial resolution of the precipitation data and completely covering the target area. Subsequently, through data resampling processing logic, for raster data with a resolution higher than the target grid, weights were assigned according to the spatial distance between the data points and the grid center, and a weighted average algorithm was used to integrate multiple data points into the corresponding target grid. For raster data with a resolution lower than the target grid, a mean-filling algorithm for effective data points within the grid was used to expand individual data points to the corresponding target grid, ensuring that all three types of data were converted into standardized data based on the same target grid system. The program automatically selected areas within the target area that were free from cloud cover, topographic obstruction, and had a single underlying surface type as cross-validation sample areas. The selection of sample areas was automatically determined by calling digital elevation model data and underlying surface type classification data. Then, each target surface type within the sample area was extracted. The precipitation inversion values ​​of the three types of data in the target grid are compared one by one with the inversion results of the three types of data in the same grid through a computer program to establish a bias identification model. When the inversion values ​​of a certain type of data in most grids of the sample area show a consistent trend of being higher or lower than the corresponding inversion values ​​of the other two types of data, it is determined that there is a systematic bias in that type of data. The fluctuation degree of the three types of data in the sample area is calculated by the program, and the data with the smallest fluctuation degree and the highest matching degree with the historical precipitation data of the same period in the sample area is selected as the reference benchmark. Based on the benchmark, the two types of data with systematic bias are corrected by the grid-by-grid data adjustment program. If the data is generally higher, the difference between the benchmark data and the data is subtracted grid by grid; if the data is generally lower, the difference is added grid by grid. After the correction is completed, the program is used to cross-compare again to confirm that the difference of the inversion values ​​of the three types of data in the same grid is within a reasonable range, that is, the systematic bias has been eliminated. Finally, the corrected three types of grid data are merged into a unified raster dataset through the data integration program to form the satellite remote sensing inversion precipitation data of the target area.

[0079] The computer program loads the structured dataset of precipitation data from the original ground stations and initiates a data integrity verification process. The program checks the field integrity of each cumulative precipitation record line by line. If a station has multiple consecutive observation times with null values, meaningless characters, or missing fields in the cumulative precipitation field within a specified time period, these records are automatically marked as invalid. Based on historical precipitation statistics and climate zoning characteristics of the target area, a database of reasonable precipitation intervals is established using the computer program. This database stores the normal fluctuation range of precipitation according to climate zoning and seasonal periods. The program calls this database and compares each cumulative precipitation record with the reasonable interval for its corresponding zoning and time period. If a record's precipitation value exceeds the reasonable interval and significantly contradicts the precipitation trend of adjacent observation times at that station, it is determined to be obviously erroneous data. Invalid records are directly removed from the structured dataset using a data filtering program. For obviously erroneous data, the program automatically retrieves the cumulative precipitation data of the time periods immediately before and after the erroneous record at that station. If the data at adjacent time points are continuous and conform to the normal variation pattern of precipitation, and other valid stations with similar spatial distances around the erroneous station are retrieved using a spatial distance calculation algorithm, the cumulative precipitation data of these neighboring stations at the same time are extracted, and the average value of the neighboring station data is calculated using an arithmetic mean algorithm. Taking into account the differences in terrain features between the erroneous station and its neighbors, the program fine-tunes and replaces the erroneous record's value. If the program finds that the adjacent time point data of the erroneous record is discontinuous or that there is no valid data support from neighboring stations, the erroneous record is marked as invalid and removed through a filtering program. After the invalid record removal and obvious error correction are completed, a secondary data verification program comprehensively checks the remaining structured dataset to confirm that no invalid records or obvious errors have been missed. The resulting standardized structured dataset is the ground station observation precipitation data for the target area. Throughout the processing, the program automatically records the reason for removing each invalid record, the basis for correcting the erroneous data, and the calculation process, generating a traceable data processing log and storing it in association with the final dataset.

[0080] The beneficial effects are as follows: through standardized data screening and format conversion processes, the source of precipitation data retrieved from multi-source satellite remote sensing is traceable, the format is uniform, and the spatiotemporal matching is good. With the help of automated spatial grid unification and cross-validation processing, systematic biases between different satellite data sources are effectively eliminated, improving the consistency and reliability of satellite precipitation data. Through structured data verification, invalid record removal, and obvious error correction operations, the integrity and accuracy of precipitation data observed by ground stations are guaranteed. The entire data processing process is fully traceable and auditable. The quality of the two types of precipitation data produced finally meets the requirements of subsequent applications, providing high-quality and standardized data support for related precipitation analysis and research.

[0081] S2. Spatiotemporally grid the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and the basic gridded ground precipitation field of the target area;

[0082] In this embodiment of the invention, the step of spatiotemporally gridding the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and the basic gridded ground precipitation field of the target area includes:

[0083] A standardized specification for the unified spatiotemporal grid of the target region is defined to obtain the target grid framework of the target region.

[0084] Based on the target grid framework, the satellite remote sensing inversion precipitation data is temporally resampled to obtain the satellite time-aligned data sequence of the target area;

[0085] Spatial reprojection and cropping are performed on the satellite time-aligned data sequence to obtain spatially aligned satellite data for the target region.

[0086] The spatially aligned satellite data is gridded to obtain the basic gridded satellite precipitation field of the target area;

[0087] Based on the target grid framework, spatial interpolation is performed on the precipitation data observed at the ground stations, and the estimated precipitation values ​​of the spatial grids in the target grid framework are statistically analyzed to obtain the basic gridded surface precipitation field of the target area.

[0088] Based on the geographical characteristics of the target area and the application requirements of precipitation data, a computer program is used to define the spatial grid division standards. Using latitude and longitude as a reference, the fixed coverage area of ​​each grid is determined to ensure complete grid coverage of the target area without overlap or omission. Simultaneously, a unified temporal resolution standard is set, specifying the data recording time intervals and time node formats. The spatial grid specifications and temporal resolution standards are integrated into a standardized parameter set. This parameter set is then transformed into a structured data framework containing unique grid identifiers, grid boundary coordinates, and time node information. This structured data framework serves as the target grid framework for the target area. During the framework generation process, the program verifies the fit between the grid boundaries and the target area boundaries to ensure that no grids exceed the area or are not covered.

[0089] The computer program reads the original timestamp information of the satellite remote sensing inversion precipitation data and compares it one by one with the time resolution and time nodes set in the target grid framework. If the original time interval of the satellite remote sensing inversion precipitation data is inconsistent with the time resolution of the target grid framework, for cases where the time interval is less than the target resolution, the program merges multiple satellite precipitation data within the same target time node and takes the average of all data within that time period as the precipitation data for the corresponding target time node. For cases where the time interval is greater than the target resolution, the program performs linear extension based on the changing trend of adjacent original data to split the precipitation data corresponding to each target time node. The continuous precipitation data sequence formed by arranging all target time nodes in sequence is the satellite time-aligned data sequence of the target area. During the processing, the program automatically records the specific operation details of time resampling to ensure data traceability.

[0090] The computer program parses the original spatial projection parameters of the satellite time-aligned data sequence and reads the standard spatial projection parameters set in the target grid frame. A projection conversion program transforms the spatial coordinate system of the satellite time-aligned data sequence into a projection system consistent with the target grid frame. During the conversion, the program uses a coordinate mapping algorithm to accurately map each spatial point of the original data to the coordinate position of the new projection system, ensuring that the spatial position accuracy of the data is not affected. After the projection conversion is completed, the program performs spatial clipping processing on the converted satellite data based on the latitude and longitude boundaries of the target grid frame, removing data that exceeds the target area boundary and retaining only the satellite precipitation data falling within the target grid frame. The clipped satellite data is the spatially aligned satellite data for the target area.

[0091] A computer program spatially matches the spatially aligned satellite data with each grid cell in the target grid framework to determine the grid cell corresponding to each satellite data. For each grid cell, the program collects all spatially aligned satellite data falling within that cell, performs consistency checks on these data, and removes outliers that significantly deviate from the overall trend of the data within that cell. Subsequently, the remaining valid data is integrated and calculated, and the average of all valid data is taken as the precipitation data value for that grid cell. The program sequentially completes the precipitation data calculation for all grid cells according to the order of their unique grid identifiers, and associates and binds the unique identifier of each grid cell with its corresponding precipitation data value to form a structured raster data set. This raster data set is the basic gridded satellite precipitation field for the target area.

[0092] The computer program reads the station coordinates and precipitation records from the ground station observation precipitation data, spatially matches the station coordinates with the grid cells of the target grid framework, and determines the grid cell to which each station belongs and its surrounding adjacent grid cells. For grid cells in the target grid framework that are covered by ground stations, the average precipitation data of all stations in that grid is directly used as the preliminary estimated precipitation value for that grid. For grid cells that are not covered by ground stations, the program selects the precipitation data of all effective ground stations within a certain range around them, determines the weight based on the spatial distance between the station and the grid cell (the closer the distance, the greater the weight), and obtains the estimated precipitation value for that grid cell through weighted calculation. The program completes the statistical calculation of the estimated precipitation values ​​of all grid cells in the target grid framework one by one, and organizes the unique identifier of each grid cell and the corresponding estimated precipitation value into raster data according to the spatial layout. This raster data is the basic gridded ground precipitation field of the target area. During the statistical process, the program records the calculation basis of the estimated precipitation value of each grid cell, including the station information and weight allocation involved in the calculation.

[0093] The beneficial effects are as follows: by establishing a unified target grid framework, a standardized benchmark is provided for precipitation data processing; temporal resampling ensures accurate matching of satellite data with target time specifications; spatial reprojection and clipping ensure a high degree of fit between the spatial location of satellite data and the target area; data gridding and spatial interpolation respectively transform satellite and ground data into gridded fields with unified structures, effectively improving the spatiotemporal consistency and integrity of the two types of precipitation data, avoiding data misalignment or information loss problems; and the resulting basic gridded satellite precipitation field and basic gridded ground precipitation field provide high-quality and standardized data support for subsequent comparative analysis and application of precipitation data.

[0094] S3. Perform multi-dimensional feature extraction on the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain a comprehensive precipitation feature dataset of the target area;

[0095] In this embodiment of the invention, the step of performing multi-dimensional feature extraction on the basic gridded satellite precipitation field and the basic gridded surface precipitation field to obtain a comprehensive precipitation feature dataset for the target area includes:

[0096] By statistically analyzing the precipitation distribution in the gridded satellite precipitation field and the gridded ground precipitation field, the statistical feature set of satellite precipitation and the statistical feature set of ground precipitation in the target area are obtained.

[0097] Spatial structure analysis is performed on the basic gridded satellite precipitation field to extract texture features that characterize the spatial continuity and gradient changes of precipitation, thereby obtaining the satellite precipitation spatial feature set of the basic gridded satellite precipitation field.

[0098] Time series analysis is performed on the basic gridded surface precipitation field to extract the time variation features that characterize the evolution of precipitation processes, thereby obtaining the surface precipitation time series feature set of the basic gridded surface precipitation field;

[0099] The satellite precipitation statistical feature set, the ground precipitation statistical feature set, the satellite precipitation spatial feature set, and the ground precipitation temporal feature set are fused and integrated to obtain the comprehensive precipitation feature set of the target area.

[0100] Complete raster data from both the basic gridded satellite precipitation field and the basic gridded ground precipitation field are read. A computer program iterates through all grids within the target area, using unique grid identifiers, and extracts the specific precipitation intensity value for each grid in both precipitation fields. The precipitation intensity values ​​of all grids in the basic gridded satellite precipitation field are then aggregated to obtain the total cumulative precipitation. Simultaneously, the highest and lowest precipitation intensity values ​​among all grids are identified to determine the distribution range of precipitation intensity. The number of grids falling within different precipitation intensity ranges and their proportion of the total number of grids are counted. The concentrated distribution range of precipitation intensity values ​​across all grids is then calculated. These statistical results are associated and bound with the spatial identifiers and time ranges of the corresponding grids to form a structured set of satellite precipitation statistical features. Using the same statistical logic and operational procedures, the total precipitation intensity values ​​of all grids in the basic gridded ground precipitation field are calculated, extreme values ​​are extracted, intervals are divided statistically, and concentrated ranges are analyzed. The relevant statistical results are associated with grid identifiers and time ranges to form a structured set of ground precipitation statistical features.

[0101] Spatial structure analysis is performed on the gridded satellite precipitation field data. A computer program sets a fixed-range neighboring grid query window for each grid cell, extracting the precipitation intensity values ​​of the current grid cell and all neighboring grid cells within the query window. The difference in precipitation intensity between the current grid cell and each neighboring grid cell is compared. If the difference is within a very small range and multiple consecutive neighboring grid cells maintain similar precipitation intensity values, the grid cell is considered to have good spatial continuity, and its continuity status and the number of consecutive neighboring grid cells are recorded. Simultaneously, the precipitation intensity difference between the current grid cell and each neighboring grid cell is calculated, and the results are then used to determine the spatial continuity of the grid cell. The magnitude determines the strength of precipitation gradient changes, and the positive or negative direction of the difference clarifies the direction of gradient changes. The target area is further divided into multiple continuous small spatial units. The precipitation distribution pattern of all grids in each unit is analyzed to determine whether the precipitation in the unit is uniformly distributed or has a gradient distribution. The uniformity and continuity characteristics of precipitation texture in the unit are extracted. The spatial continuity status of each grid, the number of continuous grids, the strength and direction of gradient changes, and the texture characteristics of the small spatial unit are associated and integrated with the unique grid identifier to form a satellite precipitation spatial feature set of the basic gridded satellite precipitation field.

[0102] The precipitation intensity records of each grid in the basic gridded surface precipitation field are sorted chronologically to form time series data for each grid. A computer program analyzes the evolution of the time series data grid by grid, comparing precipitation intensity values ​​at adjacent time points to determine the trend of precipitation intensity changes. Key time points showing significant changes in precipitation intensity and the magnitude of these changes before and after each point are recorded. The duration of the precipitation process in each grid is calculated, i.e., the time span from the first occurrence of a non-zero precipitation intensity value to the last occurrence of a non-zero precipitation intensity value. The number of times a non-zero precipitation intensity value occurs per unit time in that grid is also calculated to determine the frequency of precipitation occurrence. The fluctuation of precipitation intensity values ​​in the time series is analyzed. If the fluctuation range of precipitation intensity values ​​at multiple consecutive time points is small, the fluctuation is considered stable; if the fluctuation range is large, the fluctuation is considered violent. Information such as the time trend, key time points, fluctuation range, precipitation duration, frequency, and fluctuation status of each grid are integrated with the grid's unique identifier and time series range to form a set of surface precipitation time series features for the basic gridded surface precipitation field.

[0103] A unified feature integration framework is established using computer programs. With unique grid identifiers and data time nodes as core association fields, the framework standardizes the fields of satellite precipitation statistical feature sets, surface precipitation statistical feature sets, satellite precipitation spatial feature sets, and surface precipitation temporal feature sets, ensuring consistency in field names and data formats for fields with the same meaning across different feature sets. Based on the correspondence between unique grid identifiers and time nodes, the data of the four feature sets are matched and associated one by one, aggregating statistical, spatial, and temporal features corresponding to the same grid and time period. During aggregation, redundant information from duplicate records is automatically removed, retaining only complementary data from each feature set. All aggregated feature data undergoes integrity verification to ensure that no statistical, spatial, or temporal features are missing for each grid and time period. If a feature is found to be missing, it is reasonably supplemented by the feature data trends of adjacent time periods for that grid or by similar feature data from surrounding grids. After verification, all feature data is organized in a structured format of "grid identifier - time node - statistical feature - spatial feature - temporal feature," forming a comprehensive precipitation feature dataset that fully covers the target area and includes multi-dimensional precipitation characteristics.

[0104] The beneficial effects are as follows: by comprehensively statistically analyzing the precipitation distribution of two types of basic gridded precipitation fields, the statistical characteristics of satellite and ground precipitation are accurately obtained; the spatial structure of satellite precipitation fields and the temporal evolution of ground precipitation fields are deeply analyzed; key features characterizing the spatial continuity, gradient change, and temporal change trend of precipitation are extracted respectively; after standardization and precise correlation aggregation, redundant information is eliminated and missing features are supplemented, achieving a complete fusion of multi-dimensional precipitation features. The resulting comprehensive precipitation feature dataset covers core information in statistics, space, and time series, with good data consistency and completeness, providing comprehensive and reliable feature support for the subsequent construction of dual-weight allocation benchmarks and accurate fusion of precipitation data.

[0105] S4. Based on the comprehensive precipitation feature dataset, construct a dual weight allocation benchmark for the dry-wet fusion weight and the heavy precipitation fusion weight in the target area;

[0106] In this embodiment of the invention, the step of constructing a dual weight allocation benchmark for the target region based on the comprehensive precipitation feature dataset and the combined dry and wet precipitation weights includes:

[0107] Based on the comprehensive precipitation feature data that centrally characterizes the spatial continuity of precipitation, the dry and wet distribution statistics of the target area are performed to obtain the background precipitation area of ​​the target area;

[0108] Based on the characteristics that centrally represent the reliability of ground observations in the comprehensive precipitation feature dataset, the precipitation intensity in the target area is calibrated at multiple scales to obtain the high-precipitation area in the target area.

[0109] The spatial continuity index of precipitation in the background precipitation area is correlated with the similarity of precipitation patterns in the surrounding grid to obtain the wet-dry fusion weight value of the target area.

[0110] The observation density and data quality level of adjacent ground stations in the high-precipitation area are quantified with confidence to obtain the heavy precipitation fusion weight value of the target area. The calculation formula of the heavy precipitation fusion weight value is as follows:

[0111] ;

[0112] In the formula, Spatial grid position in the target region The aforementioned heavy precipitation fusion weight value, To be at the spatial grid position Ground observation quality influencing factors at the location It is the hyperbolic tangent function. The preset intensity difference sensitivity coefficient, To be at the spatial grid position Standardized precipitation intensity differences at various locations;

[0113] The wet and dry fusion weight value and the heavy precipitation fusion weight value are spatially seamlessly fused, and the grid at the boundary of the fused weight region is smoothed to obtain the dual weight allocation benchmark of the target region.

[0114] The formula for calculating the wet-dry fusion weight value is as follows:

[0115] ;

[0116] In the formula, Spatial grid position in the target region The dry-wet fusion weight value at the location, The row and column index coordinates of the spatial grid within the target area. The spatial continuity index of precipitation, As the precipitation intensity gradient consistency factor, The preset gradient consistency adjustment coefficient, It is a natural exponential function.

[0117] From the comprehensive precipitation feature dataset, computer programs extract relevant data characterizing the spatial continuity of precipitation, including precipitation values ​​for each grid, the magnitude of changes in precipitation values ​​between adjacent grids, and the distribution of precipitation duration. Based on these features, all grids within the target area are analyzed one by one to determine the precipitation correlation between each grid and its neighboring grids. If the precipitation values ​​of multiple consecutive grids within a certain area change gradually, the precipitation duration is consistent, and there is no obvious interruption, it is determined to be a humid continuous area. If most grids within a certain area have no precipitation records or only sporadic and discontinuous precipitation records, and there is no stable precipitation correlation between adjacent grids, it is determined to be an arid discrete area. All humid continuous areas, arid discrete areas, and transitional areas between the two within the target area are integrated as a whole to clarify the spatial boundary of each area and form a basic pattern of precipitation distribution that completely covers the target area. This pattern is the background precipitation area of ​​the target area. Throughout the statistical process, the program records the criteria for determining each area to ensure that the area division is traceable.

[0118] Feature information characterizing the reliability of ground observations is extracted from the comprehensive precipitation feature dataset. This includes the integrity of precipitation data from ground observation stations, historical observation equipment calibration records, data error correction records, and cross-site data consistency verification results. Based on this feature information, a multi-scale calibration system is constructed. First, starting from a single grid scale, the precipitation intensity data within the grid is compared one by one with the observation data from the grid and neighboring ground stations to correct precipitation intensity values ​​with large deviations from the station observation data. Next, starting from the regional scale, the target area is divided into multiple sub-regions, and the average precipitation intensity of each sub-region is calculated and compared with the historical average precipitation intensity of the same period in that sub-region to calibrate the overall precipitation intensity deviation of the region. Finally, starting from the time scale, the precipitation intensity data is checked for consistency with the ground observation data of the corresponding time period according to different precipitation occurrence periods to correct time-specific deviations. After the above multi-scale collaborative calibration, grids with precipitation intensity significantly higher than the average precipitation level of the target area and whose data reliability meets the standard after calibration are selected. These grids are integrated according to spatial distribution to form continuous or discrete regions, which are the high-precipitation areas of the target area.

[0119] The spatial continuity index of precipitation for each grid within a background precipitation area is calculated using a computer program. The calculation process involves statistically analyzing the differences in precipitation values ​​between the current grid and its neighboring grids; the smaller the difference, the higher the continuity index. Simultaneously, the precipitation morphology characteristics of each grid are analyzed, including precipitation coverage, precipitation intensity trends, and precipitation start-end time patterns. The precipitation morphology characteristics of this grid are then compared one by one with those of all grids within a certain surrounding area to determine the degree of similarity. This similarity is comprehensively judged based on dimensions such as the overlap ratio of precipitation coverage, consistency of intensity trends, and the degree of agreement between start-end times. A higher overlap ratio and a better trend indicate a higher continuity index. The greater the consistency and the higher the degree of agreement, the higher the similarity. The spatial continuity index of precipitation in each grid is correlated with the precipitation pattern similarity of the surrounding grids. Grids with both high continuity index and high similarity are assigned higher dry-wet fusion weight values, while grids with both low continuity index and low similarity are assigned lower dry-wet fusion weight values. Grids in between are assigned corresponding intermediate weight values ​​based on the specific correlation between the index and similarity. The weight values ​​of all grids in the background precipitation area are calculated one by one to form a dataset containing a unique identifier for each grid and its corresponding weight value. This dataset is the dry-wet fusion weight value of the target area.

[0120] For each grid within a high-precipitation area, a computer program is used to count the number of ground observation stations within a certain spatial range around it, determining the observation density of neighboring ground stations for that grid; the more stations, the higher the observation density. Simultaneously, based on indicators such as precipitation data integrity, data error rate, and equipment calibration frequency, the observation data quality of each station is graded, with higher grades indicating more reliable data quality. Confidence metrics are quantified for each grid based on observation density and data quality grade. Grids with high observation density and high data quality grades from surrounding stations have a high confidence quantification result and are assigned a higher heavy precipitation fusion weight value; grids with low observation density or low data quality grades from surrounding stations have a low confidence quantification result and are assigned a lower heavy precipitation fusion weight value. For grids exhibiting different combinations of observation density and data quality grade, corresponding weight values ​​are assigned according to the combined influence of both. The weight values ​​of all grids within the high-precipitation area are quantified one by one, forming a dataset containing a unique identifier for each grid and its corresponding weight value. This dataset represents the heavy precipitation fusion weight value for the target area.

[0121] The ground observation quality impact factor is obtained by assessing the reliability of ground observation data at a spatial grid location. The assessment first collects operational status data of the ground observation equipment at that grid location, including equipment calibration records, failure frequency, and data transmission stability. Simultaneously, the completeness of the observation data is analyzed, and the proportion of missing data is statistically analyzed. Then, the degree of interference from the observation environment on the data is detected, such as the presence of obstructions or the impact of extreme weather. The equipment status, data completeness, and environmental interference are quantified into corresponding values ​​according to preset rules. A weighted sum is then used to obtain a comprehensive quantitative result, which is the ground observation quality impact factor. The more reliable the observation data, the larger the factor value.

[0122] The preset intensity difference sensitivity coefficient is determined through calibration using historical heavy precipitation fusion data. During calibration, a large amount of ground observation data of precipitation processes of different intensities and precipitation data from other sources are collected. The fusion weight value under different coefficient values ​​is calculated, the fusion result is compared with the actual precipitation situation, the fusion error is statistically analyzed, and the coefficient value is adjusted until the fusion error reaches its minimum. The coefficient value at this point is the preset intensity difference sensitivity coefficient. This coefficient determines the degree of influence of standardized precipitation intensity difference on the fusion weight value.

[0123] Standardized precipitation intensity difference is the product of standardizing precipitation intensity differences at grid locations. The calculation involves first obtaining the observed precipitation intensity value and the reference precipitation intensity value at that grid location, then subtracting the reference value from the observed precipitation intensity value to obtain the original precipitation intensity difference. Raw precipitation intensity difference data for all grid locations during historical precipitation events in the region are collected, and the standard deviation of these data is calculated. The original precipitation intensity difference is then divided by this standard deviation to obtain the standardized precipitation intensity difference. This process eliminates the influence of differences in the magnitude of intensity differences across different precipitation events.

[0124] The heavy precipitation fusion weight value is an indicator that quantifies the contribution of ground observation data at a spatial grid location to the heavy precipitation fusion process. Its value directly determines the influence weight of the observation data at that location on the final fusion result.

[0125] The ground observation quality impact factor, as a fundamental factor, directly reflects the reliability of the observation data at the grid location. The higher the quality of the observation data, the larger the value of this factor, providing a reliable foundation for the fusion weight value and ensuring that high-quality observation data has a fundamental advantage in the fusion process.

[0126] The hyperbolic tangent function is used to perform a nonlinear transformation on the product of standardized precipitation intensity difference and intensity difference sensitivity coefficient. The transformation characteristic of this function is to map the input value to a fixed interval, avoiding abnormal fluctuations in the weight value due to excessive intensity difference. At the same time, when the intensity difference is within a reasonable range, the weight value can show a smooth response with the difference change, highlighting the influence of significant intensity difference while avoiding the interference of extreme values.

[0127] The standardized precipitation intensity difference reflects the degree of deviation between the observed precipitation and the reference precipitation at the grid location. After combining the intensity difference sensitivity coefficient, the intensity influence factor is obtained by hyperbolic tangent function transformation. This factor is multiplied by the ground observation quality influence factor to achieve a synergistic consideration of the reliability of observation data and the degree of intensity deviation.

[0128] The final weight value for heavy precipitation fusion integrates the quality and intensity differences of the observation data into a quantitative index. The larger the value, the more reliable the observation data at that grid location and the more reasonable the intensity deviation. It should be given a higher weight in heavy precipitation data fusion, and vice versa. By using this weight value to fuse multi-source heavy precipitation data, the accuracy and reliability of the fused data can be effectively improved, providing more accurate data support for heavy precipitation monitoring and early warning.

[0129] A computer program spatially matches the wet / dry fusion weight values ​​with the raster datasets corresponding to the heavy precipitation fusion weight values, ensuring that the grid framework, spatial projection, and resolution of the two datasets are completely consistent, and that each grid corresponds to a unique wet / dry fusion weight value and a heavy precipitation fusion weight value. For each grid, the two weight values ​​are comprehensively calculated according to the characteristics of its corresponding precipitation area. Grids with a higher proportion of background precipitation area use the wet / dry fusion weight value as the primary reference and the heavy precipitation fusion weight value as a secondary supplement, while grids with a higher proportion of high-value precipitation area use the heavy precipitation fusion weight value as the primary reference and the wet / dry fusion weight value as a secondary supplement, achieving seamless spatial fusion of the two weight values. For grids at the boundary of the fused weighted areas, the program analyzes the difference in weight values ​​between the boundary grids and the grids in the adjacent areas, and fine-tunes the weight values ​​of grids with large differences, so that the weight values ​​of the boundary grids gradually transition to the weight value level of the adjacent areas, avoiding abrupt changes in weight values. After completing the fusion of all grids and smoothing of the boundaries, a standardized weighted dataset with continuous and reasonable grid weight values ​​that fully covers the target area is formed. This dataset serves as the dual weight allocation benchmark for the target area.

[0130] The spatial continuity index of precipitation is calculated by analyzing the precipitation distribution characteristics of a spatial grid location and its neighborhood. The calculation first determines a fixed range of neighboring grids around the target grid location, extracts precipitation observations from the target grid and all neighboring grids, calculates the absolute difference between the precipitation value of the target grid and the precipitation values ​​of each neighboring grid, sums all the absolute differences, and takes the average. This average is then subtracted from a fixed baseline value to obtain a preliminary continuity value. This preliminary continuity value is then mapped to a fixed interval through a linear transformation. The values ​​within this interval are the spatial continuity index of precipitation. The more continuous the precipitation distribution between the target grid and its neighboring grids, the larger the index value.

[0131] The precipitation intensity gradient consistency factor is obtained by comparing the precipitation intensity gradient characteristics of the target grid and its neighboring grids. Precipitation intensity data of the target grid and its neighboring grids are obtained, and the precipitation intensity gradient of each grid is calculated, including the gradient direction and gradient magnitude. The gradient direction of the target grid is compared with the gradient directions of each neighboring grid, and the number of neighboring grids with the same direction is counted. The number of consistent neighboring grids is divided by the total number of neighboring grids to obtain the direction consistency ratio. At the same time, the similarity of the gradient magnitude between the target grid and the neighboring grids is calculated. The direction consistency ratio and the gradient magnitude similarity are weighted and summed according to preset weights. The result is the precipitation intensity gradient consistency factor. The larger the factor value, the more consistent the precipitation intensity change trend between the target grid and the neighboring grids.

[0132] The preset gradient consistency adjustment coefficient is determined through calibration using historical precipitation fusion data. During calibration, a large amount of multi-source precipitation data under different precipitation scenarios is collected, multiple candidate adjustment coefficient values ​​are selected, and the corresponding dry and wet fusion weight values ​​are calculated by substituting them into the formula. Then, these weight values ​​are used to fuse precipitation data, and the fusion result is compared with the measured precipitation data to calculate the fusion error corresponding to each candidate coefficient. The candidate coefficient with the smallest fusion error is selected as the preset gradient consistency adjustment coefficient. This coefficient is used to adjust the influence of the precipitation intensity gradient consistency factor on the fusion weight value.

[0133] The wet-dry fusion weight value is a core indicator that quantifies the contribution of precipitation data at a spatial grid location to the wet-dry condition-related precipitation fusion process. Its value directly determines the influence weight of the data at that location on the final fusion result.

[0134] The spatial continuity index of precipitation, as a basic factor, directly reflects the spatial coordination between the precipitation data of the target grid and the surrounding area. The larger the value of the index, the more reasonable the spatial distribution of the precipitation data at that location and the higher the reliability, thus providing a basis for the spatial rationality of the fusion weight value.

[0135] The natural exponential function part performs a nonlinear transformation on the precipitation intensity gradient consistency factor through specific calculation logic. The ratio of the gradient consistency factor to the adjustment coefficient is used as input, and the attenuation term is calculated through the natural exponential function. The attenuation term is subtracted from the fixed benchmark value to form the gradient influence adjustment factor. The larger the gradient consistency factor, the closer the adjustment factor is to the fixed benchmark value, and the stronger its effect on improving the fusion weight value. At the same time, the adjustment coefficient can precisely control the sensitivity of this improvement effect.

[0136] Multiplying the precipitation spatial continuity index by the gradient influence adjustment factor yields a wet-dry fusion weight value that synergistically considers the spatial continuity and intensity gradient consistency of precipitation data. A larger value indicates better spatial coordination and a more reasonable intensity change trend in the precipitation data at that grid location, and should be assigned a higher weight during the fusion process. Conversely, a smaller value assigns a lower weight. By integrating multi-source precipitation data through this weight value, spatially anomalies or contradictory trends can be effectively filtered out, significantly improving the accuracy of the fused precipitation data and providing reliable data support for scenarios such as wet-dry condition assessment and water resource analysis.

[0137] The beneficial effects are that by combining the spatial continuity characteristics of precipitation with the reliability characteristics of ground observations, the background precipitation area and high-value precipitation area of ​​the target region can be accurately divided. Through scientific correlation assessment and confidence quantification, dry and wet fusion weight values ​​and heavy precipitation fusion weight values ​​that fit the actual distribution of precipitation are obtained respectively. After seamless spatial fusion and smoothing at the boundary, the problem of weight abrupt change is avoided, ensuring the continuity and rationality of the dual weight allocation benchmark, fully adapting to the weight requirements of different precipitation scenarios, and providing a reliable and realistic weight basis for the accurate fusion of precipitation data in the target region.

[0138] S5. Correct the dry and wet boundaries of the basic gridded satellite precipitation field, and combine the corrected satellite precipitation field with the difference analysis of the basic gridded ground precipitation field to obtain the residual field of the target area.

[0139] In this embodiment of the invention, the step of correcting the dry and wet boundaries of the basic gridded satellite precipitation field and combining the corrected satellite precipitation field with the difference analysis of the basic gridded ground precipitation field to obtain the residual field of the target area includes identifying the dry and wet boundary features characterizing precipitation distribution based on the precipitation intensity of the grids in the basic gridded satellite precipitation field.

[0140] Based on the dry and wet boundary characteristics, the precipitation intensity of the corresponding grid in the basic gridded satellite precipitation field is adaptively adjusted to obtain the corrected satellite precipitation field of the target area;

[0141] The precipitation intensity values ​​of the grids in the corrected satellite precipitation field are compared with the precipitation intensity values ​​of the corresponding grids in the basic gridded ground precipitation field grid by grid-by-grid to obtain the preliminary residual field of the target area.

[0142] The preliminary residual field is spatially smoothed to eliminate local outliers, thus obtaining the residual field of the target region.

[0143] The step of subtracting the precipitation intensity values ​​of the grids in the corrected satellite precipitation field from the corresponding grids in the basic gridded ground precipitation field to obtain the preliminary residual field of the target area includes:

[0144] Spatial grid registration is performed between the corrected satellite precipitation field and the basic gridded ground precipitation field to obtain grid data pairs for the target area;

[0145] Based on the grid data pair, the precipitation intensity values ​​in the corrected satellite precipitation field and the basic gridded ground precipitation field are extracted synchronously according to the spatial grid order to obtain the corresponding precipitation intensity value set of the grid data pair;

[0146] Based on the corresponding precipitation intensity value set, the intensity values ​​of the basic gridded surface precipitation field are compared by difference to obtain the preliminary residual field of the target area.

[0147] Precipitation intensity information for each grid is extracted from the basic gridded satellite precipitation field. A computer program traverses all grids within the target area grid by grid, while simultaneously retrieving precipitation intensity data from neighboring grids around each grid. The precipitation intensity of the current grid is compared with that of its neighboring grids to determine if there are any abrupt changes in precipitation status. All grids with abrupt changes in precipitation status are marked. The continuously marked grids are then connected according to their spatial distribution to form closed or semi-closed boundary lines. These boundary lines clearly delineate areas within the target area that are covered by precipitation and areas that are not covered by precipitation, thus representing the dry and wet boundary characteristics of precipitation distribution. During the marking and connection process, the continuity of the boundary lines is verified by the program to ensure that there are no broken or overlapping boundary segments.

[0148] Based on the identified dry-wet boundary characteristics, a computer program determines the spatial position of each grid relative to the dry-wet boundary, clarifying whether the grid is located in a precipitation area inside the dry-wet boundary, a non-precipitation area outside, or directly on the dry-wet boundary. For grids on the dry-wet boundary, if the precipitation intensity differs significantly from the average precipitation intensity of the grids in the inner precipitation area and exceeds the reasonable range of natural precipitation transition, the program adjusts the precipitation intensity of the grid to an intermediate transition value between the average precipitation intensity of the grids in the inner precipitation area and the precipitation intensity of the grids in the outer non-precipitation area. For grids close to the dry-wet boundary with abnormally high or low precipitation intensity, the program moderately adjusts the precipitation intensity of the abnormal grids by referring to their distance from the dry-wet boundary and the precipitation intensity distribution pattern of the surrounding grids, so that the precipitation intensity of the grids on both sides of the dry-wet boundary presents a natural gradual trend, avoiding abrupt changes. The complete gridded precipitation data formed after all grids are adjusted is the corrected satellite precipitation field of the target area. During the adjustment process, the program records the adjustment basis and the precipitation intensity values ​​before and after the adjustment for each grid to ensure data traceability.

[0149] The computer program loads raster data from the corrected satellite precipitation field and the base gridded ground precipitation field. Using the program's built-in spatial matching function, it matches the grids in the two precipitation fields one-to-one based on their unique identifiers and spatial coordinates. This ensures that each grid has a matching grid in both precipitation fields that completely overlaps in spatial location. After matching, the program extracts the precipitation intensity values ​​from the corrected satellite precipitation field and the corresponding matching grids in the base gridded ground precipitation field, grid by grid. The program subtracts the precipitation intensity values ​​from the corrected satellite precipitation field (minuend) and the precipitation intensity values ​​from the base gridded ground precipitation field (subtrahend), obtaining the difference for each grid. All the difference results are organized into a raster dataset based on their original spatial locations. This raster dataset is the preliminary residual field for the target area. During the calculation, the program automatically verifies the consistency of the numerical data types for each grid to avoid calculation errors caused by differences in data types.

[0150] A computer program performs spatial smoothing on each grid in the initial residual field. The program sets a fixed range of neighboring grids for each grid, extracts the initial residual values ​​of all grids within that range, arranges these residual values ​​in ascending order, removes extreme values ​​at both ends, and then sums the remaining residual values. The summation result is divided by the number of remaining residual values ​​to obtain the average level of the residual values ​​within that range. The program uses this average level as the smoothed residual value for the current grid, replacing the original initial residual value. This process is repeated for all grids, ensuring that the neighboring grids are selected consistently and that the criteria for removing extreme values ​​are uniform. This method eliminates isolated high or low values ​​in the initial residual field caused by local data anomalies, making the spatial distribution of the residual field more continuous and reasonable. The resulting raster dataset is the residual field for the target region.

[0151] The algorithm reads the raster data metadata of the corrected satellite precipitation field and the basic gridded ground precipitation field. A computer program analyzes the core parameters of the two types of data, such as spatial projection type, grid resolution, latitude and longitude boundary range, and number of rows and columns. First, the spatial projection systems of the two types of data are uniformly converted to the standard projection preset for the target area. During the conversion process, coordinate mapping rules are used to accurately map the corner coordinates of each grid to the standard projection coordinate system, ensuring that the spatial position of the grid after projection conversion is without deviation. Then, using the grid boundary of the target grid frame as a reference, the grid resolution of the two types of data is adjusted for consistency. If the grid resolution of a certain type of data is inconsistent with the target frame, adjacent grid data are integrated or split to ensure that the grid size and row and column arrangement of the two types of data are completely matched. Finally, according to the row and column index order of the grids, a correspondence is established between each grid in the corrected satellite precipitation field and each grid in the basic gridded ground precipitation field whose spatial position completely overlaps. Each correspondence includes the unique identifier, latitude and longitude boundary, and row and column index information of the two grids. The set of all established correspondences constitutes the grid data pairs for the target area.

[0152] Based on the established grid data pairs, a fixed spatial grid traversal order is set by a computer program. This order starts from the grid in the upper left corner of the target area and traverses each grid data pair sequentially from left to right and from top to bottom. During the traversal, the program uses the unique identifier in the grid data pair to synchronously locate the corresponding grid in the corrected satellite precipitation field and the corresponding grid in the basic gridded ground precipitation field. It extracts the precipitation intensity values ​​recorded in the corrected satellite precipitation field grid one by one, and at the same time extracts the precipitation intensity values ​​recorded in the corresponding grid in the basic gridded ground precipitation field. The two precipitation intensity values ​​of the same grid data pair are bound with the unique identifier and row and column index of the grid data pair to form a structured data entry containing "grid identifier - satellite precipitation intensity value - ground precipitation intensity value". All the structured data entries generated after the traversal are completed are arranged in the traversal order to constitute the corresponding precipitation intensity value set of the grid data pair.

[0153] Each structured data entry in the corresponding precipitation intensity value set is retrieved, and a computer program extracts the corrected satellite precipitation field precipitation intensity value and the basic gridded ground precipitation field precipitation intensity value bound to each entry. The corrected satellite precipitation field precipitation intensity value is used as the minuend, and the basic gridded ground precipitation field precipitation intensity value is used as the subtrahend, and the numerical subtraction is performed to obtain the difference result corresponding to the grid data pair. During the calculation, the program automatically verifies the data type consistency of the two values ​​to ensure that there are no logical errors in the calculation. After the difference calculation of each data entry is completed, the difference result is associated and bound with the corresponding grid identifier and row and column index. After the difference calculation of all data entries is completed, according to the spatial distribution order of the grid data pairs, all difference results are filled into a raster structure consistent with the grid framework of the target area according to their corresponding grid row and column positions, forming a raster dataset that completely covers the target area and each grid corresponds to a unique difference result. This raster dataset is the preliminary residual field of the target area.

[0154] The beneficial effects are as follows: by accurately identifying the dry and wet boundary characteristics of the basic gridded satellite precipitation field, the precipitation intensity of the boundary and adjacent grids is adaptively adjusted, effectively avoiding the problem of abrupt precipitation changes, improving the rationality and authenticity of the corrected satellite precipitation field. Based on the grid-by-grid difference operation of precise grid matching, it is ensured that the preliminary residual field can accurately reflect the differences between the two types of precipitation fields. Subsequent spatial smoothing processing eliminates local outliers, making the spatial distribution of the residual field more continuous and stable. The final residual field is of reliable quality, providing accurate data support for the further fusion and optimization of precipitation data in the target area.

[0155] By accurately registering the corrected satellite precipitation field with the basic gridded ground precipitation field, spatial discrepancies between the two types of data are eliminated, ensuring accurate grid correspondence. Precipitation intensity values ​​of grid data pairs are extracted sequentially and synchronously to form a standardized set of intensity values ​​with clear correspondences. Based on this set, precise difference comparison is performed to ensure that the difference calculation logic of each grid is rigorous and the results are accurate. The resulting preliminary residual field can truly reflect the intensity differences between the two types of precipitation fields, laying a reliable foundation for subsequent optimization of the residual field.

[0156] S6. Based on the dual weight allocation benchmark and the residual field, perform pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain the satellite-ground fused precipitation field of the target area.

[0157] In this embodiment of the invention, the step of performing pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field based on the dual-weight allocation benchmark and the residual field to obtain the satellite-ground fused precipitation field of the target area includes:

[0158] Based on the fusion rule scheme in the dual weight allocation benchmark, the spatial grid of the target region is subjected to dominant rule filtering to obtain the adaptation fusion rule of the target region;

[0159] Based on the aforementioned adaptation and fusion rules, the basic gridded satellite precipitation field and the basic gridded ground precipitation field are weighted and fused to obtain the preliminary pixel-by-pixel fusion result of the target area;

[0160] The preliminary pixel-by-pixel fusion result is superimposed with the residual field after deviation compensation to obtain the deviation-corrected fused precipitation field of the target area;

[0161] The regional consistency of the deviation-corrected fused precipitation field is verified to obtain the intermediate fused precipitation field of the target region;

[0162] Based on the aforementioned dual-weight allocation benchmark, the intermediate fused precipitation field is subjected to adaptive spatial smoothing filtering in the dry-wet transition region and the boundary region between strong and weak precipitation to obtain the star-ground fused precipitation field of the target region.

[0163] The fusion rule scheme contained in the dual-weight allocation benchmark is read. This scheme clarifies two core fusion logics corresponding to wet-dry fusion weight dominance and heavy precipitation fusion weight dominance. The wet-dry fusion weight dominance rule focuses on adjusting the fusion weight ratio based on the spatial continuity of precipitation, while the heavy precipitation fusion weight dominance rule focuses on adjusting the fusion weight ratio based on the reliability of ground observation data. The wet-dry fusion weight value and the heavy precipitation fusion weight value in the dual-weight allocation benchmark are extracted grid by grid through a computer program. The two weight values ​​of the same grid are compared. If several wet fusion weight values ​​are greater than the heavy precipitation fusion weight values, the grid is determined to be suitable for the wet-dry fusion weight dominance fusion rule. If the heavy precipitation fusion weight value is greater than the wet-dry fusion weight value, the grid is determined to be suitable for the heavy precipitation fusion weight dominance fusion rule. The suitable rules of all grids are associated and bound with the grid's unique identifier and spatial coordinates to form a rule set covering all grids in the target area. This rule set is the suitable fusion rule for the target area. During the rule determination process, the consistency of the weight value comparison logic is verified by the program to ensure that no grid has ambiguity in rule determination.

[0164] Based on the adaptation and fusion rules, a computer program matches the corresponding grid precipitation intensity values ​​of the base gridded satellite precipitation field and the base gridded ground precipitation field grid by grid. For grids that adapt to the wet-dry fusion weight-dominated rule, the wet-dry fusion weight value of the grid is extracted from the dual weight allocation benchmark. This weight value is used as the fusion weight of the base gridded satellite precipitation field, and the complementary value of the weight value is used as the fusion weight of the base gridded ground precipitation field. The grid precipitation intensity values ​​of the two precipitation fields are multiplied by their corresponding weights and then summed to obtain the fused precipitation value of the grid. For grids that adapt to the heavy precipitation fusion weight-dominated rule, the heavy precipitation fusion weight value of the grid is extracted. The same weight allocation logic is used to complete the weighted summation calculation of the grid precipitation intensity values ​​of the two precipitation fields. After completing all fusion calculations grid by grid, the fused precipitation value of each grid is organized into a raster dataset according to its spatial coordinates. This raster dataset is the preliminary pixel-by-pixel fusion result of the target area.

[0165] The computer program precisely matches the preliminary pixel-by-pixel fusion results with the residual field using spatial grids, ensuring that the grid framework, spatial projection, and resolution of the two datasets are completely consistent. Each grid corresponds to a unique preliminary fused precipitation value and residual value. The fused precipitation value and residual value of the residual field are extracted from the preliminary pixel-by-pixel fusion results grid by grid. The residual value and the fused precipitation value are then superimposed to calculate the bias-corrected precipitation value for that grid. The purpose of the superposition calculation is to compensate for any systematic biases that may exist in the preliminary fusion results through the residual field. Since the residual value accurately reflects the difference between the corrected satellite precipitation field and the basic gridded ground precipitation field, the superposition can further improve the accuracy of the fused data. After the superposition calculation is completed grid by grid, a raster dataset containing the bias-corrected precipitation values ​​of all grids is formed. This raster dataset is the bias-corrected fused precipitation field of the target area.

[0166] Based on the topographic features and climate zoning patterns of the target area, a computer program divides the target area into multiple continuous regional sub-units. Each sub-unit encompasses grid areas with similar topography and climate backgrounds. The bias-corrected fused precipitation field data within each regional sub-unit is analyzed one by one, calculating the average precipitation, precipitation trend, and precipitation distribution pattern for all grids within the sub-unit. Simultaneously, historical precipitation statistics and overall trends from ground observation data for the corresponding sub-unit are retrieved. The average, trend, and pattern of the bias-corrected fused precipitation data are compared one by one with historical characteristics and ground observation trends to determine if there are significant differences. If the precipitation values ​​of some grids within a sub-unit deviate from the overall trend or historical characteristics, the precipitation values ​​of the deviating grids are appropriately adjusted by referring to the precipitation levels and historical data of most grids within the sub-unit to ensure the consistency of precipitation data within the sub-unit. After all sub-units have been verified and adjusted, they are integrated to form a complete raster dataset covering the target area. This raster dataset is the intermediate fused precipitation field for the target area.

[0167] Based on the spatial distribution characteristics of weights according to a dual-weight allocation benchmark, a computer program identifies dry-wet transition zones and areas where heavy and light precipitation meet within the target area. The identification criteria are regions where weight values ​​show a continuous, gradual change and the weight difference between adjacent grids is at a moderate level. Specifically, the dry-wet transition zone corresponds to the gradual range of the combined dry-wet weight value, and the heavy-light precipitation boundary zone corresponds to the gradual range of the combined heavy precipitation weight value. For each grid within these two boundary zones, the program sets a fixed-range adjacent grid filtering window, extracts the intermediate combined precipitation value of all grids within the window, and removes extreme values ​​that significantly deviate from the majority. For outliers, the remaining effective precipitation values ​​are summed and the sum is divided by the number of effective precipitation values ​​to obtain the average precipitation value. This average precipitation value is used as the smoothed precipitation value for the current grid, replacing the original intermediate fused precipitation value. For grids in non-boundary areas, their original intermediate fused precipitation values ​​are kept unchanged. During the smoothing process, it is ensured that the precipitation values ​​of grids in boundary areas show a natural gradual trend to avoid abrupt changes. After smoothing is completed grid by grid, a spatially continuous raster dataset with good data consistency is formed. This raster dataset is the star-ground fused precipitation field of the target area.

[0168] The beneficial effects are as follows: based on the dual-weight allocation benchmark, the fusion rules are accurately selected and adapted to ensure that the fusion logic of each grid is consistent with its own precipitation characteristics; the weighted fusion combined with residual field bias compensation effectively corrects the systematic bias of the initial fusion results and improves data accuracy; regional consistency verification ensures that the data in each sub-unit is consistent with historical characteristics and observation trends, avoiding local anomalies; adaptive smoothing filtering for the dry-wet and strong-wet precipitation boundary areas eliminates data mutation problems and enhances spatial continuity; the final star-ground fused precipitation field data is reliable, continuously distributed, and has good consistency, providing high-quality data support for precipitation analysis and application in the target area.

[0169] like Figure 2 The diagram shown is a functional block diagram of a ground-to-space precipitation fusion system for dry and wet distribution and heavy precipitation areas provided in an embodiment of the present invention.

[0170] The satellite-to-ground precipitation fusion system 100 for dry / wet distribution and heavy precipitation areas described in this invention can be installed in an electronic device. Depending on the functions implemented, the satellite-to-ground precipitation fusion system 100 may include a data acquisition module 101, a spatiotemporal gridding module 102, a multidimensional feature extraction module 103, a dual-weight construction module 104, a residual analysis module 105, and a weighted fusion module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0171] In this embodiment, the functions of each module / unit are as follows:

[0172] The data acquisition module 101 is used to acquire satellite remote sensing inversion precipitation data and ground station observation precipitation data of the target area;

[0173] The spatiotemporal gridding module 102 is used to perform spatiotemporal gridding on the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and the basic gridded ground precipitation field of the target area.

[0174] The multidimensional feature extraction module 103 is used to extract multidimensional features from the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain a comprehensive precipitation feature dataset of the target area.

[0175] The dual-weight construction module 104 is used to construct a dual-weight allocation benchmark for the dry-wet fusion weight and the heavy precipitation fusion weight in the target area based on the comprehensive precipitation feature dataset.

[0176] The residual analysis module 105 is used to perform dry and wet boundary correction on the basic gridded satellite precipitation field, and combine the corrected satellite precipitation field to perform difference analysis on the basic gridded ground precipitation field to obtain the residual field of the target area.

[0177] The weighted fusion module 106 is used to perform pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field based on the dual weight allocation benchmark and the residual field, so as to obtain the satellite-ground fused precipitation field of the target area.

[0178] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0179] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0180] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0181] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0182] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0183] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for integrating dry and wet distribution and heavy precipitation areas, characterized in that, The method includes: S1. Acquire satellite remote sensing inversion precipitation data and ground station observation precipitation data for the target area; S2. Spatiotemporally grid the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and the basic gridded ground precipitation field of the target area; S3. Perform multi-dimensional feature extraction on the basic gridded satellite precipitation field and the basic gridded surface precipitation field to obtain a comprehensive precipitation feature dataset for the target area, including: By statistically analyzing the precipitation distribution in the gridded satellite precipitation field and the gridded ground precipitation field, the statistical feature set of satellite precipitation and the statistical feature set of ground precipitation in the target area are obtained. Spatial structure analysis is performed on the basic gridded satellite precipitation field to extract texture features that characterize the spatial continuity and gradient changes of precipitation, thereby obtaining the satellite precipitation spatial feature set of the basic gridded satellite precipitation field. Time series analysis is performed on the basic gridded surface precipitation field to extract the time variation features that characterize the evolution of precipitation processes, thereby obtaining the surface precipitation time series feature set of the basic gridded surface precipitation field; The satellite precipitation statistical feature set, the ground precipitation statistical feature set, the satellite precipitation spatial feature set, and the ground precipitation temporal feature set are fused and integrated to obtain the comprehensive precipitation feature set of the target area; S4. Based on the comprehensive precipitation feature dataset, construct a dual weight allocation benchmark for the target region, comprising the combined weights of dry and wet precipitation and the combined weights of heavy precipitation, including: Based on the comprehensive precipitation feature data that centrally characterizes the spatial continuity of precipitation, the dry and wet distribution statistics of the target area are performed to obtain the background precipitation area of ​​the target area; Based on the characteristics that centrally represent the reliability of ground observations in the comprehensive precipitation feature dataset, the precipitation intensity in the target area is calibrated at multiple scales to obtain the high-precipitation area in the target area. The spatial continuity index of precipitation in the background precipitation area is correlated with the similarity of precipitation patterns in the surrounding grid to obtain the dry-wet fusion weight value of the target area. The observation density and data quality level of adjacent ground stations in the high-precipitation area are quantified with confidence to obtain the heavy precipitation fusion weight value of the target area. The calculation formula of the heavy precipitation fusion weight value is as follows: ; In the formula, Spatial grid position in the target region The aforementioned heavy precipitation fusion weight value, To be at the spatial grid position Ground observation quality influencing factors at the location It is the hyperbolic tangent function. The preset intensity difference sensitivity coefficient, To be at the spatial grid position Standardized precipitation intensity differences at various locations; The wet and dry fusion weight value and the heavy precipitation fusion weight value are spatially seamlessly fused, and the grid at the boundary of the fused weight region is smoothed to obtain the dual weight allocation benchmark of the target region. S5. Correct the dry and wet boundaries of the basic gridded satellite precipitation field, and combine the corrected satellite precipitation field with the difference analysis of the basic gridded ground precipitation field to obtain the residual field of the target area. S6. Based on the dual weight allocation benchmark and the residual field, perform pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field to obtain the satellite-ground fused precipitation field of the target area.

2. The method for integrating dry and wet distribution and heavy precipitation areas as described in claim 1, characterized in that, The acquisition of satellite remote sensing inversion precipitation data and ground station observation precipitation data for the target area includes: From the remote sensing data of on-orbit meteorological satellites, infrared, microwave, and combined precipitation data covering the target area are selected as the multi-source satellite remote sensing precipitation data for the target area; In the corresponding meteorological operational data center of the target area, the cumulative precipitation data recorded by effective surface meteorological observation stations within the same time period is extracted as the original surface station observation precipitation data of the target area. Cross-validation is performed on the multi-source satellite remote sensing inversion precipitation data to eliminate systematic biases between different satellite data sources, thereby obtaining satellite remote sensing inversion precipitation data for the target area; The original ground station precipitation data is subjected to quality control, invalid records are removed and obvious errors are corrected to obtain the ground station precipitation data for the target area.

3. The method for integrating dry and wet distribution and heavy precipitation areas as described in claim 1, characterized in that, The process of spatiotemporally gridding the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and basic gridded ground precipitation field of the target area includes: A standardized specification for the unified spatiotemporal grid of the target region is defined to obtain the target grid framework of the target region. Based on the target grid framework, the satellite remote sensing inversion precipitation data is temporally resampled to obtain the satellite time-aligned data sequence of the target area; Spatial reprojection and cropping are performed on the satellite time-aligned data sequence to obtain spatially aligned satellite data for the target region. The spatially aligned satellite data is gridded to obtain the basic gridded satellite precipitation field of the target area; Based on the target grid framework, spatial interpolation is performed on the precipitation data observed at the ground stations, and the estimated precipitation values ​​of the spatial grids in the target grid framework are statistically analyzed to obtain the basic gridded surface precipitation field of the target area.

4. The method for integrating dry and wet distribution and heavy precipitation areas as described in claim 1, characterized in that, The formula for calculating the wet-dry fusion weight value is as follows: ； In the formula, Spatial grid position in the target region The dry-wet fusion weight value at the location, The row and column index coordinates of the spatial grid within the target area. The spatial continuity index of precipitation, As the precipitation intensity gradient consistency factor, The preset gradient consistency adjustment coefficient, It is a natural exponential function.

5. The method for integrating dry and wet distribution and heavy precipitation areas as described in claim 1, characterized in that, The process of correcting the wet and dry boundaries of the basic gridded satellite precipitation field, and then performing difference analysis on the basic gridded ground precipitation field in conjunction with the corrected satellite precipitation field to obtain the residual field of the target region, includes: Based on the precipitation intensity of the grid in the basic gridded satellite precipitation field, the dry and wet boundary features characterizing the precipitation distribution are identified; Based on the dry and wet boundary characteristics, the precipitation intensity of the corresponding grid in the basic gridded satellite precipitation field is adaptively adjusted to obtain the corrected satellite precipitation field of the target area; The precipitation intensity values ​​of the grids in the corrected satellite precipitation field are compared with the precipitation intensity values ​​of the corresponding grids in the basic gridded ground precipitation field grid by grid-by-grid to obtain the preliminary residual field of the target area. The preliminary residual field is spatially smoothed to eliminate local outliers, thus obtaining the residual field of the target region.

6. The method for integrating dry and wet distribution and heavy precipitation areas as described in claim 5, characterized in that, The step of subtracting the precipitation intensity values ​​of the grids in the corrected satellite precipitation field from the corresponding grids in the basic gridded ground precipitation field to obtain the preliminary residual field of the target area includes: Spatial grid registration is performed between the corrected satellite precipitation field and the basic gridded ground precipitation field to obtain grid data pairs for the target area; Based on the grid data pair, the precipitation intensity values ​​in the corrected satellite precipitation field and the basic gridded ground precipitation field are extracted synchronously according to the spatial grid order to obtain the corresponding precipitation intensity value set of the grid data pair; Based on the corresponding precipitation intensity value set, the intensity values ​​of the basic gridded surface precipitation field are compared by difference to obtain the preliminary residual field of the target area.

7. The method for integrating dry and wet distribution and heavy precipitation areas as described in claim 1, characterized in that, The step of performing pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field based on the dual-weight allocation benchmark and the residual field to obtain the satellite-ground fused precipitation field of the target area includes: Based on the fusion rule scheme in the dual weight allocation benchmark, the spatial grid of the target region is subjected to dominant rule filtering to obtain the adaptation fusion rule of the target region; Based on the aforementioned adaptation and fusion rules, the basic gridded satellite precipitation field and the basic gridded ground precipitation field are weighted and fused to obtain the preliminary pixel-by-pixel fusion result of the target area; The preliminary pixel-by-pixel fusion result is superimposed with the residual field after deviation compensation to obtain the deviation-corrected fused precipitation field of the target area; The regional consistency of the deviation-corrected fused precipitation field is verified to obtain the intermediate fused precipitation field of the target region; Based on the aforementioned dual-weight allocation benchmark, the intermediate fused precipitation field is subjected to adaptive spatial smoothing filtering in the dry-wet transition region and the boundary region between strong and weak precipitation to obtain the star-ground fused precipitation field of the target region.

8. A ground-based precipitation fusion system for wet and dry distribution and heavy precipitation areas, characterized in that, For implementing the method for fusing dry and wet distribution and heavy precipitation areas according to claim 1, the system comprises: The data acquisition module is used to acquire satellite remote sensing inversion precipitation data and ground station observation precipitation data for the target area; The spatiotemporal gridding module is used to perform spatiotemporal gridding on the satellite remote sensing inversion precipitation data and the ground station observation precipitation data to obtain the basic gridded satellite precipitation field and the basic gridded ground precipitation field of the target area. A multidimensional feature extraction module is used to extract multidimensional features from the basic gridded satellite precipitation field and the basic gridded surface precipitation field to obtain a comprehensive precipitation feature dataset of the target area, including: By statistically analyzing the precipitation distribution in the gridded satellite precipitation field and the gridded ground precipitation field, the statistical feature set of satellite precipitation and the statistical feature set of ground precipitation in the target area are obtained. Spatial structure analysis is performed on the basic gridded satellite precipitation field to extract texture features that characterize the spatial continuity and gradient changes of precipitation, thereby obtaining the satellite precipitation spatial feature set of the basic gridded satellite precipitation field. Time series analysis is performed on the basic gridded surface precipitation field to extract the time variation features that characterize the evolution of precipitation processes, thereby obtaining the surface precipitation time series feature set of the basic gridded surface precipitation field; The satellite precipitation statistical feature set, the ground precipitation statistical feature set, the satellite precipitation spatial feature set, and the ground precipitation temporal feature set are fused and integrated to obtain the comprehensive precipitation feature set of the target area; A dual-weighting construction module is used to construct a dual-weighting allocation benchmark for the target region based on the comprehensive precipitation feature dataset, comprising: Based on the comprehensive precipitation feature data that centrally characterizes the spatial continuity of precipitation, the dry and wet distribution statistics of the target area are performed to obtain the background precipitation area of ​​the target area; Based on the characteristics that centrally represent the reliability of ground observations in the comprehensive precipitation feature dataset, the precipitation intensity in the target area is calibrated at multiple scales to obtain the high-precipitation area in the target area. The spatial continuity index of precipitation in the background precipitation area is correlated with the similarity of precipitation patterns in the surrounding grid to obtain the dry-wet fusion weight value of the target area. The observation density and data quality level of adjacent ground stations in the high-precipitation area are quantified with confidence to obtain the heavy precipitation fusion weight value of the target area. The calculation formula of the heavy precipitation fusion weight value is as follows: ； In the formula, Spatial grid position in the target region The aforementioned heavy precipitation fusion weight value, To be at the spatial grid position Ground observation quality influencing factors at the location It is the hyperbolic tangent function. The preset intensity difference sensitivity coefficient, To be at the spatial grid position Standardized precipitation intensity differences at various locations; The wet and dry fusion weight value and the heavy precipitation fusion weight value are spatially seamlessly fused, and the grid at the boundary of the fused weight region is smoothed to obtain the dual weight allocation benchmark of the target region. The residual analysis module is used to correct the wet and dry boundaries of the basic gridded satellite precipitation field, and combine the corrected satellite precipitation field to perform difference analysis on the basic gridded ground precipitation field to obtain the residual field of the target area. The weighted fusion module is used to perform pixel-by-pixel weighted fusion of the basic gridded satellite precipitation field and the basic gridded ground precipitation field based on the dual weight allocation benchmark and the residual field, so as to obtain the satellite-ground fused precipitation field of the target area.

Images