A geographic information surveying and mapping data intelligent analysis method and system

Abstract

This invention proposes an intelligent analysis method and system for geographic information mapping data, belonging to the field of mapping data analysis technology. The method includes: determining the area to be analyzed and acquiring geographic information mapping data; gridding the area to be analyzed into multiple spatial locations, constructing a time series, and extracting multi-dimensional trend feature vectors within different time windows based on the time series; for each spatial location, selecting final candidate neighborhoods; selecting effective neighborhoods based on the multi-dimensional trend feature vectors; interpolating the time series of the target spatial locations based on the effective neighborhoods to obtain a complete time series for each spatial location; performing spatiotemporal change pattern analysis on the complete time series, identifying spatiotemporal evolution patterns, and generating a spatiotemporal change analysis report. This invention can accurately identify and intelligently analyze the spatiotemporal evolution patterns of geographic mapping data, providing interpretable analysis results with geographical significance.

Description

Technical Field

[0001] This invention relates to the field of surveying and mapping data analysis technology, and in particular to an intelligent analysis method and system for geographic information surveying and mapping data. Background Technology

[0002] With the rapid development of remote sensing technology, Global Positioning System (GPS), Geographic Information System (GIS), and surveying and mapping technology, geographic information mapping data has seen significant improvements in spatial resolution, temporal coverage, and data accuracy. Modern geographic mapping data is characterized by multi-source heterogeneity, massive real-time volume, and spatiotemporal coupling, encompassing various types such as satellite remote sensing data, aerial photography data, ground-based measured data, and UAV aerial survey data. This data provides a rich foundation for a deeper understanding of the spatiotemporal evolution of geographic phenomena, monitoring of surface change processes, and supporting urban planning and environmental protection. However, due to the influence of various factors such as satellite revisit cycles, weather conditions, equipment failures, and task scheduling, geographic mapping data often exhibits severe irregularities and incompleteness in the temporal dimension, and also shows uneven density characteristics in spatial distribution, posing significant challenges to subsequent data analysis and processing.

[0003] Existing geographic data analysis techniques mainly fall into two categories: spatial analysis methods and time series analysis methods. Spatial analysis methods focus on uncovering the distribution patterns and correlations of data in the spatial dimension, such as spatial interpolation, spatial clustering, and spatial regression. These methods can effectively handle the spatial distribution characteristics of data but lack consideration for temporal evolution. Time series analysis methods, on the other hand, focus on revealing the changing trends and periodic patterns of data in the time dimension, such as trend analysis, period detection, and anomaly identification. However, they typically assume that the data is equally spaced and continuous in time, an assumption that is difficult to satisfy in actual geographic mapping data. In recent years, spatiotemporal data analysis methods have gradually become a research hotspot, aiming to utilize the temporal and spatial information of data for comprehensive analysis. However, existing spatiotemporal analysis methods still have technical shortcomings in handling temporal irregularities, intelligently filtering spatial information, considering multi-scale features, and adaptive processing.

[0004] For example, the invention patent CN117171533A, "Real-time Acquisition and Processing Method and System for Geographic Surveying and Mapping Operation Data," focuses on the real-time acquisition, coordinate transformation, attribute field standardization, and basic quality verification of field surveying and mapping data by the operation terminal. Its core technology is an end-to-cloud collaborative transmission compression strategy and a rule-based field mapping table. This solution achieves real-time acquisition and preliminary processing of surveying and mapping data through modules such as operation data aggregation, data preprocessing, and overlapping data filtering, offering advantages in data acquisition efficiency and transmission optimization. However, this solution primarily addresses the technical problems in the data acquisition stage and does not provide effective solutions for key technical issues such as spatiotemporal sparsity, trend conflicts, and intelligent enhancement faced after surveying and mapping data acquisition is completed. Summary of the Invention

[0005] In view of this, the present invention provides an intelligent analysis method for geographic information mapping data, which aims to solve the technical defects of existing geographic information mapping data analysis technology in terms of handling irregularities in time, intelligent selection of spatial information, and consideration of multi-scale features. By establishing an intelligent neighborhood selection mechanism based on trend consistency constraints and a multi-scale feature vector matching algorithm, the method can accurately identify and intelligently analyze the spatiotemporal evolution of geographic mapping data, and provide interpretable analysis results with geographical significance.

[0006] The technical solution of this invention is implemented as follows:

[0007] On the one hand, this invention provides an intelligent analysis method for geographic information mapping data, including:

[0008] S1. Determine the area to be analyzed, obtain geographic information mapping data within the area to be analyzed, including spatial coordinates, timestamps and attribute values, standardize the data points and generate data quality scores.

[0009] S2. The area to be analyzed is gridded into multiple spatial locations. The standardized data points are assigned to the corresponding spatial locations according to their spatial coordinates. For each spatial location, the corresponding data points are arranged according to timestamps and attribute values ​​to construct a time series. Based on the time series, multi-dimensional trend feature vectors are extracted in different time windows.

[0010] S3. For each spatial location, construct an initial candidate neighborhood based on spatial distance. Use data quality scoring to filter and eliminate unqualified locations from the initial candidate neighborhood. When the number of candidate neighbors is insufficient, use an adaptive expansion mechanism to reconstruct the neighborhood and obtain the final candidate neighborhood.

[0011] S4. Based on the multi-dimensional trend feature vector, calculate the trend feature matching degree between each candidate neighbor in the final candidate neighborhood and the target spatial location, determine the optimal similarity threshold, and filter to obtain the effective neighborhood.

[0012] S5. Interpolate the time series of the target spatial location based on the effective neighborhood to obtain the complete time series of each spatial location;

[0013] S6. Perform spatiotemporal change pattern analysis on the complete time series, identify spatiotemporal evolution laws, and generate a spatiotemporal change analysis report.

[0014] Preferably, step S1 includes:

[0015] S11. Standardize the format of geographic information mapping data and unify the coordinate system and time reference.

[0016] S12. Establish an R-tree spatial index structure to support efficient spatial neighborhood queries;

[0017] S13. Use the Tukey fence method to identify and handle abnormal data points;

[0018] S14. Calculate the quality score for each data point based on data completeness, temporal density, and spatial coverage;

[0019] S15. Associate and store the standardized data points with their corresponding data quality scores.

[0020] Preferably, step S2 includes:

[0021] S21. Determine the grid resolution based on the range and data density of the area to be analyzed, and divide the area to be analyzed into a regular grid to form multiple spatial locations;

[0022] S22. Establish the mapping relationship between spatial location and data points, and assign the standardized data points to the corresponding grid spatial locations according to their spatial coordinates.

[0023] S23. For each spatial location, collect all data points belonging to it, sort them by timestamp, and combine them with attribute numerical information to construct a time series for that location;

[0024] S24. Set three different time windows for each spatial location: short-term, medium-term, and long-term. Calculate the linear trend slope using the least squares method within each time window and extract the second-order curvature features using local weighted polynomial fitting. Combine the features from the three time windows to construct a six-dimensional multi-dimensional trend feature vector.

[0025] Preferably, the formula for calculating the slope of the linear trend is:

[0026]

[0027] Where i is the spatial location index, w kk = {1, 2, 3} is the time window. For spatial location i within time window w k The slope of the linear trend within the time window w, where n is the spatial location i within the time window w. k The number of data points within, t is the timestamp of the data point, V i (t) represents the attribute value of the data point corresponding to spatial location i at timestamp t. This indicates that spatial location i is within time window w. k The set of timestamps for all data points within the set.

[0028] Preferably, step S3 includes:

[0029] S31. Take each spatial location as the target spatial location in turn, and select other nearby spatial locations centered on the target spatial location according to Euclidean distance to construct the initial candidate neighborhood.

[0030] S32. For each candidate spatial location in the initial candidate neighborhood, calculate the comprehensive quality score of the spatial location based on the data quality score of its associated data point, and eliminate spatial locations with a comprehensive quality score lower than the quality threshold.

[0031] S33. When the number of candidate spatial locations after screening is less than the preset minimum value, an adaptive expansion mechanism is activated to gradually expand the search radius and reconstruct the final candidate neighborhood that meets the quantity requirements.

[0032] Preferably, step S4 includes:

[0033] S41. Extract the common time period between each candidate spatial location and the target spatial location in the final candidate neighborhood;

[0034] S42. Recalculate the multi-dimensional trend feature vectors of both based on the common time period;

[0035] S43. Calculate the trend matching degree based on the linear trend slope, and calculate the comprehensive similarity by combining the recalculated multi-dimensional trend feature vector.

[0036] S44. Determine the optimal similarity threshold and neighborhood number parameters;

[0037] S45. Filter candidate spatial locations based on the optimal similarity threshold, and select the top few locations in descending order of comprehensive similarity to form an effective neighborhood.

[0038] Preferably, the formula for calculating the trend matching degree M(i,j) is:

[0039]

[0040] Among them, w k k = {1, 2, 3} is the time window. and Spatial positions i and j are respectively located in time window w. k The slope of the linear trend within the interval, ∈ is a small positive number to prevent the denominator from being zero;

[0041] The formula for calculating the overall similarity S(i,j) is:

[0042] S(i,j)=α1·M(i,j)+α2·|cos(v′ i ,v′ j )|

[0043] Where, v′ i and v′ j These are the multi-dimensional trend feature vectors recalculated for spatial locations i and j based on a common time period, cos(v′) i ,v′ j ) is v′ i and v′ j The cosine similarity is given by α1 and α2, which are the weights of trend matching degree and cosine similarity, respectively, and α1+α2=1, α1,α2>0.

[0044] Preferably, step S5 includes:

[0045] S51. Calculate the spatial distance weight from each spatial location in the effective neighborhood to the target spatial location;

[0046] S52. Spatial interpolation calculation is performed on the time series of each spatial location within the effective neighborhood using the inverse distance weighting method.

[0047] S53. When original time series data exists for the target spatial location, a weighted fusion is performed based on the comprehensive quality score of the time series and the interpolation results.

[0048] S54. Use the results obtained by fusion or interpolation as the complete time series of that spatial location.

[0049] Preferably, step S6 includes:

[0050] S61. Perform time-domain decomposition on the complete time series for each spatial location to identify the trend component, periodic component, and random component of each spatial location.

[0051] S62. Identify spatial locations with similar change patterns and aggregate them into spatial regions through spatial autocorrelation analysis;

[0052] S63. Classify and identify the spatiotemporal evolution process at the regional level based on the identified spatial regions;

[0053] S64. Organize the analysis results according to spatial regions and generate a spatiotemporal change analysis report of the region to be analyzed, which includes hotspot areas of change, evolution trend prediction, and anomaly detection.

[0054] On the other hand, the present invention also provides an intelligent analysis system for geographic information mapping data, the system being used to implement the method described in any of the above claims, the system comprising:

[0055] The data preprocessing module is used to standardize the format of the input geographic information mapping data, handle abnormal data, and calculate quality scores.

[0056] The gridding module is used to divide the area to be analyzed into a regular grid and construct a time series.

[0057] The neighborhood construction module is used to build candidate neighborhoods for each target spatial location and perform quality screening, outputting the final candidate neighborhoods for each spatial location;

[0058] The feature matching module is used to calculate the trend feature matching degree of spatial locations within the candidate neighborhood and to filter the effective neighborhood.

[0059] The interpolation processing module is used to perform time series interpolation or fusion processing based on the effective neighborhood and output a complete time series.

[0060] The analysis report module is used to perform spatiotemporal change analysis and generate analysis reports.

[0061] The present invention has the following advantages over the prior art:

[0062] (1) This invention transforms the traditional distance-oriented spatial neighborhood selection into a trend-consistency-oriented selection by constructing an intelligent analysis framework based on trend consistency constraints, fundamentally solving the problem of analysis accuracy of geographic information mapping data under conditions of spatiotemporal sparsity and irregular distribution. This method can effectively handle the spatiotemporal evolution analysis of multi-source heterogeneous geographic data, improving the accuracy of spatiotemporal pattern recognition and the interpretability of analysis results while maintaining geographic authenticity;

[0063] (2) The dynamic neighborhood construction mechanism based on data quality scoring and trend feature matching established in this invention screens high-quality spatial neighbors through multi-dimensional quality evaluation indicators, effectively avoiding the negative impact of low-quality neighborhood data on the analysis results in traditional methods.

[0064] (3) The six-dimensional multi-scale trend feature vector constructed in this invention extracts linear trend and nonlinear acceleration features through three time windows: short-term, medium-term and long-term, which can comprehensively depict the evolution characteristics of geographical phenomena at different time scales.

[0065] (4) The confidence-based spatial augmentation method adopted in this invention comprehensively considers the quality of the original observation data and the reliability of the spatial interpolation results. Through a weighted fusion strategy, it supplements missing information while preserving the authenticity of the original data. This method effectively avoids the information loss and distortion problems that may occur with traditional interpolation methods, improving data integrity and analytical accuracy. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart of the method of the present invention;

[0068] Figure 2 This is a schematic diagram of the system of the present invention. Detailed Implementation

[0069] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0070] like Figure 1 As shown, this invention provides an intelligent analysis method for geographic information mapping data, comprising:

[0071] S1. Determine the area to be analyzed, obtain geographic information mapping data within the area to be analyzed, including spatial coordinates, timestamps and attribute values, standardize the data points and generate data quality scores.

[0072] S2. The area to be analyzed is gridded into multiple spatial locations. The standardized data points are assigned to the corresponding spatial locations according to their spatial coordinates. For each spatial location, the corresponding data points are arranged according to timestamps and attribute values ​​to construct a time series. Based on the time series, multi-dimensional trend feature vectors are extracted in different time windows.

[0073] S3. For each spatial location, construct an initial candidate neighborhood based on spatial distance. Use data quality scoring to filter and eliminate unqualified locations from the initial candidate neighborhood. When the number of candidate neighbors is insufficient, use an adaptive expansion mechanism to reconstruct the neighborhood and obtain the final candidate neighborhood.

[0074] S4. Based on the multi-dimensional trend feature vector, calculate the trend feature matching degree between each candidate neighbor in the final candidate neighborhood and the target spatial location, determine the optimal similarity threshold, and filter to obtain the effective neighborhood.

[0075] S5. Interpolate the time series of the target spatial location based on the effective neighborhood to obtain the complete time series of each spatial location;

[0076] S6. Perform spatiotemporal change pattern analysis on the complete time series, identify spatiotemporal evolution laws, and generate a spatiotemporal change analysis report.

[0077] This invention adopts the core technical approach of "trend consistency constraint + multi-scale feature matching," transforming the traditional distance-oriented spatial neighborhood selection into trend consistency-oriented selection. By constructing a six-dimensional multi-scale trend feature vector (covering the linear trend slope and second-order curvature features of three time windows: short-term, medium-term, and long-term), and combining it with the comprehensive similarity calculation of cosine similarity and sigmoid mapping, a dynamic screening threshold mechanism based on statistical variation characteristics is established. This enables adaptive spatial enhancement processing through intelligent neighborhood screening and confidence fusion, thereby solving the problem of accurate analysis of geographic information mapping data under conditions of spatiotemporal sparsity and irregular distribution. Finally, through time domain decomposition and spatial autocorrelation analysis, it achieves intelligent identification of spatiotemporal evolution patterns and interpretable analysis of geographical significance.

[0078] Specifically, in one embodiment of the present invention, step S1 includes:

[0079] S11. Standardize the format of geographic information mapping data and unify the coordinate system and time reference.

[0080] S12. Establish an R-tree spatial index structure to support efficient spatial neighborhood queries;

[0081] S13. Use the Tukey fence method to identify and handle abnormal data points;

[0082] S14. Calculate the quality score for each data point based on data completeness, temporal density, and spatial coverage;

[0083] S15. Associate and store the standardized data points with their corresponding data quality scores.

[0084] In this embodiment, the geographical region boundary for which spatiotemporal change analysis needs to be performed is first determined. This region can be an administrative boundary, a natural boundary, or any user-defined polygonal region.

[0085] Acquire geographic information mapping data for this area. This data includes spatial coordinate information (latitude and longitude coordinates or projected coordinates), timestamp information (recording the specific time of data collection), and attribute numerical information (such as quantitative or qualitative descriptive values ​​of geographic elements like vegetation index, surface temperature, and land use type). Data sources can be various types, including satellite remote sensing data, aerial photography data, ground measurement data, and UAV aerial survey data.

[0086] The acquired raw data undergoes format standardization, including unifying the coordinate system and time base to ensure all data points are based on the same spatial reference system and time standard. An R-tree spatial index structure is established to support efficient spatial neighborhood query operations. This index structure can quickly locate data points within a specified area and supports range queries and nearest neighbor queries. The Tukey fence method is used for anomaly detection and processing. This method identifies statistical outliers by calculating the interquartile range (IQR). The specific formula is: the anomaly determination condition is that the data value is less than Q1 - 1.5 × IQR or greater than Q3 + 1.5 × IQR, where Q1 and Q3 are the first and third quartiles, respectively, and IQR = Q3 - Q1 is the IQR. Identified outlier data points are marked, and a decision is made on whether to remove or correct them based on the specific circumstances.

[0087] The data quality score for each data point is calculated based on three dimensions: data completeness, temporal density, and spatial coverage. The data completeness score is calculated by checking the completeness of the data point's attribute fields; higher completeness results in a higher score. The temporal density score is calculated based on the data distribution density of the data point within its nearest time window; higher density indicates better temporal continuity. The spatial coverage score is calculated based on the distribution density of other data points within a certain range around the data point; higher coverage indicates stronger spatial representativeness. The three scores are combined using a weighted average to create the final data quality score, ranging from 0 to 1, with higher scores indicating better data quality. Standardized data points are then associated and stored with their corresponding data quality scores to establish a complete data quality profile.

[0088] Specifically, in one embodiment of the present invention, step S2 includes:

[0089] S21. Determine the grid resolution based on the range and data density of the area to be analyzed, and divide the area to be analyzed into a regular grid to form multiple spatial locations;

[0090] S22. Establish the mapping relationship between spatial location and data points, and assign the standardized data points to the corresponding grid spatial locations according to their spatial coordinates.

[0091] S23. For each spatial location, collect all data points belonging to it, sort them by timestamp, and combine them with attribute numerical information to construct a time series for that location;

[0092] S24. Set three different time windows for each spatial location: short-term, medium-term, and long-term. Calculate the linear trend slope using the least squares method within each time window and extract the second-order curvature features using local weighted polynomial fitting. Combine the features from the three time windows to construct a six-dimensional multi-dimensional trend feature vector.

[0093] In this embodiment, the grid resolution is first determined based on the range and data density of the region to be analyzed, and the region to be analyzed is divided into regular grids to form multiple spatial locations. The selection of the grid resolution needs to comprehensively consider the spatial distribution density of data points, the required analysis accuracy, and the balance of computational efficiency, and is set to 1 to 3 times the average spacing between data points.

[0094] After standardization, data points are assigned to corresponding grid spatial locations based on their spatial coordinates, establishing a mapping relationship between spatial locations and data points. The center point coordinates of each grid cell serve as the representative coordinates of that spatial location, and all data points falling within the same grid cell belong to that spatial location. For each spatial location, all assigned data points are collected, sorted by timestamp, and combined with attribute numerical information to construct a time series for that location. The time series consists of a series of ordered pairs of timestamps and corresponding attribute values, in the form (t1, V). i (t1)),(t2,V i (t2)),...,(t n V i (t n ))$, where i is the spatial location index, t j For timestamps, V i (t j ) represents the attribute value corresponding to this timestamp.

[0095] For each spatial location, three different time windows {w1, w2, w3} are set: short-term, medium-term, and long-term. The length of the time window is determined according to the total time span of the time series and the analysis requirements. Typically, the short-term window covers 15% to 25% of the total time span, the medium-term window covers 30% to 50%, and the long-term window covers 60% to 90%.

[0096] The slope of the linear trend is calculated using the least squares method within each time window, using the following formula:

[0097]

[0098] Where i is the spatial location index, w k k = {1, 2, 3} is the time window. For spatial location i within time window w k The slope of the linear trend within the time window w, where n is the spatial location i within the time window w. kThe number of data points within, t is the timestamp of the data point, V i (t) represents the attribute value of the data point corresponding to spatial location i at timestamp t. This indicates that spatial location i is within time window w. k The set of timestamps for all data points within the set.

[0099] Simultaneously, second-order curvature features are extracted using local weighted polynomial fitting, by fitting a quadratic polynomial V. i (t)=a2t 2 +a1t+a0 extracts the second-order coefficient a2 as the curvature feature. This feature reflects the acceleration or deceleration trend of attribute changes. A six-dimensional multi-dimensional trend feature vector is constructed by combining the linear trend slope and second-order curvature features from three time windows. This vector can simultaneously capture the linear trend and nonlinear acceleration characteristics of spatial location across short-term, medium-term, and long-term time scales.

[0100] For spatial locations with insufficient time series data points, when the number of data points within a certain time window is less than 3, data from adjacent time windows are used to supplement the data, or the feature dimension is set to null and specially marked in subsequent processing. The six-dimensional trend feature vector of each spatial location is normalized using the Z-score standardization method to ensure that the numerical range of different feature dimensions is consistent. The normalization formula is (x-μ) / σ, where μ is the mean of the feature dimension and σ is the standard deviation.

[0101] Specifically, in one embodiment of the present invention, step S3 includes:

[0102] S31. Take each spatial location as the target spatial location in turn, and select other nearby spatial locations centered on the target spatial location according to Euclidean distance to construct the initial candidate neighborhood.

[0103] S32. For each candidate spatial location in the initial candidate neighborhood, calculate the comprehensive quality score of the spatial location based on the data quality score of its associated data point, and eliminate spatial locations with a comprehensive quality score lower than the quality threshold.

[0104] S33. When the number of candidate spatial locations after screening is less than the preset minimum value, an adaptive expansion mechanism is activated to gradually expand the search radius and reconstruct the final candidate neighborhood that meets the quantity requirements.

[0105] In this embodiment, each spatial location is first processed sequentially as a target spatial location. Then, the Euclidean distance from each target spatial location to all other spatial locations is calculated based on their spatial coordinates. The Euclidean distance calculation formula is:

[0106]

[0107] Where (x) i ,y i ) and (x j ,y j ) are the coordinates of the target spatial location and the candidate spatial location, respectively.

[0108] Sort the data in ascending order according to the calculated Euclidean distance, and select the nearest spatial locations to construct the initial candidate neighborhood. The size of the initial candidate neighborhood is determined based on the spatial distribution density of the data and the required analytical precision; typically, 10 to 20 nearest neighbor spatial locations are selected as the initial candidate neighborhood.

[0109] For each candidate spatial location in the initial candidate neighborhood, a comprehensive quality score is calculated based on the data quality scores of its associated data points. The comprehensive quality score is obtained by weighted averaging of the quality scores of all associated data points for that spatial location, using the following formula:

[0110]

[0111] Q i For the overall quality score of spatial location i, n i q represents the number of data points to which this location belongs. k For the quality score of the k-th data point, ω k These are the corresponding weights. The weights are determined based on the temporal relevance and spatial representativeness of the data points; data points that are more recent in time and have stronger spatial representativeness have greater weights.

[0112] A quality threshold is set to filter candidate spatial locations, eliminating those with an overall quality score below the threshold. The quality threshold is set between 0.6 and 0.8, with the specific value determined based on the overall distribution of data quality and the required analytical precision. After quality filtering, the number of remaining candidate spatial locations is counted to determine if they meet the preset minimum requirement.

[0113] The preset minimum number of candidate locations is set to 5 to 8 to ensure the stability and reliability of subsequent interpolation processing. When the number of candidate spatial locations after filtering is less than the preset minimum, an adaptive expansion mechanism is activated to reconstruct the candidate neighborhood.

[0114] The adaptive expansion mechanism increases the number of candidate spatial locations by gradually expanding the search radius. The initial search radius is set to r0 = 3 × grid resolution, and each expansion increases the radius by Δr = r0. The expansion process is executed at most twice, meaning the search radius is expanded to a maximum of r0 + 2Δr. After each expansion, the comprehensive quality score of the newly added spatial locations within the expanded range is recalculated, and they are filtered according to the same quality threshold. Spatial locations that meet the requirements are added to the candidate neighborhood. If the number of candidate spatial locations is still insufficient after two expansions, the quality threshold is lowered and the filtering is repeated, or existing candidate locations are used for further processing, and the data quality level of that area is marked as low reliability. The final candidate neighborhood obtained through the above processing contains spatial locations that meet the requirements in both spatial proximity and data quality dimensions.

[0115] Specifically, in one embodiment of the present invention, step S4 includes:

[0116] S41. Extract the common time period between each candidate spatial location and the target spatial location in the final candidate neighborhood;

[0117] S42. Recalculate the multi-dimensional trend feature vectors of both based on the common time period;

[0118] S43. Calculate the trend matching degree based on the linear trend slope, and calculate the comprehensive similarity by combining the recalculated multi-dimensional trend feature vector.

[0119] S44. Determine the optimal similarity threshold and neighborhood number parameters;

[0120] S45. Filter candidate spatial locations based on the optimal similarity threshold, and select the top few locations in descending order of comprehensive similarity to form an effective neighborhood.

[0121] In this embodiment, the common observation time period T between each candidate spatial location j and the target spatial location i in the final candidate neighborhood is first determined. common =T i ∩T j T i T is the set of observation times for the target's spatial location. j This is the set of observation times for candidate spatial locations, which is also the set of timestamps.

[0122] The multi-dimensional trend feature vectors of the two spatial locations are recalculated within a common observation period to ensure comparability of the feature vectors. For the target spatial location i, the six-dimensional trend feature vector v′ is recalculated based on data from the common time period. i For candidate spatial location j, its six-dimensional trend feature vector v′ is also recalculated based on the common time period. j .

[0123] The trend matching degree between two spatial locations is calculated based on the recalculated linear trend slope feature:

[0124]

[0125] Among them, w k k = {1, 2, 3} is the time window. and Spatial positions i and j are respectively located in time window w. k The slope of the linear trend within the range, ∈ is a small positive number to prevent the denominator from being zero.

[0126] Among them, adopt As a normalization benchmark, the absolute value of the larger slope is used as the reference standard; when both slopes are small, the denominator is close to ε, and the similarity is mainly determined by the absolute difference; when one slope is large and the other is small, the denominator is determined by the larger slope, thus avoiding the bias of relative comparison.

[0127] This represents the absolute difference between two slopes. Provide a normalized benchmark. The relative difference is represented by 1; the similarity component of the time window is obtained by weighted multiplication of the similarity components of the three time windows.

[0128] The cosine similarity is calculated by combining the recalculated multi-dimensional trend feature vectors. The formula is as follows:

[0129]

[0130] This similarity measure is the angular similarity between two feature vectors in a multidimensional space.

[0131] The overall similarity is obtained by multiplying the trend matching degree by the cosine similarity:

[0132] S(i,j)=α1·M(i,j)+α2·|cos(v′ i ,v′ j )|

[0133] Where, v′ i and v′ j These are the multi-dimensional trend feature vectors recalculated for spatial locations i and j based on a common time period, cos(v′) i ,v′ j ) is v′ i and v′ jThe cosine similarity is calculated, where α1 and α2 are the weights of trend matching and cosine similarity, respectively, and α1 + α2 = 1, α1, α2 > 0. The comprehensive similarity considers both trend consistency and the overall similarity of feature vectors.

[0134] In this embodiment, leave-one-out cross-validation is used to search for parameters within a similarity threshold range of 0.3 to 0.8 with a step size of 0.1 and a neighborhood size range of 3 to 12. For each parameter combination, historical data is used for validation, the average error of the interpolation prediction is calculated, and the parameter combination with the smallest error is selected as the optimal setting.

[0135] Based on the optimal similarity threshold determined by cross-validation, spatial locations that meet the required overall similarity are selected from the candidate neighborhoods. If the selection results meet the requirement for the number of optimal neighborhoods, the top few locations are selected in descending order of overall similarity. If the selection results are insufficient for the number of optimal neighborhoods, all locations that pass the threshold selection are chosen as valid neighborhoods.

[0136] When the number of locations filtered by the similarity threshold is insufficient to meet the minimum neighborhood requirement, a threshold of 3 is typically set, employing a decreasing threshold strategy. Each time, the similarity threshold is reduced by 0.05, and the filtering process is re-executed until a sufficient number of valid neighborhoods are obtained. A minimum threshold limit of 0.1 is set to prevent the filtering from being too lenient.

[0137] Specifically, in one embodiment of the present invention, step S5 includes:

[0138] S51. Calculate the spatial distance weight from each spatial location in the effective neighborhood to the target spatial location;

[0139] S52. Spatial interpolation calculation is performed on the time series of each spatial location within the effective neighborhood using the inverse distance weighting method.

[0140] S53. When original time series data exists for the target spatial location, a weighted fusion is performed based on the comprehensive quality score of the time series and the interpolation results.

[0141] S54. Use the results obtained by fusion or interpolation as the complete time series of that spatial location.

[0142] In this embodiment, the effective neighborhood N′ is first... i For each spatial location j, its spatial distance weight to the target spatial location i is calculated, and the weight coefficient is calculated using a Gaussian distance decay function.

[0143]

[0144] Where d ij For the ellipsoidal distance, σ dThis is the distance attenuation parameter, which is determined based on the spatial distribution density of the data and is typically set to 0.5 to 1.5 times the average neighbor distance.

[0145] Spatial interpolation is performed on the time series of each spatial location within the effective neighborhood using an inverse distance weighting method. The interpolation calculation formula for the target spatial location at time point t is as follows:

[0146]

[0147] Where V j (t) represents the observation value of spatial location j in the effective neighborhood at time t, w ij These are the corresponding distance weighting coefficients. When the target spatial location i has original observation data V at time point t... i When (t), confidence fusion processing is required based on the comprehensive quality score of the observation data and the interpolation results.

[0148] The formula for calculating confidence fusion is as follows: This formula integrates the original observation data with the spatial interpolation results, preserving the authenticity of the original data while supplementing any missing or erroneous information through spatial data.

[0149] Based on the different stage characteristics of the target location's time series, a segmented adaptive enhancement strategy is adopted for differentiated processing. During the inverse distance weighted interpolation process, when the coefficient of variation (CV) of the target location's neighborhood data is large, it indicates significant data fluctuation in that region; in this case, the conservatism of the interpolation should be appropriately increased. In the rapidly changing stage, i.e., when the coefficient of variation (CV) is high... i >CV high At this stage, the base threshold τ0 is increased to rely more on high-matching nearest neighbors, ensuring that the interpolation results accurately reflect rapidly changing features. In the stationary phase, i.e., CV... i <CV low When interpolating, the threshold is appropriately lowered to expand the effective neighbor range and utilize more spatial information to improve the stability of interpolation. During transition phases, special attention is paid to neighbor locations that also experience transitions, and their contribution to interpolation is enhanced through weighted adjustments. For cases where observation data is completely missing at certain time points for the target spatial location, the spatial interpolation result is directly used as the value for that time point, and the data source is labeled as spatial augmentation. The time series data obtained after interpolation or confidence fusion undergoes an integrity check to ensure that all time points have corresponding values, while recording the processing method and quality level information for each data point. The processed complete time series is used as the final time series data for that spatial location.

[0150] Specifically, in one embodiment of the present invention, step S6 includes:

[0151] S61. Perform time-domain decomposition on the complete time series for each spatial location to identify the trend component, periodic component, and random component of each spatial location.

[0152] S62. Identify spatial locations with similar change patterns and aggregate them into spatial regions through spatial autocorrelation analysis;

[0153] S63. Classify and identify the spatiotemporal evolution process at the regional level based on the identified spatial regions;

[0154] S64. Organize the analysis results according to spatial regions and generate a spatiotemporal change analysis report of the region to be analyzed, which includes hotspot areas of change, evolution trend prediction, and anomaly detection.

[0155] In this embodiment, the complete time series of each spatial location is first decomposed in the time domain to establish a geographic semantic perception time series decomposition framework. This framework not only performs mathematical component decomposition, but also identifies and labels time patterns with geographical significance.

[0156] The process of constructing a geographic semantic pattern library is as follows: First, basic geographic change patterns are defined, including five basic patterns: monotonically increasing, monotonically decreasing, periodic fluctuations, step changes, and abrupt changes. Each pattern is represented by a standardized time series template, with a fixed template length of 30 time points. For the input time series, a subsequence of the same length is extracted using a sliding window method, and the Pearson correlation coefficient is calculated between this subsequence and the template in the pattern library. The pattern corresponding to the template with the highest correlation coefficient is the semantic annotation result for that time period.

[0157] Establish a pattern matching degree calculation mechanism, the calculation formula is as follows:

[0158]

[0159] Among them PMD k (t) represents the matching degree of the k-th semantic pattern at time t, DTW(·) is the dynamic time warping distance function, and V i (t) represents the observation sequence. For the i-th feature sequence of pattern k, w i The corresponding weights are σ, a standardization parameter. The DTW distance is converted into similarity using an exponential function; the smaller the distance, the higher the similarity. The time series is then decomposed as follows:

[0160] X(t) = T(t) + S(t) + R(t)

[0161] Where T(t) is the trend component, S(t) is the periodic component, and R(t) is the random component. A multi-level intelligent time anomaly identification system is established, constructing a three-layer anomaly detection architecture. The statistical anomaly layer is based on the improved CUSUM method, and the calculation formula is as follows:

[0162]

[0163] Introducing adaptive control limit h t =h0·(1+γ·CV) t ), where CV t This is the rolling coefficient of variation. The trend anomaly layer is based on trend change rate detection, and the calculation formula is:

[0164]

[0165] When TCD>TCD threshold The time is identified as an anomaly in the trend. The pattern anomaly layer is based on semantic pattern deviation detection, and the calculation formula is:

[0166]

[0167] When MAD>MAD threshold It was identified as a pattern anomaly.

[0168] An intelligent anomaly detection rule is established: isolated point anomalies are detected solely by the statistical layer; trend reversal anomalies are detected jointly by the statistical and trend layers; pattern disruption anomalies are detected jointly by the statistical and pattern layers; and systematic anomalies are detected simultaneously by all three layers. Specifically, the statistical anomaly layer uses the 3σ principle: data points deviating from the mean by more than three times the standard deviation are considered statistically abnormal. The trend anomaly layer calculates the slope of the linear trend over a 30-day sliding window; when the absolute value of the slope exceeds the 95th percentile of the historical slope distribution, it is considered a trend anomaly. The pattern anomaly layer identifies patterns as abnormal when the highest correlation coefficient between the time series and all patterns in the semantic pattern library is below 0.3, indicating the emergence of unknown change patterns. The Hurst exponent of the time series is calculated to assess its long-term memory and predictability; the Ljung-Box test is used to evaluate the white noise characteristics of the series; and an ARIMA or state-space model is established to describe the mathematical characteristics of the time series, providing quantitative evidence for short-term forecasting and trend extrapolation.

[0169] Spatial autocorrelation analysis was used to identify spatial locations with similar change patterns. Moran's I index was then used to calculate the spatial autocorrelation, with the following formula:

[0170]

[0171] Where n is the total number of spatial locations, w ij Let x be the spatial weight matrix. i For the attribute value of spatial location i, This represents the mean of the attribute values.

[0172] Based on spatial autocorrelation results and time series similarity measures, K-means clustering or hierarchical clustering methods are used to aggregate spatial locations with similar spatiotemporal variation patterns into spatial regions, forming spatiotemporal evolution pattern partitions. Based on the identified spatial regions, the spatiotemporal evolution process is classified and labeled at the region level. The spatial adjacency relationships and transition probabilities between different pattern regions are calculated, and a Markov chain model of spatiotemporal evolution is constructed to reveal the macroscopic evolution mechanism of geographical phenomena. Multi-scale visualization analysis results are generated, including spatiotemporal change trajectory maps, trend intensity distribution maps, pattern partition maps, and anomaly detection result maps. The analysis results are organized by spatial region, and a structured analysis report is output, containing key information such as hotspot identification, evolution trend prediction, anomaly detection, change rate statistics, turning point identification, and confidence interval estimation. Multiple data formats such as JSON, XML, and GeoJSON are supported for output, forming a complete spatiotemporal change analysis report for the analyzed region.

[0173] In addition, such as Figure 2 As shown, the present invention also provides an intelligent analysis system for geographic information mapping data, the system being used to implement the method described in any of the above-mentioned embodiments, the system comprising:

[0174] The data preprocessing module is used to standardize the format of the input geographic information mapping data, handle abnormal data, and calculate quality scores.

[0175] The gridding module is used to divide the area to be analyzed into a regular grid and construct a time series.

[0176] The neighborhood construction module is used to build candidate neighborhoods for each target spatial location and perform quality screening, outputting the final candidate neighborhoods for each spatial location;

[0177] The feature matching module is used to calculate the trend feature matching degree of spatial locations within the candidate neighborhood and to filter the effective neighborhood.

[0178] The interpolation processing module is used to perform time series interpolation or fusion processing based on the effective neighborhood and output a complete time series.

[0179] The analysis report module is used to perform spatiotemporal change analysis and generate analysis reports.

[0180] In this embodiment, the intelligent analysis system for geographic information mapping data adopts a modular architecture design, consisting of six core functional modules. These modules work collaboratively to achieve intelligent analysis and processing of geographic information mapping data. The data preprocessing module, as the system's data input, is responsible for receiving geographic information mapping data from different data sources, including GPS measurement data, remote sensing image data, topographic map data, engineering measurement data, and other data types. This module has built-in parsers for various data formats, supporting automatic recognition and parsing of mainstream data formats such as ESRI Shapefile, GeoJSON, KML, GPX, and CSV. The format standardization processing subunit uses the GDAL / OGR open-source geographic data processing library to achieve unified coordinate system conversion, supporting mutual conversion between multiple coordinate systems such as WGS84, UTM, and Gauss-Krüger projection, ensuring that all input data uses a unified spatial reference system. The time reference standardization processing uses UTC time as the unified reference, automatically recognizing and converting various time formats, including ISO8601 standard time, local time, and timestamps. The anomaly data processing subunit integrates multiple anomaly detection algorithms, including statistical distribution-based anomaly detection, clustering-based anomaly detection, and time series-based anomaly detection methods, which can automatically identify and label data points that significantly deviate from the normal range. The quality score calculation subunit establishes a comprehensive evaluation system and generates a quality score value between 0 and 1 for each data point.

[0181] The gridding module is responsible for spatially discretizing the geographic area to be analyzed, automatically calculating the optimal grid resolution based on the input geographic boundary range and data density analysis results. This module uses a quadtree data structure to optimize spatial partitioning efficiency, employing a finer grid resolution in data-dense areas and a relatively coarser grid resolution in data-sparse areas, achieving an adaptive grid partitioning strategy. The spatial indexing construction sub-unit uses the R-tree algorithm to build an efficient spatial index structure, supporting fast spatial neighborhood queries and range retrieval operations. The time series construction sub-unit sorts data points belonging to the same grid unit according to timestamps, constructs the original time series for that spatial location, and automatically identifies the sampling frequency and time interval characteristics of the time series. The multi-dimensional trend feature extraction sub-unit implements an automatic division mechanism for short-term, medium-term, and long-term time windows, using sliding window technology to extract trend features at each time scale, including multi-dimensional feature parameters such as linear trend slope and second-order curvature features.

[0182] The neighborhood construction module intelligently builds high-quality candidate neighborhoods for each target spatial location. This module employs a multi-level neighborhood construction strategy to ensure the spatial representativeness and data quality of the neighborhoods. The spatial distance calculation subunit uses an ellipsoidal distance calculation method, taking into account the Earth's surface features to calculate accurate spatial distances, and supports distance calculations under different projected coordinate systems. The initial candidate neighborhood construction subunit quickly retrieves candidate spatial locations around the target location using a spatial index structure based on a set search radius or neighbor number threshold. The quality screening subunit filters candidate spatial locations based on their comprehensive quality score, eliminating locations with a quality score below a set threshold to ensure the data quality of the candidate neighborhoods. The adaptive expansion mechanism automatically activates when the number of candidate neighbors is insufficient, increasing the number of candidate locations by gradually expanding the search radius, while dynamically adjusting the quality threshold to balance the requirements of neighborhood size and data quality.

[0183] The feature matching module implements an intelligent matching algorithm based on multi-dimensional trend features, ensuring that the selected effective neighborhoods are highly similar to the target location in terms of trend features. The common time period extraction subunit automatically identifies the overlapping time series intervals between candidate and target spatial locations, ensuring fairness and consistency in feature comparison. The trend feature recalculation subunit recalculates the multi-dimensional trend feature vectors of candidate and target locations based on the common time period, eliminating the impact of time range differences on feature comparison. The trend matching degree calculation subunit calculates the trend matching degree using a weighted geometric average. The comprehensive similarity calculation subunit combines the trend matching degree with cosine similarity to form a comprehensive similarity metric. The dynamic threshold calculation subunit determines the optimal similarity threshold and neighborhood number parameters based on historical data, filters candidate spatial locations according to the optimal similarity threshold, and selects the top few locations in descending order of comprehensive similarity to form the effective neighborhood.

[0184] The interpolation module performs intelligent interpolation on the target spatial location based on the selected effective neighborhood, generating high-quality complete time series data. The spatial weight calculation subunit uses a Gaussian distance decay function to calculate the weight coefficients of each spatial location within the effective neighborhood; closer locations contribute more to the interpolation. The inverse distance weighted interpolation subunit implements a classic spatial interpolation algorithm, calculating a weighted average of the time series data for each location within the neighborhood based on spatial weights. The quality fusion processing subunit, for locations with existing original observation data, intelligently fuses the original data with the interpolation results based on their quality scores, preserving the authenticity of the original data while enhancing quality using spatial information. The integrity verification subunit performs a quality check on the generated complete time series, ensuring that all time points have corresponding values ​​and marking the source and processing method of each data point.

[0185] The analysis report module performs in-depth spatiotemporal variation analysis on complete time series data, generating comprehensive analysis reports and visualization results. The time domain decomposition sub-unit employs advanced algorithms such as empirical mode decomposition and wavelet decomposition to decompose the time series into components, identifying trend components, periodic components, and random components, revealing the inherent structural characteristics of the time series. The spatial autocorrelation analysis sub-unit uses spatial statistical methods such as Moran's I index and Geary's C index to analyze the correlation between spatial locations, identifying spatial location groups with similar change patterns. The clustering analysis sub-unit uses methods such as K-means clustering and hierarchical clustering to aggregate similar spatial locations into spatial regions, forming geographically significant spatial partitions. The spatiotemporal evolution analysis sub-unit classifies the spatiotemporal evolution processes at the regional level based on the identified spatial regions, constructing Markov chain models to describe the transformation relationships between regions. The report generation sub-unit integrates the analysis results into a structured analysis report, including identification of change hotspot areas, evolution trend prediction, anomaly detection, and statistical summary information. The visualization output sub-unit generates various types of charts and maps, including spatiotemporal change trajectory maps, trend intensity distribution maps, pattern zoning maps, and anomaly detection result maps, supporting multiple image format outputs. The system also provides a flexible data interface, supporting result output in multiple formats such as JSON, XML, and GeoJSON, facilitating integration with other geographic information systems and data analysis platforms.

[0186] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent analysis of geographic information mapping data, characterized in that, include: S1. Determine the area to be analyzed, obtain geographic information mapping data within the area to be analyzed, including spatial coordinates, timestamps and attribute values, standardize the data points and generate data quality scores. S2. The area to be analyzed is gridded into multiple spatial locations. The standardized data points are assigned to the corresponding spatial locations according to their spatial coordinates. For each spatial location, the corresponding data points are arranged according to timestamps and attribute values ​​to construct a time series. Based on the time series, multi-dimensional trend feature vectors are extracted in different time windows. Step S2 includes: S21. Determine the grid resolution based on the range and data density of the area to be analyzed, and divide the area to be analyzed into a regular grid to form multiple spatial locations; S22. Establish the mapping relationship between spatial location and data points, and assign the standardized data points to the corresponding grid spatial locations according to their spatial coordinates. S23. For each spatial location, collect all data points belonging to it, sort them by timestamp, and combine them with attribute numerical information to construct a time series for that location; S24. Set up three different time windows for each spatial location: short-term, medium-term and long-term. Calculate the linear trend slope using the least squares method in each time window and extract the second-order curvature feature using local weighted polynomial fitting. Combine the features of the three time windows to construct a six-dimensional multi-dimensional trend feature vector. The formula for calculating the slope of a linear trend is: ； in, For spatial location index, For time window, For spatial location In the time window The slope of the linear trend within, For spatial location In the time window Number of data points within, For the timestamp of the data point, For spatial location In timestamp The attribute values ​​corresponding to the data points, Indicates spatial location In the time window The set of timestamps for all data points within; S3. For each spatial location, construct an initial candidate neighborhood based on spatial distance. Use data quality scoring to filter and eliminate unqualified locations from the initial candidate neighborhood. When the number of candidate neighbors is insufficient, use an adaptive expansion mechanism to reconstruct the neighborhood and obtain the final candidate neighborhood. S4. Based on the multi-dimensional trend feature vector, calculate the trend feature matching degree between each candidate neighbor in the final candidate neighborhood and the target spatial location, determine the optimal similarity threshold, and filter to obtain the effective neighborhood. Step S4 includes: S41. Extract the common time period between each candidate spatial location and the target spatial location in the final candidate neighborhood; S42. Recalculate the multi-dimensional trend feature vectors of both based on the common time period; S43. Calculate the trend matching degree based on the linear trend slope, and calculate the comprehensive similarity by combining the recalculated multi-dimensional trend feature vector. S44. Determine the optimal similarity threshold and neighborhood number parameters; S45. Filter candidate spatial locations based on the optimal similarity threshold, and select the top few locations in descending order of comprehensive similarity to form an effective neighborhood. Trend matching degree The calculation formula is: ； in, For time window, and Spatial location and spatial location In the time window The slope of the linear trend within, To prevent small positive numbers with a denominator of zero; Overall similarity The calculation formula is: ； in, and Spatial location and spatial location The multi-dimensional trend feature vector is recalculated over a common time period. for and cosine similarity, , These are the weights for trend matching and cosine similarity, respectively. ; S5. Interpolate the time series of the target spatial location based on the effective neighborhood to obtain the complete time series of each spatial location; S6. Perform spatiotemporal change pattern analysis on the complete time series, identify spatiotemporal evolution laws, and generate a spatiotemporal change analysis report.

2. The intelligent analysis method for geographic information mapping data according to claim 1, characterized in that, Step S1 includes: S11. Standardize the format of geographic information mapping data and unify the coordinate system and time reference. S12. Establish an R-tree spatial index structure to support efficient spatial neighborhood queries; S13. Use the Tukey fence method to identify and handle abnormal data points; S14. Calculate the quality score for each data point based on data completeness, temporal density, and spatial coverage; S15. Associate and store the standardized data points with their corresponding data quality scores.

3. The intelligent analysis method for geographic information mapping data according to claim 1, characterized in that, Step S3 includes: S31. Take each spatial location as the target spatial location in turn, and select other nearby spatial locations centered on the target spatial location according to Euclidean distance to construct the initial candidate neighborhood. S32. For each candidate spatial location in the initial candidate neighborhood, calculate the comprehensive quality score of the spatial location based on the data quality score of its associated data point, and eliminate spatial locations with a comprehensive quality score lower than the quality threshold. S33. When the number of candidate spatial locations after screening is less than the preset minimum value, an adaptive expansion mechanism is activated to gradually expand the search radius and reconstruct the final candidate neighborhood that meets the quantity requirements.

4. The intelligent analysis method for geographic information mapping data according to claim 1, characterized in that, Step S5 includes: S51. Calculate the spatial distance weight from each spatial location in the effective neighborhood to the target spatial location; S52. Spatial interpolation calculation is performed on the time series of each spatial location within the effective neighborhood using the inverse distance weighting method. S53. When original time series data exists for the target spatial location, a weighted fusion is performed based on the comprehensive quality score of the time series and the interpolation results. S54. Use the results obtained by fusion or interpolation as the complete time series of that spatial location.

5. The intelligent analysis method for geographic information mapping data according to claim 1, characterized in that, Step S6 includes: S61. Perform time-domain decomposition on the complete time series for each spatial location to identify the trend component, periodic component, and random component of each spatial location. S62. Identify spatial locations with similar change patterns and aggregate them into spatial regions through spatial autocorrelation analysis; S63. Classify and identify the spatiotemporal evolution process at the regional level based on the identified spatial regions; S64. Organize the analysis results according to spatial regions and generate a spatiotemporal change analysis report of the region to be analyzed, which includes hotspot areas of change, evolution trend prediction, and anomaly detection.

6. A geographic information mapping data intelligent analysis system, characterized in that, The system is used to implement the method as described in any one of claims 1-5, the system comprising: The data preprocessing module is used to standardize the format of the input geographic information mapping data, handle abnormal data, and calculate quality scores. The gridding module is used to divide the area to be analyzed into a regular grid and construct a time series. The neighborhood construction module is used to build candidate neighborhoods for each target spatial location and perform quality screening, outputting the final candidate neighborhoods for each spatial location; The feature matching module is used to calculate the trend feature matching degree of spatial locations within the candidate neighborhood and to filter the effective neighborhood. The interpolation processing module is used to perform time series interpolation or fusion processing based on the effective neighborhood and output a complete time series. The analysis report module is used to perform spatiotemporal change analysis and generate analysis reports.

Images