Methods and Systems for Monitoring Dynamic Changes in Geographic Information Based on Spatial Big Data

By constructing a multi-source spatial big data fusion benchmark framework and calculating the temporal difference index, a trend prediction model was established, solving the problems of data fusion, change detection, and trend prediction in geographic information change monitoring, and realizing efficient and intelligent dynamic change monitoring and visual early warning.

CN123068889APending Publication Date: 2026-11-10CHONGQING PLANNING & NATURAL RESOURCES INFORMATION CENT
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611106836.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-24
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing geographic information change monitoring technologies suffer from problems such as insufficient data fusion capabilities, limited change detection accuracy, weak dynamic trend prediction capabilities, low level of intelligence in compliance judgment, and imperfect expression and early warning mechanisms for monitoring results.

Method used

By constructing a multi-source spatial big data fusion benchmark framework, using the calculation of temporal difference index and the quantification of change intensity, a trend prediction model is established to conduct a comprehensive compliance score, and a visual variable mapping and hierarchical early warning model is established to achieve dynamic change monitoring.

Benefits of technology

It achieves efficient fusion of multi-source heterogeneous data, accurately identifies areas of change and quantitatively describes the characteristics of change, provides forward-looking trend prediction and intelligent compliance judgment, dynamically visualizes monitoring results and triggers tiered early warnings.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123068889A_ABST
    Figure CN123068889A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for monitoring dynamic changes in geographic information based on spatial big data, belonging to the fields of geographic information science and spatial big data processing technology. It includes: S1, collecting multi-source spatial big data and performing quality assessment and weighted fusion to construct a unified benchmark framework data; S2, identifying potential change areas through temporal difference index calculation and extracting change features through change intensity quantification, generating a change detection result layer; S3, constructing a change rate time series and establishing a trend prediction model, outputting a change trend prediction layer and an anomaly area layer; S4, intelligently comparing the identified change detection result layer and the trend prediction layer, outputting a change compliance judgment result; S5, establishing a mapping model from change intensity to visual variables, triggering a tiered early warning model, generating a monitoring report, and issuing early warning information.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of geographic information science and spatial big data processing technology, specifically relating to a method and system for monitoring dynamic changes in geographic information based on spatial big data. Background Technology

[0002] Dynamic monitoring of geographic information changes is a core technological means to understand the spatiotemporal evolution of Earth's surface resources and environment. With the rapid development of remote sensing, Internet of Things, and mobile Internet technologies, spatial big data is experiencing explosive growth, providing a data foundation for multi-scale, high-frequency, and high-precision dynamic monitoring of geographic information changes.

[0003] Existing geographic information change monitoring technologies mainly suffer from the following shortcomings:

[0004] First, the data fusion capability is insufficient. Existing methods typically process single data sources and lack a systematic fusion mechanism for multi-source heterogeneous spatial data such as satellite remote sensing imagery, aerial photography data, ground sensor data, and vector geographic boundary data. This results in monitoring results with a single information dimension, failing to comprehensively reflect the multi-level changing characteristics of geographic information.

[0005] Second, the accuracy of change detection is limited. Traditional change detection algorithms are mostly based on single-pixel spectral differences or simple threshold judgments, without fully considering the spatial context, temporal evolution, and multi-scale feature differences of land cover types. This results in a high false alarm rate and a high false negative rate, making it difficult to meet the needs of accurate monitoring.

[0006] Third, the ability to predict dynamic trends is weak. Existing technologies focus on the post-event detection and identification of changes, lacking a forward-looking prediction mechanism for geographic information change trends. They cannot achieve quantitative prediction of the rate and direction of change, resulting in the passivity and lag of monitoring work.

[0007] Fourth, the level of intelligence in compliance assessment is low. Existing compliance analysis relies heavily on manual spatial overlay and rule comparison, lacking the ability to transform regulatory constraints into mathematical judgment models. This results in low assessment efficiency and high subjectivity, making it difficult to meet the needs of large-scale dynamic monitoring.

[0008] Fifth, the expression and early warning mechanisms for monitoring results are inadequate. The existing system's visualization is mainly static, lacking a dynamic mechanism to map the intensity of change, trends, and compliance status into intuitive visual expressions and tiered early warning signals, making it difficult for decision-makers to quickly obtain key information. Summary of the Invention

[0009] This invention aims to at least solve the technical problems existing in the prior art, and in particular, it innovatively proposes a method and system for monitoring dynamic changes in geographic information based on spatial big data.

[0010] To achieve the above-mentioned objectives of this invention, this invention provides a method for monitoring dynamic changes in geographic information based on spatial big data, comprising the following steps:

[0011] S1. Collect multi-source spatial big data and perform quality assessment and weighted fusion to construct a unified benchmark framework data;

[0012] S2. Potential change areas are identified by calculating the temporal difference index and change features are extracted by calculating the change intensity quantification value, generating a change detection result layer;

[0013] S3. Construct a time series of change rates and establish a trend prediction model, outputting a change trend prediction layer and an anomaly region layer;

[0014] S4. Spatially overlay and compare the identified change detection result layer with the trend prediction layer and perform difference analysis to conduct a comprehensive compliance score and output the change compliance judgment result.

[0015] S5. Establish a mapping model from change intensity to visual variables, including hue, symbol size, and transparency, and trigger a graded early warning model to generate a monitoring report and issue early warning information.

[0016] In the preferred embodiment of the above technical solution, the multi-source spatial big data in S1 includes time-series satellite remote sensing images, aerial photogrammetry data, ground sensor network data, and vector geographic boundary data.

[0017] In the preferred embodiment of the above technical solution, the quality assessment and weighted fusion in S1 includes verifying the geometric precision constraints and calculating the time freshness factor for each data source, determining the fusion weight of each data source based on the test results, and generating a unified benchmark framework data with geometric precision constraints and time freshness factor calculation through weighted fusion.

[0018] In the preferred embodiment of the above technical solution, the extraction of change features in S2 includes calculating a quantitative value of change intensity to describe the significance of the change, and extracting change direction features to determine the type of change, and generating a change detection result layer by combining change intensity and change direction.

[0019] In the preferred embodiment of the above technical solution, the establishment of the prediction model in S3 includes establishing a trend prediction model based on the historical rate of change time series, predicting the intensity of change in future periods by analyzing the time evolution law of the rate of change, and outputting a trend prediction layer.

[0020] In the preferred embodiment of the above technical solution, the abnormal region layer in S3 is obtained by establishing an abnormality identification model. This model identifies regions that deviate from the normal change trend as abnormal regions and outputs an abnormal region layer based on the normal fluctuation range of the change rate time series.

[0021] In the preferred embodiment of the above technical solution, the comprehensive compliance score in S4 includes, based on the spatial superposition relationship between the change area and the regulatory constraint area, combined with the compliance judgment rules for the change type, quantitatively assessing the compliance level of the change, and generating a compliance judgment result that includes a compliance score and judgment basis.

[0022] In the preferred embodiment of the above technical solution, the step S5 of mapping the intensity of change to a visual variable includes a graded mapping of color depth and symbol size based on the magnitude of the intensity of change, so that the degree of change is presented intuitively through visual variables.

[0023] In the preferred embodiment of the above technical solution, triggering the graded early warning in S5 includes dividing the intensity of change into multiple early warning levels based on a preset threshold, determining the urgency and release priority of the early warning information according to the early warning level, and automatically releasing the early warning information through multiple channels.

[0024] The present invention also discloses a computer system, comprising:

[0025] processor;

[0026] Memory used to store instructions executed by the processor;

[0027] The processor is configured to implement any of the methods described above when executing the instructions.

[0028] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0029] This invention addresses the challenges of data fusion caused by diverse data sources, heterogeneous formats, and inconsistent coordinate systems by establishing a fusion mechanism and unified benchmark framework for multi-source heterogeneous spatial data, thus providing a high-quality data foundation for subsequent change monitoring. It proposes an innovative method for extracting and quantifying dynamic change features of geographic information, accurately identifying change areas and quantitatively describing the intensity, type, and extent of change through multi-scale spatiotemporal joint analysis. A change trend prediction and anomaly identification model is constructed to achieve forward-looking judgments on the evolution patterns of geographic information and promptly detect abnormal changes deviating from normal evolution trends. An intelligent change compliance judgment system is established, automatically comparing changes with regulatory constraints in both spatial and attribute aspects to output compliance judgment conclusions. Finally, it enables dynamic visualization of monitoring results and tiered early warning dissemination, transforming complex spatial analysis results into intuitive visual presentations and actionable early warning information.

[0030] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0031] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0032] Figure 1 This is a flowchart of the multi-source spatial big data fusion and benchmark framework construction process;

[0033] Figure 2 This is a flowchart of the extraction and quantification process for dynamic changes in geographic information;

[0034] Figure 3 This is a flowchart for predicting dynamic trends and identifying anomalies;

[0035] Figure 4 This is a flowchart of the intelligent determination process for compliance with changes.

[0036] Figure 5 This is a flowchart for the visualization of dynamic monitoring results and the release of early warnings. Detailed Implementation

[0037] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0038] like Figures 1 to 5 As shown, this invention discloses a method and system for monitoring dynamic changes in geographic information based on spatial big data.

[0039] To address the aforementioned technical challenges, this invention comprehensively enhances the technical level and application effectiveness of dynamic change monitoring of geographic information by constructing a multi-source data fusion benchmark framework, innovating change feature extraction algorithms, establishing trend prediction and anomaly identification models, designing intelligent compliance judgment rules, and developing a dynamic visualization early warning system.

[0040] S1. Construct a multi-source spatial big data fusion and benchmark framework; unify and integrate multi-source spatial big data from different data sources with different spatial resolutions, temporal resolutions and data formats to establish a spatiotemporally consistent monitoring benchmark framework, laying the foundation for subsequent dynamic change monitoring.

[0041] First, multi-source spatial big data is collected, including time-series satellite remote sensing imagery, aerial photogrammetry data, data collected by ground-based IoT sensors, and vector geographic boundary data. These data types have different spatial resolutions, temporal frequencies, and data structures.

[0042] To address the quality differences among multi-source data, this invention proposes a multi-source data quality assessment and weight allocation formula to calculate the fusion weight of each data source. The formula is as follows:

[0043] ;

[0044] in, This represents the fusion weight of the i-th data source, and is a dimensionless parameter. Its value range is determined based on the data fusion normalization requirements; the sum of all weights equals 1 to ensure the weighted average characteristic of the fusion result. This represents the quality score of the i-th data source, reflecting the inherent quality level of the data source. The value range is [0,1], and the closer the value is to 1, the higher the data quality. This represents the time freshness factor of the i-th data source, reflecting the timeliness of the data. The calculation formula is as follows: ,in Indicates the current monitoring time. This represents the actual data collection time of the i-th data source. This represents the time half-life constant, expressed in days. This represents the spatial coverage completeness of the i-th data source, reflecting the degree of data spatial coverage. Its value range is [0,1], and the closer the value is to 1, the more complete the spatial coverage. The value represents the overall contribution of the i-th data source, which is a dimensionless parameter. The product of the three values ​​quantifies the overall performance of the i-th data source in terms of quality, timeliness, and spatial coverage. n represents the total number of data sources. , , These represent the quality score, time freshness factor, and spatial coverage completeness of the j-th data source, respectively.

[0045] In the multi-source data quality assessment and weight allocation formula, the data sources participating in the fusion calculation must meet the comprehensive fusion contribution value of at least one data source. Greater than zero, to ensure the denominator is not zero; if a data source's , or If the value is zero, then the fusion weight of this data source is zero. A data source with a contribution value of zero and not participating in the fusion calculation is selected, and at least one data source with a comprehensive fusion contribution value greater than zero is retained as a valid fusion data source.

[0046] The fusion weight of each data source is determined by three dimensions: data quality, timeliness, and spatial coverage. The numerator is the product of the three indicators of the i-th data source, and the denominator is the sum of the products of the three indicators of all data sources, thus ensuring that the sum of all weights equals 1, achieving normalization. The time freshness factor adopts an exponential decay model, which can reasonably reflect the decreasing timeliness of spatial data over time.

[0047] By scientifically allocating weights, high-quality, timely, and comprehensive data are given a larger proportion in the fusion process, improving the overall reliability of the benchmark framework. The calculation results of the fusion weights directly affect the accuracy of the fused benchmark data, and thus the accuracy of subsequent change detection. This achieves a leap from qualitative description to quantitative fusion of multi-source heterogeneous data, avoiding the subjective defects of traditional methods that rely on experience to set weights.

[0048] Quality rating The evaluation is based on a comprehensive assessment of cloud coverage, radiometric correction accuracy, and geometric correction accuracy of the imagery. The specific calculation method is as follows: ;in, This represents the cloud coverage score, with a value range of [0,1]. The lower the cloud coverage, the higher the score. , The percentage of cloud coverage recorded in the image metadata; This represents the radiometric correction accuracy score, with a value range of [0,1]. It is based on the uncertainty assessment of the radiometric calibration coefficients. The uncertainty of the radiometric calibration coefficients is determined by the radiometric calibration error parameters recorded in the satellite image metadata, or obtained through a laboratory calibration report. When the radiometric calibration uncertainty is less than 2%, Use 1.0 for 2%-5%, 0.8 for 5%-10%, 0.6 for greater than 10%, and 0.4 for more than 10%. This represents the geometric correction accuracy score, with a value range of [0,1]. It is based on the geometric positioning error assessment; when the positioning error is less than 0.5 pixels... Use 1.0 for 0.5 to 1.0 pixels, 0.8 for 1.0 to 2.0 pixels, and 0.4 for more than 2.0 pixels. , , These are the weighting coefficients for cloud coverage, radiometric correction accuracy, and geometric correction accuracy, respectively, satisfying... The weighting coefficients were determined based on industry practices for remote sensing data quality assessment. Cloud coverage directly affects the usable area of ​​the image and has the most significant impact on data availability. A value of 0.4 is used; the accuracy of radiometric correction affects the accuracy of ground reflectance inversion and is crucial for spectral analysis in change detection, therefore... Take 0.35; geometric correction accuracy affects the registration accuracy of multi-temporal images, and its impact on the accuracy of change positioning is secondary, therefore... The value is set to 0.25. The above weighting coefficients can be fine-tuned in practical applications based on the specific monitoring task requirements, but must meet the normalization constraint conditions.

[0049] time half-life The setting is based on the rate of change of land cover types in the monitoring area, where rapidly changing areas refer to areas where the monthly average change intensity exceeds a threshold of 20, and the time half-life. Values ​​are taken over 30 days; the slow-change region refers to the area where the monthly average change intensity is below the threshold of 5, and the time half-life. Values ​​taken over 365 days; moderate variation area Take 90 to 180 days; spatial coverage integrity Calculated based on the ratio of the actual coverage area to the target monitoring area. Threshold 20 and threshold 5 refer to the quantified values ​​of the change intensity. The average monthly change.

[0050] Secondly, based on the fusion weights of each data source, a multi-source spatial data fusion formula is established to generate a unified fusion formula for benchmark framework data:

[0051] ;

[0052] in, This represents the fused baseline data value at time t and coordinate position (x, y); The fusion weight of the i-th data source is calculated by the multi-source data quality assessment and weight allocation formula in S1; This represents the data value of the i-th data source at time t and coordinate position (x, y). The data from each data source are unified to the fusion reference time through temporal normalization processing. x represents the horizontal coordinate in the east-west direction, in meters; y represents the horizontal coordinate in the north-south direction, in meters. Indicates the fusion reference time. This represents the actual data collection time of the i-th data source; n represents the total number of data sources participating in the fusion.

[0053] The fused baseline data value calculated using this formula is a weighted average of data values ​​from all data sources. The weights are determined by a multi-source data quality assessment and weight allocation formula, ensuring that high-weight data sources have a greater influence on the fusion result. This weighted fusion method is suitable for the unified fusion processing of raster data, vector rasterized data, and sensor data. Its function is to generate spatiotemporally consistent fused baseline data, serving as a reference benchmark for change detection in S2. The accuracy of the fused data directly determines the sensitivity and reliability of change detection. It achieves seamless fusion of multi-source heterogeneous data within a unified coordinate framework, eliminating systematic differences between different data sources.

[0054] Specific numerical settings are based on: the coordinate position (x, y) adopts the National Coordinate System 2000; and the reference time is integrated. The time variable t is based on Julian days, and is either the weighted average of the collection times of each data source or the latest data source collection time. The total number of data sources n is determined based on the actual available data, and its value range is [3,7].

[0055] For slowly changing landforms, such as urban boundaries and topography, the temporal difference does not exceed [a certain value]. The data source is normalized using linear trend extrapolation. The extrapolation formula is: ,in This represents the average historical rate of change for that location. For rapidly changing land cover types, such as vegetation and built-up land, data sources with a temporal difference exceeding 30 days are excluded from fusion; the latest data source is used directly as the fusion result. Temporal normalization and time freshness factor are also considered. The relationship is as follows: temporal normalization is used to eliminate time differences between data sources, and the time freshness factor is used to quantify the reliability of the normalized data. The two are not calculated repeatedly.

[0056] Finally, the geometric precision constraints and time freshness factor calculations for the baseline framework are established. The formula for the geometric precision constraint is:

[0057] ;

[0058] in, This refers to the geometric precision of the fused data, expressed in meters. Indicates the east-west coordinate deviation. Indicates the north-south coordinate deviation. , The deviation between the coordinates of the merged data and the coordinates of the data source with the highest geometric accuracy among those participating in the fusion is defined using the National Coordinate System 2000 as the reference system. This indicator is used to constrain the geometric accuracy of the merged data to not exceed a preset threshold. .when When this happens, the data recalibration process is triggered. Set to half the resolution of the optimal data source in the fused data. Maximum permissible geometric error. The value range is 1.0 to 5.0 meters, and this range is determined based on the resolution of the data source: when the resolution of high-resolution satellite imagery is 2 meters... When the resolution is 10 meters and the multispectral image is 1.0 meter, Use 5.0 meters as the reference, and use linear interpolation for other resolutions.

[0059] After S1 is completed, the output includes the fused time-series benchmark dataset, fusion weight records of each data source, geometric accuracy assessment report, and benchmark framework metadata information. These outputs are then passed as input to S2. The fused time-series benchmark dataset serves as the input data for calculating the temporal difference index in S2. The geometric accuracy assessment report confirms that the data quality meets the requirements for subsequent analysis, and the benchmark framework metadata information provides key parameters such as coordinate system and spatial resolution.

[0060] S2 identifies potential change areas by calculating the temporal difference index and extracts change features by calculating the change intensity quantification value, generating a change detection result layer.

[0061] By constructing a fusion benchmark framework, changes in geographic information data from different time periods are detected, and features such as the type, range, intensity, and direction of changes are extracted and quantitatively quantified.

[0062] S2-1, Calculate the phase difference index to identify potential areas of change. The formula for calculating the phase difference index is:

[0063] ;

[0064] in, This indicates that at coordinate position (x, y), from Time's up The phase difference index at time t is a dimensionless parameter with a value range of [0, +∞). Indicates in The reflectance value of the kth spectral band at (x,y) is the surface reflectance value after atmospheric correction and radiometric calibration. Each band is independently normalized to the range of [0,1]. The normalization method is minimum-maximum normalization to ensure that the dimensions of each band are consistent. Indicates in b represents the reflectance value of the k-th spectral band at (x, y) at time k; b represents the number of spectral bands involved in the calculation; x represents the horizontal coordinate in the east-west direction; y represents the horizontal coordinate in the north-south direction. Indicates the initial data collection time; This indicates the time of data collection in the later stages.

[0065] The temporal phase difference index is the average Euclidean distance between two temporal images across all spectral bands used in the calculation. This index comprehensively considers spectral variation information across all bands, effectively reflecting the degree of spectral difference in land cover between two temporal phases. When the index exceeds a preset threshold, it indicates that a change may have occurred at that location. This temporal phase difference index is a fundamental indicator for change detection, providing the initial basis for subsequent change type determination and change intensity quantification. The time series of this index is then used for subsequent trend analysis and prediction.

[0066] Compared to the single-band difference method, the multi-band comprehensive difference index effectively reduces the interference of single-band noise on change detection results and improves the robustness of change identification. For multispectral images, the number of bands b is [4,8], including blue, green, red, and near-infrared bands; for hyperspectral images, the value of b can range from tens to hundreds. The difference index threshold is set as the mean of the difference indices of all pixels plus twice the standard deviation. This threshold setting is based on the principle of statistical significance testing; the mean plus twice the standard deviation covers approximately 95% of the normal fluctuation range. Changes exceeding this range are considered significant changes, which conforms to the common threshold setting conventions in the field of remote sensing change detection.

[0067] S2-2 uses the temporal difference index to calculate the quantitative value of change intensity, providing a quantitative description of the significance of the changed region. The formula for calculating the quantitative value of change intensity is:

[0068] ;

[0069] in, This represents the quantized value of the change intensity at coordinate position (x, y), which is a dimensionless parameter and takes values ​​in the range [0, 100]. This indicates the temporal difference index at that location; This represents the maximum value of the temporal phase difference index for all pixels within the monitoring area, used for normalization; This represents the area of ​​the varying pixels within a neighborhood window centered at (x,y), expressed in square meters. This represents a reference area unit, with the preferred value being 100 times the pixel area of ​​the monitoring area.

[0070] Calculate the phase difference index for all pixels within the monitoring area, and take the maximum value as the result. In practice, to avoid the influence of extreme outliers, the 99.5 quantile of the phase difference index distribution is taken as... When the temporal phase difference index of all pixels within the monitoring area is zero, The value is 0.001, or the change intensity of all pixels can be directly quantized. The value is assigned to zero. This value is recalculated in each monitoring cycle. A reference area unit is taken as 100 times the spatial resolution of the monitoring data. For example, when the spatial resolution is 10 meters, the area of ​​a single pixel is 100 square meters. Take 10,000 square meters. Its function is to standardize the area gain factor to a uniform dimension, ensuring that the intensity of change in data at different resolutions is comparable. The neighborhood window boundary processing rule is that for pixels located at the boundary of the monitoring area, the neighborhood window takes the intersection with the monitoring area. Calculated based on the proportion of the actual effective neighborhood area.

[0071] The quantification of change intensity is determined by two factors: the first is the normalized temporal phase difference index, reflecting the relative intensity of spectral changes; the second is the area logarithmic gain factor, reflecting the spatial clustering scale of the change. Larger, contiguous areas of change result in higher quantified change intensity values. The use of the logarithmic function ensures that large areas of change do not lead to excessively high intensity values ​​that could cause distortion. The quantified change intensity value provides crucial input for S4 compliance assessment; high-intensity change areas require priority for compliance review. Simultaneously, the change intensity value is also an important basis for S5-level early warning.

[0072] The neighborhood window size is set according to the monitoring scale, ranging from 3×3 to 15×15 pixel windows. This range is determined based on the spatial autocorrelation scale: 3×3 is suitable for fine-scale monitoring, 15×15 is suitable for macro-scale monitoring, and the intermediate value is selected based on the spatial continuity characteristics of land cover types. Reference area. The value is set to 100 times the pixel area to ensure that the parameter of the logarithmic function is greater than 0.

[0073] S2-3, Extract the directional features of change to determine the development trend of the change. The formula for the directional features of change is:

[0074] ;

[0075] In the formula ; .

[0076] Where atan2 is the arctangent function in the fourth quadrant, in degrees, with a range of (-180, 180]. When the calculation result is negative, add 360° to convert it to the range of [0, 360]. This represents the direction of change at coordinate position (x, y) in degrees, with a value range of [0, 360]. This represents the change in reflectivity of the k-th band, i.e., the difference between the later and earlier periods. This represents the weighting coefficient for the k-th band, which is set based on the band's sensitivity to changes in ground features. π represents the mathematical constant pi, with a value of 3.14159; K represents the set of bands sensitive to changes, and J represents the set of stable reference bands. The division of sets K and J is based on the dominant land cover type in the monitoring area and is determined by ranking the variance of each band in historical sample data: the bands with the largest variance are ranked first... One band is assigned to K, and the rest to J, where b is the total number of bands. Indicates rounding up; This represents the smallest positive number that is not divisible by zero, and its value is 0.0001.

[0077] The numerator is the sum of the weighted changes of all bands in the set of change-sensitive bands K, and the denominator is the sum of the weighted changes of all bands in the set of stable reference bands J plus a very small positive number. Because sets K and J contain different bands, the mathematical forms of the numerator and denominator differ. Therefore, the result of the atan2 function accurately reflects the directional characteristics of change, rather than being a constant 45 degrees. The angle of change reflects the dominant direction of land cover transformation from one type to another. For example, the transformation from vegetation to built-up land exhibits a specific angle along the normalized vegetation index (NDI) direction. By mapping multidimensional band changes to angle values ​​using the atan2 function, a quantitative expression of the direction of change is achieved. Using a change-sensitive band in the numerator amplifies the change signal, while using a stable reference band in the denominator provides a stable benchmark. The ratio of these two accurately characterizes the directional features of the change. This formula reduces the dimensionality of multidimensional spectral space changes to intuitive directional angles, facilitating subsequent analysis and visualization.

[0078] The set of wavelength bands K sensitive to vegetation changes includes near-infrared and short-wave infrared bands, with weights... Determined based on the reciprocal of the band variance; the reference band set J includes stable green light bands.

[0079] Finally, based on the above calculation results, a change detection result layer is output, including a changed / unchanged binary layer, a change intensity distribution layer, and a change direction angle layer. The changed / unchanged binary layer uses a change indicator function. The intensity distribution layer uses CI as the pixel value, and the direction angle layer uses CI as the pixel value. These are pixel values; the spatial resolution and coordinate range of each layer are consistent with the input image. The above output results are passed to S3 for trend prediction and anomaly identification. Specifically, the quantized value of change intensity (CI) is input into the S3's rate of change calculation formula and trend prediction model, along with the change direction and angle. Entering the S4 change type compliance determination, the binary layer is used to define the spatial range of the change area.

[0080] S3, based on the change feature data extracted in S2, establishes a time series analysis model to quantitatively predict the change trend of geographic information, identifies change areas that deviate from the normal change pattern, and outputs a change trend prediction layer and anomaly area layer.

[0081] S3-1, Construct a time series of the rate of change, and establish the formula for calculating the rate of change:

[0082] ;

[0083] in, This indicates that the coordinate position (x, y) is in The rate of change over time, in months; This indicates that the location is in Quantitative value of the intensity of change at any given time; This represents the duration of the m-th monitoring period, i.e. minus The unit is month, and .in This represents the m-th monitoring time, where m is an integer ≥ 1. hour, To establish a baseline for monitoring, a fixed time period prior to the first monitoring session is used; one month is used for areas of rapid change, and three months is used for areas of slow change. For the initial data, the rate of change is calculated as the ratio of the initial change intensity to the base period duration. The rate of change describes the cumulative speed of change intensity per unit time. A higher rate of change indicates a more drastic change process in the region. By calculating the rate of change over multiple consecutive monitoring periods, a time series of the rate of change can be constructed to analyze the acceleration or deceleration trend of the change process. The rate of change is the core input variable of the trend prediction model. Subsequent calculations in S3 will establish a trend equation and predict future changes based on the time series of the rate of change; S4 will use the prediction results for a forward-looking compliance assessment. Monitoring Period Based on the monitoring targets, the monitoring period is 1 to 3 months for areas of rapid change and 6 to 12 months for other areas; the rate of change is measured in monthly units.

[0084] S3-2, Based on the time series of change rates, a trend prediction model is established to predict the intensity of change at future times. The formula for the trend prediction model is:

[0085] ;

[0086] in, This indicates the coordinate position (x, y) at the prediction time. The predicted change intensity value is a dimensionless parameter; Indicates the current time The quantified value of the intensity of change at this location; Indicates the rate of change at the current moment; This represents the time interval from the current moment to the predicted moment, in months. This indicates the average intensity of change at that location during the same historical period, obtained by analyzing historical monitoring data from at least three years of the same period. This represents the weight coefficient of the current state, with a value range of [0.3, 0.6]. This represents the trend extrapolation weighting coefficient, with a value range of [0.2, 0.5]. This represents the weighting coefficient for the historical period, with a value range of [0.1, 0.3]. The three weighting coefficients satisfy... The constraints.

[0087] Based on the dominant factors of land cover change in the monitoring area, parameters are selected from a preset set. For areas dominated by human activities, Take 0.4, Take 0.4, Take 0.2; for areas dominated by the natural environment, Take 0.3, Take 0.3, Take 0.4; for mixed-type regions, Take 0.35, Take 0.35, The value is set to 0.3. The aforementioned preset parameter set was determined through time-series cross-validation of historical monitoring data: the historical monitoring data was divided into training and validation sets according to time series, with the objective function being the minimization of the root mean square error of prediction, under the constraints... Within their respective value ranges, the optimal weight combination is solved using the grid search method; changes in areas dominated by human activities are driven by short-term policies, and the current state and trend extrapolation have higher weights; changes in areas dominated by the natural environment are affected by seasonal patterns, and the historical same period has higher weights; for mixed-type areas, the equilibrium value of the three is taken.

[0088] The predicted intensity of future change is synthesized from three contributions. The first part is the continuation contribution of the current state of change, with a weight of [weight missing]. The second part is the trend extrapolation contribution based on the current rate of change, with a weight of [weight missing]. Assuming that change continues at the current rate, the third part represents the periodic contribution of historical change patterns, with a weight of [weight missing]. This formula reflects the seasonal or periodic patterns of changes in land features. It comprehensively considers information from three dimensions: current state, development trend, and historical patterns, thus improving the accuracy and reliability of change prediction.

[0089] S3-3, Establish an anomaly identification model to identify areas that deviate from the normal trend. The formula for calculating the anomaly index is:

[0090] ;

[0091] This formula is applicable to two scenarios:

[0092] (1) Threshold training phase: Calculate the anomaly index using complete historical monitoring data, including... Actual observation value at time The 95th percentile was used as the statistical measure. The baseline value;

[0093] (2) Real-time monitoring phase: at the current moment Predicting the future based on trend forecasting formula Intensity of change at time Pending actual acquisition Observations at time Then, calculate the anomaly index and compare it with... The comparison is performed, and if the threshold is exceeded, a backtracking anomaly warning is triggered. For scenarios requiring immediate warnings, a simplified prediction model is used. That is, assuming the intensity of future changes is the same as at the present moment, then... , , .

[0094] in, The anomaly index represents the coordinate position (x, y), is a dimensionless parameter, and takes values ​​in the range [0, +∞). This indicates the predicted intensity of change; Indicates in The actual observed intensity of change at any given moment; The standard deviation of the historical change intensity at that location is used to standardize the prediction bias. When the value is zero, the mean of the standard deviation of the historical change intensity within the neighborhood window of that location is taken as the replacement, or 0.001 is taken; This represents the spatial neighborhood anomalous gain coefficient, with a value range of [0.5, 2.0]. This represents a spatial neighborhood anomalous correlation function.

[0095] The complete definition is: ;

[0096] in, This represents the spatial neighborhood anomaly correlation function, with a value range of [0,1]. The closer the value is to 1, the higher the degree of anomaly clustering within the neighborhood. It represents the set of all cell locations within a neighborhood window centered at (x,y); This indicates the total number of cells within the neighboring window; Indicates an indicator function, when The value is 1 if the condition is met, and 0 otherwise, where k takes the values ​​1.5 or 2.0. This represents the quantized value of the intensity of the change in UV at a location. This indicates the average intensity of change in UV at a given location over the same historical monitoring period.

[0097] The anomaly index consists of two parts. The first part is the absolute value of the standardized prediction bias, reflecting the degree of deviation between the actual change and the predicted trend, standardized using historical standard deviations. The second part is the spatial neighborhood anomaly correlation gain, reflecting the spatial clustering characteristics of anomalous changes. Spatial Neighborhood Anomaly Correlation Function The spatial clustering of anomalous changes is quantified by statistically analyzing the proportion of pixels within a neighborhood window whose change intensity exceeds the historical average for the same period. When anomalous changes also occur in surrounding areas, the anomalous index of the current area will further increase, which aligns with the objective law that changes in human activities often occur in concentrated clusters. The anomalous index is the core basis for the S5-level early warning system. Areas with high anomalous indices will trigger higher-level warnings. In compliance assessments, S4 will prioritize the review of areas with high anomalousness. This formula achieves quantitative identification of anomalous change events, considering both prediction bias in the temporal dimension and clustering characteristics in the spatial dimension.

[0098] Historical standard deviation Calculated using variation intensity data from at least 3 years of the same period; the spatial neighborhood range is a 5×5 pixel window; the threshold is 1.5 times the standard deviation of the historical average for the same period; spatial neighborhood anomaly gain coefficient. Set the threshold to 1.0. Use the 95th quantile of the anomaly index in historical monitoring data for the same period as the threshold. Specifically, collect at least three years of historical monitoring data for the same period, calculate the anomaly index for each period, and use the 95th quantile of the distribution of these anomaly indices as the threshold. The baseline value is set at 2.0. If the historical data is less than 3 years, the default value of 2.0 is used, meaning that an anomaly is determined when the prediction deviation exceeds twice the historical standard deviation. This threshold can be manually adjusted according to the actual management needs of the monitoring area. In areas requiring strict monitoring, the threshold can be appropriately reduced to 1.5, while in general monitoring areas, it can be maintained at 2.0.

[0099] S3-4, combining the trend prediction results and anomaly identification results, outputs a trend prediction layer and an anomaly region identification layer. The trend prediction layer uses... The anomaly index (AI) is used as the pixel value. The anomaly detection layer uses AI as the pixel value, and the spatial resolution and coordinate range of each layer are consistent with the input image. When the anomaly index (AI) exceeds a preset threshold... At this point, the location is marked as an area of ​​abnormal change and enters the priority compliance review process in S4. The above output is passed to S4, specifically conveying the predicted intensity of change. The AI-generated anomaly index serves as input data for S4 compliance assessment, the anomaly area identification layer is used to determine the spatial scope for priority review, and the change rate time series is used to assist in judging the urgency of changes.

[0100] S4 performs spatial overlay comparison and difference analysis on the change detection result layer identified by S2 and S3 and the trend prediction layer to automatically determine the change compliance status result.

[0101] S4-1, the spatial boundary data of the legally constrained areas is obtained from the latest version of the territorial spatial planning data released by the natural resources authorities. This includes vector layers of ecological protection red lines, permanent basic farmland, urban development boundaries, and water source protection areas. The data format is Shapefile or GeoDatabase, and the plane coordinate system is consistent with the National 2000 coordinate system. The data update frequency is synchronized with the territorial spatial planning adjustment cycle, generally once a year, with the latest version used during planning adjustments. When there is spatial overlap between multi-level legally constrained areas, the strictest constraint rule is applied, i.e., the overlay indicator function for overlapping areas. The value is 1, and the weight is... Take the maximum value from the overlap rules; this is the spatial overlay composite index at that location. Calculated based on the weight of the single rule that is actually in effect.

[0102] S4-2, Establish a spatial overlay analysis model to determine the spatial relationship between the changing region and the regulatory constraint region. The spatial overlay analysis formula is:

[0103] ;

[0104] in, The spatial superposition composite index represents the coordinate position (x,y), is a dimensionless parameter, and takes values ​​in the range [0,+∞). This represents the weight coefficient of the r-th rule, with a value range of [0,1]. The weight is determined according to the level of legal force of the rule. Let represent the superposition indicator function of the r-th rule at location (x, y). If the geometric center of pixel (x, y) lies within the constraint region of the r-th rule, or the area of ​​the intersection between the pixel and the constraint region exceeds 50%, then... This indicates that the change is subject to Article r; otherwise... This indicates that the change is not subject to Article r. This indicates the total number of rules and regulations that govern legal constraints.

[0105] The legally constrained areas include ecological protection red line areas, permanent basic farmland areas, urban development boundary areas, and water source protection areas, etc. The spatial extent of these areas is determined by vector boundary data. (Overlay indicator function) The spatial overlay analysis shows that if the geometric center of the pixel (x,y) is located within the r-th legal constraint region, or if the area ratio of the pixel to the constraint region exceeds 50%, the indicator function is set to 1; otherwise, it is 0.

[0106] The method for determining it is as follows: The spatial overlay comprehensive index is the sum of the weights of all regulatory constraints, and this value is the theoretical maximum. It is the result of spatially overlaying all regulatory constraints at a specific location. Each rule is assigned a different weight according to its importance; when the area of ​​change overlaps with the constraint area of ​​a rule, that rule contributes its weight value. The larger the comprehensive index, the stricter the regulatory constraints faced in that area, and the higher the compliance risk of the change. The spatial overlay comprehensive index is the core input for compliance scoring. This index transforms complex, multi-layered regulatory constraints into a single quantitative indicator, facilitating subsequent processing. S5 will adjust the warning level based on this index. This formula achieves a unified quantitative expression of multi-layered regulatory constraints, supporting differentiated processing of regulations at different levels of effectiveness.

[0107] The regulations and rules governing these areas include ecological protection red lines, permanent basic farmland, urban development boundaries, and water source protection zones, and their number... The number of ecological protection red lines ranges from 4 to 10. Take 1.0, permanent basic farmland 0.9, urban development boundary 0.7, and other constraints 0.5.

[0108] S4-3, Establish a comprehensive compliance scoring model to quantitatively assess the compliance of each change. The compliance scoring formula is:

[0109] ;

[0110] in, The compliance score represents the coordinate position (x, y), with a value range of [0, 100]. The closer the value is to 100, the better the compliance. This indicates the spatial overlay composite index at that location; This represents the theoretical maximum value of the spatial superposition comprehensive index. That is, the sum of the weights of all regulatory constraints, used to normalize the spatial superposition comprehensive index of each location to the range of [0,1]. This represents the quantitative value of the change intensity at this location; κ represents the adjustment coefficient for the degree of impact of the change intensity on compliance, with a value range of [0.5, 2.0].

[0111] The method for determining κ is based on setting an adjustment coefficient according to the management strictness of the monitoring area. For Level 1 strictly controlled areas, such as national nature reserves and core drinking water source areas, κ is set to 2.0, where the decay of compliance scores due to high-intensity changes accelerates. For Level 2 controlled areas, such as provincial ecological function zones, κ is set to 1.5. For general controlled areas, κ is set to 1.0, where the intensity of change has a linear impact on the compliance score. For areas with relaxed management, κ is set to 0.5. The specific value of κ is determined by the regional environmental functional zoning level and is automatically matched according to the functional zoning of the monitoring area during system initialization. The compliance score is composed of the product of two factors. The first factor is the regulatory constraint avoidance factor, reflecting the degree to which the changing area avoids regulatory constraints. The smaller the spatial superposition comprehensive index, the closer this factor is to 1, indicating that the changing area is less subject to regulatory constraints. The second factor is the change intensity adjustment factor, reflecting the impact of change intensity on compliance. The greater the change intensity, the smaller this factor, indicating a higher compliance risk from high-intensity changes. The existence of the index κ makes the decay of compliance scores due to high-intensity changes exhibit a non-linear characteristic. The compliance score serves as the direct basis for S5 visualization and early warning issuance. Areas with scores below the threshold will be prominently displayed and trigger early warnings. This formula unifies the two dimensions of regulatory constraint strength and change intensity into a single compliance score, achieving automation and quantification of compliance determination.

[0112] The adjustment coefficient κ is set based on the strictness of management in the monitoring area; κ is 2.0 for Level 1 control areas and 1.0 for general areas. The compliance score threshold ranges from 40 to 60 points, with 60 points corresponding to the basic compliance line, and anything below this value is considered non-compliant; 40 points corresponds to the attention line, and anything between 40 and 60 points is considered to require attention. This threshold is determined based on land and space management practices, taking into account both monitoring sensitivity and enforcement feasibility.

[0113] Establish rules for determining the compliance of change types, judging the compliance of change types based on the characteristics of the direction of change. The formula for change type compliance is:

[0114] ;

[0115] in, The compliance index represents the type of change in coordinate position (x, y), which is a dimensionless parameter with a value range of (-∞, +∞). This represents the probability that the position belongs to the c-th type of change, obtained by combining the change direction feature of S2 with the classification model; The compliance coefficient represents the c-th type of change, and its value ranges from [-1, +1]. A positive value indicates a positive compliant change, and a negative value indicates a non-compliant change. This indicates the total number of variation types.

[0116] The method for determining it is as follows: the angle of change calculated by S2. Angle similarity matching is performed with a pre-established change type direction feature library. This change type direction feature library extracts typical direction angles for each change type by statistically clustering the spectral change vectors of verified change samples in historical monitoring periods. A pre-established library of directional features for each variation type stores the typical directional angles of each variation type in the multidimensional spectral space. Calculate cosine similarity. ,in The typical direction angle for the c-th type of change is determined by the average direction angle of this type of change in the historical sample database; it is then normalized using the softmax function. The probability distribution of each change type is then obtained. The change type with the highest probability is the dominant change type at that location.

[0117] In accordance with relevant national laws, regulations, and policy documents, a compliance coefficient is set for each type of change. Positive ecological changes such as vegetation increase and water body restoration are included. Positive values ​​are taken, ranging from [0.3, 0.9]; negative changes such as the expansion of construction land encroaching on cultivated land and the destruction of vegetation are also considered. Negative values ​​are taken, ranging from [-0.9, 0.3]; neutral changes include seasonal vegetation cover fluctuations. The value is set to 0. The specific value is determined based on the severity of the violation of the change type; the more severe the violation, the higher the value. The larger the absolute value, the greater the compliance index. The change type compliance index is the probability-weighted average of the compliance coefficients for each change type. When the probability distribution of a change type is biased towards compliance, the index is positive. When it is biased towards non-compliance, the index is negative. The absolute value of this index reflects the confidence level of the compliance judgment. The change type compliance index provides another dimension of compliance judgment besides spatial overlay analysis; the judgment results from both dimensions will be used comprehensively. S5 will issue a warning based on the combined judgment results from both dimensions.

[0118] Types of change include vegetation increase, vegetation decrease, expansion of construction land, and changes in water areas. Vegetation increase... We use 0.8 for vegetation reduction, -0.6 for construction land expansion, -0.9 for water area change, and 0.3 for change type probability. It is obtained by combining the changing direction and angle with a historical sample database through angle similarity matching.

[0119] S4-4, combining the overall compliance score and the change type compliance index, outputs the compliance judgment result for each change, including the compliance score, change type compliance index, list of main regulatory constraints, and compliance judgment conclusion. The judgment conclusion is divided into three levels: compliant, requiring attention, and non-compliant. and It is deemed compliant at the time; when and ,or It is determined to require attention at this time; when or If the condition is deemed non-compliant, the result is transmitted to S5 for visualization and alerts. Specifically, the transmitted content includes: compliance score. and compliance index of change type As direct inputs to the S5 warning level calculation and visual mapping, the judgment conclusion level is used to determine the priority of warning issuance and response process, while the list of main regulatory constraints is used to generate detailed content of the warning notification.

[0120] S5. Establish a mapping model from change intensity to visual variables, including hue, symbol size, and transparency, to intuitively express change information. The visual mapping formula is:

[0121] ;

[0122] in, This represents the hue value of the coordinate position (x, y) in the visualization, with a value range of [0, 360]. This represents the base hue value, preferably 0 degrees or 240 degrees; This represents the quantified value of the intensity of change at that location; This represents the maximum hue shift, preferably 60 degrees. When the change intensity is 0, the hue is taken as the baseline value. When the intensity of the change reaches its maximum value of 100, the hue shift reaches [a certain value]. .

[0123] Tone mapping uses continuous hues to represent the intensity of change, and warning levels are distinguished by superimposed symbol size or transparency, forming a dual visual channel: a hue gradient from blue to magenta represents an increase in intensity, symbol size from small to large represents warning levels from 1 to 3, and transparency from semi-transparent to opaque represents urgency from low to high. The hue value shifts continuously with increasing intensity, making areas with higher intensity changes more visually prominent. This continuous mapping method avoids the step effect in hierarchical representation and can finely present the intensity gradient of change. The visual mapping result is the foundation of the final visualization product, directly affecting the decision-maker's intuitive understanding of the spatial distribution of change. This formula achieves a continuous mapping from intensity to hue, making the spatial distribution pattern of change immediately apparent. Baseline hue. 240 degrees is used to represent the blue color scheme. A 60-degree angle represents a shift towards magenta, creating a continuous gradient from blue to red, with blue indicating a low intensity of change and red indicating a high intensity of change.

[0124] S5-1 triggers a tiered early warning model, determining the warning level based on the anomaly index and compliance score. The tiered early warning model is as follows:

[0125] ;

[0126] in, The warning level indicates the coordinate position (x, y), and the value is a positive integer of 1, 2 or 3; This indicates the anomaly index at that location; This represents the threshold for the abnormality index, with a preferred value of 2.0. This represents the compliance score for that position; the round-up function ⌈x⌉ represents the smallest integer not less than x, ensuring that the calculation result is at least 1; the min function means that when the calculation result is greater than 3, it is uniformly taken as 3.

[0127] The warning level is obtained by adding two parts. The first part is the anomaly level component; the higher the anomaly index, the higher this component. The second part is the compliance risk component; the lower the compliance score, the higher this component. The two parts are added together and rounded up to obtain the final warning level. Warning level 1 represents a low warning, 2 represents a medium warning, and 3 represents a high warning. Anomaly index threshold. The threshold is 2.0. When the anomaly index is 0 and the compliance score is 100, the warning level is 1. When the anomaly index is 4 and the compliance score is 0, the warning level is 3.

[0128] S5-2, Establish a dynamic monitoring report generation model to automatically summarize the statistical analysis results of all changes within the monitoring area. The statistical summary formula is:

[0129] ;

[0130] in, This indicates the percentage of the area of ​​change within the monitored region. Indicates a change indicator function, when The value is 1 when the pixel is determined to have changed, and 0 otherwise. That is, the indicator function takes a value of 1 when the pixel is determined to have changed, and a value of 0 when the pixel has not changed. The minimum detectable threshold representing the intensity of the change; This represents the total number of pixels within the monitoring area. The summation range covers all pixel locations within the entire monitoring area. The setting references the normalization mechanism of the change intensity quantization formula, taking the quantized values ​​of change intensity of all pixels within the monitoring area. The 10th percentile is used as the minimum detectable threshold to ensure the exclusion of spurious changes caused by noise; for different land cover types, separate thresholds can be set. .

[0131] This formula provides a macroscopic statistical indicator of the overall degree of change in the monitored area. The percentage of the changed area is the ratio of the number of changed pixels to the total number of pixels. Only pixels whose change intensity exceeds the minimum detectable threshold are included in the changed area, thus excluding minor spurious changes caused by noise.

[0132] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for monitoring dynamic changes in geographic information based on spatial big data, characterized in that, Includes the following steps: S1. Collect multi-source spatial big data and perform quality assessment and weighted fusion to construct a unified benchmark framework data; S2. Potential change areas are identified by calculating the temporal difference index and change features are extracted by calculating the change intensity quantification value, generating a change detection result layer; S3. Construct a time series of change rates and establish a trend prediction model, outputting a change trend prediction layer and an anomaly region layer; S4. Spatially overlay and compare the identified change detection result layer with the trend prediction layer and perform difference analysis to conduct a comprehensive compliance score and output the change compliance judgment result. S5. Establish a mapping model from change intensity to visual variables, including hue, symbol size, and transparency, and trigger a graded early warning model to generate a monitoring report and issue early warning information.

2. The method according to claim 1, characterized in that, The multi-source spatial big data in S1 includes time-series satellite remote sensing images, aerial photogrammetry data, ground sensor network data, and vector geographic boundary data.

3. The method according to claim 1, characterized in that, The quality assessment and weighted fusion in S1 includes verifying the geometric precision constraints and calculating the time freshness factor for each data source, determining the fusion weight of each data source based on the verification results, and generating a unified benchmark framework data with geometric precision constraints and time freshness factor calculation through weighted fusion.

4. The method according to claim 1, characterized in that, The step S2 involves extracting change features, including calculating a quantitative value of change intensity to describe the significance of the change, and extracting change direction features to determine the type of change. The change intensity and change direction are combined to generate a change detection result layer.

5. The method according to claim 1, characterized in that, The prediction model established in S3 includes establishing a trend prediction model based on historical change rate time series, predicting the intensity of change in future periods by analyzing the time evolution law of the change rate, and outputting a change trend prediction layer.

6. The method according to claim 1, characterized in that, The abnormal region layer in S3 is obtained by establishing an abnormality identification model. This model identifies regions that deviate from the normal trend as abnormal regions based on the normal fluctuation range of the change rate time series and outputs the abnormal region layer.

7. The method according to claim 1, characterized in that, The comprehensive compliance score in S4 includes a quantitative assessment of the compliance level of the change based on the spatial overlap between the change area and the regulatory constraint area, combined with the compliance judgment rules for the change type, and generates a compliance judgment result that includes a compliance score and the judgment basis.

8. The method according to claim 1, characterized in that, In step S5, mapping the intensity of change to visual variables includes hierarchical mapping of color depth and symbol size based on the magnitude of the intensity of change, so that the significance of the change is presented intuitively through visual variables.

9. The method according to claim 1, characterized in that, The S5 triggering of graded early warning includes dividing the intensity of change into multiple early warning levels based on preset thresholds, determining the urgency and release priority of early warning information according to the early warning level, and automatically releasing early warning information through multiple channels.

10. A computer system, characterized in that, include: processor; Memory used to store instructions executed by the processor; The processor is configured to implement the method according to any one of claims 1 to 9 when executing the instructions.