Comprehensive investigation and cause mechanism analysis method for ecological geological conditions in medium-high mountain and rocky desertification area

Abstract

The invention discloses a comprehensive investigation and cause mechanism analysis method for ecological geological conditions in a medium-high mountain stony desertification area, and belongs to the technical field of ecological environment monitoring. Comprising the following steps: fusing multi-source remote sensing with historical geology, soil and climate data, and constructing a regional ecological geology factor data set; space clustering and terrain analysis are adopted to optimize sample plot layout, and multi-dimensional ecological geological parameters such as lithology, soil and vegetation are collected on site; quantifying human activity intensity by combining multi-temporal remote sensing and social economic data, and establishing a human activity-ecological disturbance-geological response dynamic model; comprehensively decomposing the quantitative contribution and interaction of each dynamic factor by utilizing geostatistics, geographically weighted regression and a structural equation model, and judging the stony desertification cause; and based on a GIS platform, reconstructing a stony desertification space-time expansion process, concluding main control factors and forming a discrimination index and a prediction model. The method can scientifically reveal a master control mechanism of medium-high mountain stony desertification, and provides a scientific basis for ecological management and risk early warning of a stony desertification area.

Description

Technical Field

[0001] This invention belongs to the field of ecological environment monitoring technology, and more specifically relates to a method for comprehensive investigation and analysis of the formation mechanism of ecological geological conditions in mid-to-high mountain rocky desertification areas. Background Technology

[0002] Rocky desertification is a typical ecological and environmental problem in karst regions, characterized by the destruction of surface vegetation, soil erosion, and the expansion of bare rock areas, posing a serious threat to regional ecological security and sustainable development. Traditional methods for monitoring and analyzing the causes of rocky desertification mainly rely on field surveys and static remote sensing interpretation, which are insufficient to fully characterize the spatial heterogeneity and dynamic evolution mechanisms of the rocky desertification process, and also lack sufficient analysis of the spatiotemporal impacts of human activities.

[0003] With advancements in spatial information technologies such as remote sensing and Geographic Information Systems (GIS), the ability to acquire and process multi-source high-resolution data has significantly improved, providing new avenues for dynamic monitoring, quantitative analysis of driving mechanisms, and process simulation of rocky desertification. Simultaneously, the integrated application of methods such as geostatistical analysis, spatial regression, structural equation modeling, and spatial simulation helps to reveal the coupling mechanisms between natural ecological and geological elements and human disturbances in the formation and evolution of rocky desertification, improving the accuracy of identifying key controlling factors and enhancing predictive capabilities.

[0004] However, existing technologies still suffer from problems such as low integration, insufficient adaptability to spatial scale, and limited spatiotemporal analysis granularity in multi-source data integration, dynamic mechanism analysis, and simulation prediction. There is an urgent need to develop a new technology system for dynamic modeling and risk assessment of rocky desertification that is oriented towards multiple factors and multiple scales, so as to support the decision-making needs of scientific governance and ecological restoration in rocky desertification areas. Summary of the Invention

[0005] This invention aims to address the technical problems existing in rocky desertification monitoring and process simulation, such as insufficient multi-source data integration capabilities, unclear dynamic factor coupling mechanisms, and limited accuracy in spatiotemporal dynamic reconstruction and risk assessment. It aims to achieve accurate simulation of the spatiotemporal evolution process of the ecological geological system in rocky desertification areas under multi-factor driving and identification of the main control mechanisms, providing scientific and quantitative technical support for regional rocky desertification prevention and ecological restoration.

[0006] To achieve the above objectives, the present invention employs the following technical solution: the method comprises: Collect multi-source remote sensing data and historical geological, soil and climate data of the high mountain area in the target area, and establish a regional ecological geological factor dataset. Spatial clustering and topographic analysis methods were used to select and deploy sample plots for representative ecogeological units with different lithology, slope aspect, and vegetation cover. At the selected sample plot locations, ecological geological parameters such as lithology, weathering degree, soil thickness, water content, and vegetation structure are obtained to form a multidimensional raw data matrix; By combining multi-temporal remote sensing, ground surveys and socio-economic data, the intensity of human activities in the region is dynamically assessed, and a dynamic relationship model of human activities, ecological disturbances and geological responses is established. By pooling the aforementioned natural ecological geological parameters and human activity disturbance data from multiple sources, and based on geostatistics, geographical weighted regression, and structural equation modeling methods, the quantitative contributions and interactions of various dynamic factors are comprehensively quantified to determine the causes of rocky desertification. Using a GIS platform, inputting dynamic factor models, and combining historical time-series data, we reconstruct the spatiotemporal expansion process of rocky desertification, identify and summarize its main controlling ecological and geological formation mechanisms, and output a targeted discrimination index system and predictive model.

[0007] In one scheme, the establishment of a regional ecological geological factor dataset includes: spatial registration and rasterization fusion with field survey data of geology, soil, and climate through radiometric correction, atmospheric correction, geometric correction, and band registration; calculation of ecological geological factors such as surface exposure NDVI, soil moisture, slope, lithology type, and annual precipitation; and completion of ecological geological condition classification and rocky desertification risk prediction using the random forest method.

[0008] In one approach, the spatial clustering and topographic analysis method is based on the distribution data of ecological geological factors. It uses lithology, slope aspect, vegetation cover, soil thickness, and surface bareness index as multidimensional features. The spatial clustering algorithm is used to unsupervised group the pixels in the study area. The sample plots are selected by combining topographic analysis and accessibility constraints, and the maximum spatial dispersion optimization is adopted.

[0009] In one approach, the multidimensional raw data matrix is ​​used to set up standardized survey sample areas on-site. Soil thickness is determined by geological drilling. Soil moisture content is measured using time domain reflectance (TDR) or a portable rapid soil analyzer. Canopy coverage and average vegetation height are obtained by using a plant canopy meter or drone photography. Soil and vegetation samples are collected for laboratory physicochemical and physiological property analysis. All sample plot parameters are recorded with GPS information of the center point.

[0010] In one approach, a dynamic relationship model of human activity, ecological disturbance, and geological response is established. This model utilizes multi-temporal remote sensing imagery, ground surveys, and socio-economic statistics to dynamically monitor the intensity of human activities and ecological disturbances. It quantifies land use / cover types and remote sensing change indices, and combines ground survey information with socio-statistical data to model the spatial and temporal distribution of human activity intensity. Furthermore, through dynamic monitoring of ecological disturbance factors and a three-dimensional time-series model of human activity, ecological disturbance, and geological response, the geological environmental response in rocky desertification areas is quantitatively analyzed.

[0011] In one approach, the determination of the causes of rocky desertification includes: spatially registering and scaling natural ecological geological parameters and human activity disturbance data to form a multi-source data raster matrix; using a semivariance function to analyze the spatial structure; and using a geographic weighted regression model to reveal the spatial heterogeneity contribution of each dynamic factor. We introduce structural equation modeling to quantitatively decompose the direct, indirect, and interactive effects between natural factors, human disturbance factors, and rocky desertification response, and analyze the coupling mechanism of the rocky desertification dynamic system.

[0012] In one scheme, the reconstruction of the spatiotemporal expansion process of rocky desertification includes: using a GIS platform, inputting multi-factor weights and spatial distribution parameters, combining remote sensing inversion of rocky desertification distribution and ecological, geological, and socio-economic time-series data, constructing a grid cell state space for rocky desertification, dynamically reconstructing the spatiotemporal process of rocky desertification using spatial regression and simulation methods, and identifying the main controlling factors through principal component analysis and sensitivity analysis, and establishing a rocky desertification discrimination index system and predictive model.

[0013] Beneficial effects of this invention: This invention integrates multi-source remote sensing and spatial big data to establish a comprehensive analysis and modeling method for human activity disturbance and dynamic response of ecological geological systems in rocky desertification areas, effectively improving the accuracy of spatiotemporal dynamic simulation and risk assessment of rocky desertification processes.

[0014] Compared to traditional methods, this invention can more comprehensively reveal the coupling relationship between natural environmental elements and human activities, accurately identify the main controlling factors and high-risk areas of rocky desertification, and provide solid data and model support for the scientific formulation of ecological environment governance and rocky desertification prevention and control measures. Furthermore, the method of this invention has good adaptability and scalability, providing a reference for the dynamic monitoring and management of similar complex ecological geological systems. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method of the present invention; Figure 2 Flowchart for establishing a regional ecological geological factor dataset for this invention; Figure 3 This is a flowchart illustrating the selection and layout of sample plots for representative ecogeological units in this invention. Figure 4 A flowchart for establishing a dynamic relationship model of human activities, ecological disturbances, and geological responses for this invention; Figure 5 The flowchart illustrates the multi-scale coupling and weighted discrimination process for the causes of rocky desertification in this invention. Detailed Implementation

[0016] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0017] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. To facilitate understanding, the invention will now be described more fully with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to make the disclosure of the invention more thorough and complete.

[0018] like Figure 1 As shown, a comprehensive investigation and analysis of the formation mechanism of ecological geological conditions in mid-to-high mountain rocky desertification areas is presented, with the following specific steps: like Figure 2 As shown, step 1: Collect multi-source remote sensing data (high-resolution optical images, radar images) and historical geological, soil and climate data of the high mountain area in the target. Establish a regional ecological geological factor dataset through a classification model based on machine learning to provide quantitative support for subsequent site selection and systematic investigation.

[0019] S101. For the target area in the mid-to-high mountain rocky desertification zone, acquire multi-source remote sensing data, including high-resolution optical imagery and synthetic aperture radar (SAR) imagery. After preprocessing (including radiometric correction, atmospheric correction, geometric correction, and band registration), this remote sensing data is spatially registered and fused with historically accumulated field survey data on geology, soil, and climate. The data fusion process uses rasterization to unify various factors to the same spatial resolution and geographic reference system, facilitating subsequent data analysis.

[0020] S102. After integrating multi-source data, extract or calculate the characterization indicators for each pixel based on the main ecological and geological factors related to rocky desertification (NDVI, soil moisture, slope, lithology, and annual precipitation). The Normalized Difference Vegetation Index (NDVI) is calculated using remote sensing imagery; its mathematical expression is: in, The reflectivity is in the near-infrared band. This refers to the reflectivity in the red light band. Soil moisture can be measured using the radar backscattering coefficient. A relevant inversion is performed. The terrain slope is then obtained from the digital elevation model (DEM) through spatial difference calculation: in, and These represent the rates of change of elevation in the x and y directions, respectively.

[0021] S103. After extracting all indicators, and combining the attributes of the field survey sample points (vegetation type, measured soil properties, geological type), a supervised machine learning classification method is used for modeling. Training the random forest model involves inputting a feature matrix. Each of them The fused multidimensional feature vector and label vector To differentiate the results of the field survey, Random Forest uses ensemble decision trees to determine the ecological geological type or rocky desertification risk level of each pixel. The principle of the Random Forest model can be described by the following voting formula: in, Let represent the classification result made by the b-th decision tree, where B is the total number of decision trees. This indicates that the majority voting principle determines the final classification output.

[0022] Through the above process, a dataset of ecological and geological factors covering the study area can be obtained, enabling the prediction of ecological and geological conditions related to rocky desertification and providing high-precision, multi-dimensional geoscience basic information for subsequent automated plot layout and systematic investigation.

[0023] like Figure 3 As shown, step 2: Using the factor distribution data obtained in step 1, spatial clustering and terrain analysis methods are adopted, combined with field accessibility, to realize intelligent selection and layout of sample plots for representative ecological geological units with different lithology, slope aspect, vegetation cover, etc., thereby improving survey efficiency and representativeness.

[0024] S201. Based on the distribution data of ecological and geological factors obtained in step 1, the sample plot layout achieves intelligent selection of survey sample plots with high representativeness and high operability through multi-factor spatial optimization. First, factors such as lithology (L), slope aspect (A), vegetation cover (V), soil thickness (S), and surface exposure index (E) are composed into a spatial multi-dimensional feature vector. Each pixel is assigned a value to construct a multi-factor feature space for the entire study area. To ensure the representativeness of the sample plots, a spatial clustering algorithm (k-means clustering) is used to analyze the feature vector set of all pixels in the area. Unsupervised grouping is performed with the objective function of minimizing the sum of squared distances from samples to cluster centers: in, The specified number of clusters (representing the number of ecological and geological types to be divided). Let k be the set of samples in the k-th class. Let be the centroid of the eigenvector of the k-th class. Based on the clustering results, the main ecological and geological units of the region can be systematically divided.

[0025] S202. Subsequently, within each cluster category (i.e., different ecogeological units), sample plots are further optimized using topographic analysis (such as topographic relief, slope length, slope gradient, etc.) and field accessibility constraints. Accessibility can be modeled as cost weights. And combined with the shortest distance between the candidate sample plots and the road Define sampling weights: in, The reachability penalty score for candidate samples. , This is an empirical weighting coefficient. A threshold is set to filter out areas that are too costly or inaccessible, ultimately selecting the most representative and accessible plot locations within each ecological and geological unit category.

[0026] S203. Furthermore, to ensure the uniformity of spatial distribution and the geoscientific representativeness of the survey, the principle of maximum spatial dispersion can be adopted. From the screened candidate sampling points, several points with the largest or smallest spatial distance function correlation can be selected to optimize the layout scheme. Let the final selected sample plot locations be... Its spatial distribution uniformity can be achieved by maximizing the minimum distance: in, Geographical distance.

[0027] Through the integrated process of multi-factor clustering, terrain suitability analysis, and spatial optimization constraints, the intelligent layout of representative ecological and geological plots in mid-to-high mountain rocky desertification areas is ultimately achieved. This method not only improves the accuracy of the coupling between plot distribution and geological / ecological units, but also significantly reduces the cost and risk of actual surveys.

[0028] Step 3: At the selected sample plot location, use multi-parameter field measurement instruments (such as rapid soil analyzer, portable ground-penetrating radar, plant canopy analyzer, etc.) to simultaneously acquire ecological and geological parameters such as lithology, weathering degree, soil thickness, water content, and vegetation structure, and collect necessary soil and vegetation samples for detailed indoor analysis to form a multi-dimensional raw data matrix.

[0029] For the selected sample sites, multi-parameter ecological geological data collection and analysis were conducted simultaneously on-site. First, standardized rectangular or circular survey areas were established, and the coordinates of the center point of each sample site were determined. Based on lithology and weathering degree, portable ground-penetrating radar (GPR) was used to rapidly scan the subsurface profile, acquiring radar reflection signals at different depths. The attenuation of radar echo amplitude with depth can be characterized by the following formula: in, Let z be the echo amplitude at depth z. The initial amplitude of the incident wave, The attenuation coefficient of the medium is used to quantitatively correlate different attenuation changes with stratigraphic structure, weathering intensity, etc., in combination with lithological parameters.

[0030] Soil thickness can be accurately obtained through geological drilling combined with GPR profile verification. Soil moisture content is measured in-situ using time-domain reflectometry (TDR) or a portable rapid soil analyzer; the principle is to analyze the relationship between the reflection time of the probe as it travels through the soil and the moisture content, thus determining the soil volumetric moisture content. Available from: in, The dielectric constant of the soil, ( , () is the calibration coefficient. Multiple points of moisture content were recorded on-site, and the profile moisture distribution was obtained by integrating the soil depth.

[0031] Vegetation structure parameters are obtained through plant canopy meters or drone image interpretation, yielding indicators such as canopy cover (CC) and average vegetation height (H). Canopy cover can be expressed as: in, It is the area covered by the projection of vegetation. This represents the total area of ​​the sample plot. For multi-layered vegetation consisting of trees, shrubs, and grasses, the average coverage and height of different layers can be obtained using a stratified fixed-point measurement method.

[0032] In addition, representative soil and vegetation samples need to be collected on-site and their corresponding GPS information recorded. These samples should be brought back to the laboratory for detailed analysis of particle size distribution, physicochemical properties (organic matter content, pH, cation exchange capacity), and plant species / physiological indicators. The organic matter content (OM) is calculated using the potassium dichromate oxidation-titration method, and the formula is as follows: in, For blank titration value, For sample titration value, The concentration of the titrant is... For sample quality.

[0033] Through the above-mentioned multi-dimensional instrument measurements and sample collection and analysis, a complete data matrix was formed for each sample plot. ,in The i-th ecological geological parameter provides high-precision, standardized raw data support for subsequent factor correlation and mechanism modeling.

[0034] like Figure 4 As shown, step 4: Combine multi-temporal remote sensing, ground survey and socio-economic data to dynamically assess the intensity of human activities (grazing, mining, land use change) in the region, and establish a dynamic relationship model of human activities-ecological disturbance-geological response to provide disturbance variable input for causal analysis.

[0035] S401. Utilizing multi-temporal remote sensing imagery, ground surveys, and socioeconomic statistics, dynamic monitoring and quantitative modeling of regional human activity intensity and ecological disturbance are conducted. Specifically, high-resolution remote sensing data from typical years are selected, and change detection and time-series analysis methods are employed to quantify the temporal trajectory of land use / cover type (LUCC). Remote sensing change indices such as the Normalized Difference in Building Index (NDBI), Normalized Difference in Bare Land Index (NDSI), and Normalized Difference in Vegetation Index (NDVI) are calculated using the following formula: in, This represents the reflectance for the corresponding band. Based on the difference in multi-temporal remote sensing indices. In addition to land use classification results, combined with ground survey information, the intensity of human activity (HAI) can be quantified using the index superposition method or multiple regression method, as shown in the following formula: in, Let be the change of the i-th indicator (such as change in bare land area, change in building area, change in cultivated land, decrease in pasture coverage, etc.) at time t. The weights of each indicator are determined based on the actual degree of impact. At this point, ground surveys and social statistics supplement the data with information such as the number of mining enterprises, grazing density, population changes, and economic activity indices. By overlaying these data with remote sensing results into vector rasterization, a dynamic distribution of human activity intensity at both spatial and temporal levels can be achieved.

[0036] S402. Dynamic monitoring of ecological disturbance factors (surface vegetation destruction, land degradation, and soil erosion) is achieved through remote sensing-ground collaborative surveys, extracting disturbance indices (DI), including: disturbance patch density and average disturbance area, expressed mathematically as follows: in, Let be the area of ​​the j-th disturbed patch, A be the total area of ​​the study area, and m be the total number of disturbed patches. The disturbance intensity varies with time series and can be further characterized as follows: To analyze trends in human activity disturbances.

[0037] S403. Finally, to couple the dynamic relationship between human activities and ecological / geological factors, a three-dimensional time-series model of human activities-ecological disturbance-geological response is established, using structural equation modeling (SEM) or regression modeling. Its basic expression is: in, For geological response variables at time t (such as the rate of desertification expansion, the proportion of exposed rock, etc.). These are ecological and environmental background factors (precipitation, slope, lithology). Intensity of human activity, It is an ecological disturbance factor. This represents the residual term. The function is fitted using machine learning methods such as multiple linear regression and random forest regression. Parameter significance can be tested using t-tests or F-tests. Through model parameter and sensitivity analysis, the quantitative effects of major perturbation variables on geological responses are clarified, providing scientific support for in-depth analysis of the causes and dynamic evolution mechanisms of rocky desertification.

[0038] like Figure 5 As shown, step 5: Multi-source aggregation of the aforementioned natural ecological geological parameters and human activity disturbance data, based on geostatistics and geographical weighted regression, structural equation modeling and other methods, comprehensively quantifies the quantitative contribution and interaction of various dynamic factors, and realizes multi-scale coupling and weighted discrimination of the causes of rocky desertification.

[0039] S501. First, spatially register and scale-standardize the natural ecological geological parameters (lithology, slope, weathering degree, soil thickness, vegetation structure, and water content) and human activity disturbance data (including land use change, human activity intensity index, and disturbance index) obtained in the preceding steps to form a multi-source data raster matrix covering the study area. For each dynamic factor (natural factor set)... Human disturbance factor set ) and spatial distribution characteristics of rocky desertification (outcome variable) To establish the quantitative correlation, geostatistical methods were first applied, using the semivariogram function to analyze the spatial structure of the parameters. in, For a certain factor at a point in space The value of , This represents spatial distance. By analyzing the variogram, we can focus on the spatial correlation scale of each factor, providing support for multi-scale modeling.

[0040] S502. Subsequently, a geographically weighted regression (GWR) model was used to characterize the contribution of each factor to the spatial heterogeneity of rocky desertification. Its mathematical expression is as follows: in, As a factor At a point in space The local regression coefficients, For the corresponding factor value, The residuals are used to analyze the spatial distribution of regression coefficients, which reveals the weight changes and spatial interaction patterns of various dynamic factors in different regions.

[0041] S503. To further quantify the direct, indirect, and interaction effects among factors, a structural equation model (SEM) is introduced, integrating natural, perturbation, and response variables. The general expression of SEM is as follows: in, This is an endogenous latent variable (rock desertification response). These are exogenous latent variables (natural and human disturbance factors). For the factor loading matrix, and The path coefficient matrix, This represents the error term. By estimating and testing the significance of the model path coefficients, the direct effects (main effects), indirect effects (indirect influence through mediating factors), and interaction effects of various dynamic factors are quantitatively characterized. Simultaneously, the model goodness-of-fit (MOF) is used... The explanatory power of a model can be evaluated using metrics such as RMSEA and CFI.

[0042] Finally, by integrating multi-scale spatial statistics, multi-factor regression, and structural equation path decomposition results, the coupling mechanism of the rocky desertification dynamic system in the study area was analyzed, yielding systematic quantitative conclusions including the weights of multiple source factors, spatial differentiation characteristics, and the identification of the main controlling factors. This provides solid data support and scientific basis for differentiated strategies for rocky desertification control and ecological restoration.

[0043] Step 6: Using GIS and a dedicated simulation platform, input the dynamic factor model from Step 5, combine it with historical time series data, reconstruct the spatiotemporal expansion process of rocky desertification, identify and summarize its main controlling ecological and geological formation mechanisms, and output a targeted discrimination index system and predictive model.

[0044] Based on the aforementioned dynamic factor coupling model, a GIS spatial analysis and dedicated simulation platform (such as CLUE-S, CA-Markov, DINAMICA EGO, etc.) is used to dynamically reconstruct and simulate the spatiotemporal process of rocky desertification. Specifically, the multi-factor weights and their spatial distribution parameters obtained in step 5 are first parameterized and input into the simulation platform. Combined with historical rocky desertification distribution data retrieved from remote sensing and related ecological, geological, and socio-economic time-series data, the initial state space of the rocky desertification grid cells is constructed. Geographically weighted regression or spatial autoregressive (SAR) models are used to assign values ​​to the rocky desertification probability field. For example, Logistic regression is used to describe the conditional probability of rocky desertification occurring in a certain grid cell. in, For position The probability of desertification occurring at time t. For constant terms, Let be the regression coefficient of the i-th controlling dynamic factor. The value of this factor is given. The model iterates and evolves based on historical or future scenario data. In each update, it combines spatial neighborhood effects and dynamic changes of the main control factors to simulate the expansion and transfer process of rocky desertification, thereby achieving dynamic reconstruction of the spatiotemporal pattern of the entire region.

[0045] Based on the spatial simulation results, principal component analysis (PCA) was used to identify the main eco-geological mechanisms controlling the expansion of rocky desertification, and the most explanatory main variables and their modes of action were summarized. A rocky desertification identification index system was simultaneously output, including key thresholds and combination patterns of natural and human factors. The rocky desertification identification index system can be formally represented as: in, To determine the comprehensive assessment score for rocky desertification risk, Let i be the weight of the i-th indicator. These are the normalized corresponding factors (soil thickness < threshold, slope > threshold, vegetation cover < threshold). For future scenario analysis, predictive regression or machine learning models are further constructed to achieve quantitative prediction of the spatial distribution and dynamic evolution of rocky desertification. A typical predictive model is expressed as: in, For time Time position The state of rocky desertification This is the initial state. For each controlling factor, For spatial neighborhood influence, For fitting or machine learning prediction functions, This represents the model error term. By summarizing the spatiotemporal evolution paths of the simulation results, the main controlling mechanisms, key turning points, and high-risk areas of rocky desertification can be effectively identified, providing a scientific basis and quantitative tools for the formulation of indicator systems, early warning, and precise governance of regional rocky desertification prevention and ecological restoration.

[0046] Example: This embodiment takes a typical rocky desertification area (E: 105.23°-105.45°, N: 26.31°-26.48°, area of ​​about 20 km²) as the research object, and systematically carries out a comprehensive investigation of the ecological and geological conditions and the analysis of the formation mechanism of the mid-to-high mountain rocky desertification area according to the following steps.

[0047] I. Multi-source data acquisition and factor dataset establishment 1. Remote sensing and historical data collection 2. Data Processing and Factor Calculation Remote sensing data underwent radiometric / atmospheric / geometric correction and multi-band registration, spatially aligned with field point data, and then rasterized and fused. Relevant surface ecological and geological factors, including NDVI, soil moisture, slope, lithology, and annual precipitation, were extracted. The parameters are statistically analyzed as follows: The random forest algorithm was used to classify the ecological and geological conditions of regional spatial units and obtain a rocky desertification risk prediction map. The high-risk area of ​​rocky desertification covers an area of ​​approximately 5.8 km².

[0048] II. Sample Site Selection and Standardized Survey Based on spatial clustering (K-means, number of clusters K=6) and topographic analysis, and combined with dimensions such as lithology (limestone / sandstone / shale), slope aspect (south slope / north slope), and vegetation cover (high / low), 12 representative sample plots were selected, and their specific distribution is as follows: Each sample plot was set up with 20x20m standardized quadrats, and the following parameters were collected (excerpts of data from some sample plots): Soil and vegetation samples were sent to the laboratory for analysis to measure parameters such as pH, total nitrogen, total phosphorus, organic matter, and vegetation diversity index. All data were accompanied by GPS location data.

[0049] III. Quantification of Human Activities and Ecological Disturbance By combining land use / cover remote sensing interpretation, village distribution, and human activity factors such as roads / mines / arable land, the intensity of human activity (HAI, 0-1) is quantified, and key time series from 2015, 2018, and 2021 are selected: Combining remote sensing surface change indices (NDVI change rate, exposure index), field surveys, GDP per capita, population density, etc., a spatiotemporal model of human activity-ecological disturbance-geological response is established, as follows: IV. Quantitative Decomposition and Model Analysis of Dynamic Factors of Rocky Desertification 1. Spatial data unification and scale standardization Data from multiple sources, including remote sensing, geology, humidity, land cover, and human activity disturbance, are processed into 30m raster data to construct a multidimensional data matrix. Spatial semivariogram analysis and geographic weighted regression (GWR) modeling are then performed on the region.

[0050] 2. Example table of principal factor contribution rates in Geographically Weighted Regression (GWR) 3. Example of path coefficients in structural equation modeling (SEM) SEM models show that human activities indirectly drive the expansion of rocky desertification through two main pathways: vegetation cover reduction and soil thinning.

[0051] V. Reconstruction of the Spatiotemporal Expansion of Rocky Desertification and Establishment of an Indicator System Using remote sensing and ground survey data from 2015, 2018, and 2021, combined with GWR / principal component and regression sensitivity analysis, the distribution and evolution trend of rocky desertification were reconstructed using a GIS platform.

[0052] 1. Sensitivity analysis of principal control factors 2. Summary of core indicators for rocky desertification identification and predictive models High-risk areas include areas with NDVI < 0.18, soil thickness < 6 cm, human disturbance HAI > 0.5, slope greater than 25°, and limestone areas.

[0053] The established multi-factor prediction model (random forest AUC=0.91) can simulate the spatial distribution of rocky desertification in the region over the next 5 years, providing support for comprehensive governance decision-making.

[0054] This embodiment systematically integrates multi-source remote sensing, field surveys, spatial analysis, multivariate statistics, and modeling methods in a typical mid-to-high mountain rocky desertification area in Guizhou Province. It meticulously collects and analyzes ecological and geological driving factors and their interactions with human activities, constructs a dynamic mechanism model of rocky desertification, and achieves high-precision identification of rocky desertification causes and spatiotemporal process reconstruction. The summarized main controlling factors and indicator system have significant reference value for rocky desertification prevention and ecological restoration practices.

[0055] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM).

[0056] It should be understood that the above detailed description of the technical solutions of the present invention with reference to preferred embodiments is illustrative and not restrictive. Those skilled in the art can modify the technical solutions described in the embodiments or substitute some of the technical features based on reading this specification; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for comprehensive investigation and analysis of the ecological and geological conditions and formation mechanisms in mid-to-high mountain rocky desertification areas, characterized by: The method includes: Collect multi-source remote sensing data and historical geological, soil and climate data of the high mountain area in the target area, and establish a regional ecological geological factor dataset. Spatial clustering and topographic analysis methods were used to select and deploy sample plots for representative ecogeological units with different lithology, slope aspect, and vegetation cover. At the selected sample plot locations, ecological geological parameters such as lithology, weathering degree, soil thickness, water content, and vegetation structure are obtained to form a multidimensional raw data matrix; By combining multi-temporal remote sensing, ground surveys and socio-economic data, the intensity of human activities in the region is dynamically assessed, and a dynamic relationship model of human activities, ecological disturbances and geological responses is established. By pooling the aforementioned natural ecological geological parameters and human activity disturbance data from multiple sources, and based on geostatistics, geographical weighted regression, and structural equation modeling methods, the quantitative contributions and interactions of various dynamic factors are comprehensively quantified to determine the causes of rocky desertification. Using a GIS platform, inputting dynamic factor models, and combining historical time-series data, we reconstruct the spatiotemporal expansion process of rocky desertification, identify and summarize its main controlling ecological and geological formation mechanisms, and output a targeted discrimination index system and predictive model.

2. The method for comprehensive investigation and analysis of the formation mechanism of ecological geological conditions in mid-to-high mountain rocky desertification areas according to claim 1, characterized in that: The establishment of the regional ecological geological factor dataset includes: spatial registration and rasterization fusion with field survey data of geology, soil, and climate through radiometric correction, atmospheric correction, geometric correction, and band registration; Ecological geological factors such as surface exposure NDVI, soil moisture, slope, lithology, and annual precipitation were calculated, and the random forest method was used to classify ecological geological conditions and predict the risk of rocky desertification.

3. The method for comprehensive investigation and analysis of the formation mechanism of ecological geological conditions in mid-to-high mountain rocky desertification areas according to claim 1, characterized in that: The spatial clustering and topographic analysis method described above is based on the distribution data of ecological and geological factors. It takes lithology, slope aspect, vegetation coverage, soil thickness, and surface bareness index as multidimensional features, performs unsupervised grouping of pixels in the study area through spatial clustering algorithm, selects sample plots by combining topographic analysis and accessibility constraints, and adopts maximum spatial dispersion optimization.

4. The method for comprehensive investigation and analysis of the formation mechanism of ecological geological conditions in mid-to-high mountain rocky desertification areas according to claim 1, characterized in that: The aforementioned multidimensional raw data matrix, standardized survey sample areas were set up on site, soil thickness was determined by geological drilling, and soil moisture content was measured by time domain reflectance (TDR) or portable rapid soil analyzer. Canopy coverage and average vegetation height were obtained by using a plant canopy meter or drone photography, and soil and vegetation samples were collected for laboratory physicochemical and physiological property analysis. GPS information of the center point was recorded for all sample plot parameters.

5. The method for comprehensive investigation and analysis of the formation mechanism of ecological geological conditions in mid-to-high mountain rocky desertification areas according to claim 1, characterized in that: The aforementioned model establishes a dynamic relationship between human activities, ecological disturbances, and geological responses. It utilizes multi-temporal remote sensing images, ground surveys, and socio-economic statistics to dynamically monitor the intensity of human activities and ecological disturbances, and quantifies land use / cover types and remote sensing change indices. By combining ground survey information and social statistical data, spatial and temporal distribution models of human activity intensity are constructed. Then, through dynamic monitoring of ecological disturbance factors and a three-dimensional time series model of human activity-ecological disturbance-geological response, the geological environment response in rocky desertification areas is quantitatively analyzed.

6. The method for comprehensive investigation and analysis of the formation mechanism of ecological geological conditions in mid-to-high mountain rocky desertification areas according to claim 1, characterized in that: The aforementioned determination of the causes of rocky desertification includes: spatially registering and scaling natural ecological geological parameters and human activity disturbance data to form a multi-source data raster matrix, and using semivariance function to analyze the spatial structure; We introduce structural equation modeling to quantitatively decompose the direct, indirect, and interactive effects between natural factors, human disturbance factors, and rocky desertification response, and analyze the coupling mechanism of the rocky desertification dynamic system.

7. The method for comprehensive investigation and analysis of the formation mechanism of ecological geological conditions in mid-to-high mountain rocky desertification areas according to claim 1, characterized in that: The process of reconstructing the spatiotemporal expansion of rocky desertification includes: using a GIS platform to input multi-factor weights and spatial distribution parameters, and combining remote sensing inversion of rocky desertification distribution and ecological, geological, and socio-economic time series data; A gridded cellular state space for rocky desertification was constructed, and the spatiotemporal process of rocky desertification was dynamically reconstructed using spatial regression and simulation methods. Principal component analysis and sensitivity analysis were used to identify the main controlling factors, and a rocky desertification discrimination index system and predictive model were established.

Images