CN-121999381-A - Mountain canyon area ecological geological background rapid investigation and evaluation method based on remote sensing and geological information

Abstract

The invention discloses a rapid investigation and evaluation method for ecological geological background of a mountain canyon region based on remote sensing and geological information, and belongs to the technical field of ecological geological investigation and evaluation. The method accurately extracts the structural features of the topography factors and the landforms through the information integration of multi-source remote sensing and geological, landform, soil and vegetation databases, radiometric calibration, atmospheric correction and spatial registration. And carrying out landform unit subdivision by combining fractal dimension and landform morphological index, introducing a geological-ecological coupling self-supervision graph neural network model, and carrying out space heterogram expression on different geological and ecological attributes. And realizing automatic partition evaluation on regional ecological sensitivity and geological stability by utilizing space statistics and cluster analysis, and checking and optimizing model parameters through field sample plots. The method can efficiently realize the ecological geological background unit identification of the complex region of the high mountain canyon, and provides scientific support for ecological protection and geological disaster risk management.

Inventors

• CHEN QINGSONG
• ZHANG YU
• LI YIYING
• Gao Xujiao
• SHI JIALIANG
• ZHU JIANENG
• MA YIQI
• YIN LINHU
• LIU JING
• WANG YIYUE
• LI RUI
• TU CHUNLIN
• ZHANG QIUYU
• HUANG AN
• SHI YU
• Tao Lanchu
• LIU ZHENNAN
• HE CHENGZHONG
• Cun Dexin
• JIANG XIN
• YANG JINGUO
• LI QIANQIAN

Assignees

• 中国地质调查局昆明自然资源综合调查中心

Dates

Publication Date

20260508

Application Date

20260120

Claims (7)

1. 1. A method for quickly investigating and evaluating ecological geological background of mountain canyon area based on remote sensing and geological information is characterized by comprising the following steps: Acquiring and preprocessing remote sensing and geological multisource data, and acquiring multisource remote sensing data covering a target mountain canyon area and the information of the existing geology, landform, soil and vegetation databases; The method comprises the steps of extracting the feature factors and the feature structural features of the landforms, introducing topological relations of the landform structures as constraint conditions on the basis of automatically extracting gradient and slope-direction basic feature parameters based on DEM data, and finely dividing landform units by combining fractal dimension and morphological indexes; The geological and ecological information is fused, and a geological-ecological coupling self-supervision graph neural network model is adopted to construct an isomerism map containing geological units, lithology and ecological factor multidimensional node properties; The ecological geological background is evaluated in a rapid partition mode, ecological sensibility and geological stability indexes of various landforms and lithology-vegetation combinations are automatically quantized by using a space statistical analysis method, and a space continuous block is divided into a plurality of types of ecological geological background units by using a multi-criterion clustering fusion algorithm; And (3) field sample plot checking and model optimization, preferably carrying out field checking on a representative sample plot, collecting and carrying out field investigation, carrying out field verification on key landforms, geological and ecological features, collecting geological and vegetation feature data of the sample plot, and carrying out model optimization.
2. 2. The method for rapidly investigating and evaluating the ecological geological background of the mountain canyon region based on remote sensing and geological information, which is characterized by comprising the steps of acquiring high-resolution optical images, multi-polarization radar images, digital elevation models, existing geological and landform databases, soil type distribution information and meteorological observation data according to natural landform features of a research area; carrying out radiation calibration, atmospheric correction, orthographic correction, unified projection coordinate conversion, resolution resampling, mask processing and noise suppression on multi-source data, and ensuring space consistency and data quality; and vectorizing and spatially registering the digitized geologic map, soil and vegetation distribution databases to provide high-quality, multidimensional spatial basis data for subsequent analysis.
3. 3. The rapid investigation and evaluation method of the mountain isthmus region ecological geological background based on remote sensing and geological information is characterized in that the feature extraction of the topographic factors and the landforms comprises the steps of automatically acquiring slope, slope and curvature basic topographic parameters by using a space analysis tool based on a digital elevation model, and automatically extracting watershed lines and Gu Dexian by using a watershed analysis method and flow directions, flow grids and elevation thresholds; And combining fractal dimension and landform morphological index parameters, carrying out multidimensional constraint and space connectivity inspection on the landform units to form a layered and topologically clear landform unit system, and carrying out iterative optimization adjustment on partition boundaries.
4. 4. The rapid investigation and evaluation method of the mountain canyon region ecological geological background based on remote sensing and geological information is characterized by comprising the steps of integrating multisource remote sensing interpretation results, geological partitions, landform units and ecological variables into a space heterogeneous map by adopting a geological-ecological coupling self-supervision graph neural network model, taking the geological and ecological units as nodes, relating the space adjacent relation, geological structure connection, water system topology, slope surface upstream and downstream and ecological functions as edges, and enabling node attribute feature vectors to comprise geological unit types, lithology codes, slopes, slope directions, digital elevation model statistical features, vegetation coverage and soil humidity index multidimensional attributes.
5. 5. The rapid investigation and evaluation method of the ecological geological background of the mountain canyon area based on remote sensing and geological information, which is characterized by comprising the steps of automatically quantifying ecological sensitivity and geological stability of a relief block and lithology-vegetation combined space unit based on a space statistical analysis method; The ecological sensitivity indexes comprise normalized vegetation indexes, soil humidity, topography fluctuation degree and gradient multi-factor weighting, and the geological stability indexes are fused with lithology weather resistance, fracture and landslide potentials, slope stability and hydrologic permeability parameters; and adopting K-means clustering to automatically partition each space unit according to the ecological sensitivity and geological stability index vector.
6. 6. The method for rapidly investigating and evaluating the ecological geological background of the mountain canyon area based on remote sensing and geological information is characterized in that the field sample land verification and model optimization comprises the steps of optimizing field sample land areas through multi-level space representative analysis according to an automatic partitioning result and combining administrative areas, landform types, lithology combinations and ecological function areas, acquiring key terrain, landform and surface coverage data by utilizing high-resolution images and ground real points of an unmanned aerial vehicle, and synchronously carrying out on-site geological investigation and ecological variable measurement.
7. 7. The method for rapidly investigating and evaluating the ecological geological background of the mountain canyon region based on remote sensing and geological information, according to claim 4, is characterized in that the geological-ecological coupling self-supervision graph neural network model fuses neighbor node information of different types of edges through a message transmission mechanism in the model reasoning process, and performs multiple training by combining attribute reconstruction loss, neighborhood prediction and contrast loss self-supervision learning mechanisms; the multi-source space attribute high-dimensional topology embedding is realized, the regional distinction and attribute prediction are carried out through the graph classification, the node classification and the clustering algorithm, and the intelligent discrimination capability of the ecological and geological theme characteristics is improved.

Description

Mountain canyon area ecological geological background rapid investigation and evaluation method based on remote sensing and geological information Technical Field The invention belongs to the technical field of ecological geological investigation and evaluation, and particularly relates to a rapid investigation and evaluation method for ecological geological background of a mountain canyon region based on remote sensing and geological information. Background Along with the continuous improvement of ecological protection and geological disaster risk prevention and control demands, quantitative identification and regional evaluation of regional ecological geological background become important foundation of multi-disciplinary cross attention such as geology, ecology and homeland space planning. However, the traditional ecological geological background investigation and partitioning method is dependent on manual interpretation, qualitative judgment and single factor analysis, so that the complex heterogeneity and the multielement coupling characteristic inside the region are difficult to comprehensively reflect, and particularly, the outstanding problems of fuzzy region boundary, unclear attribute superposition, subjective partitioning standard and the like exist in regions with highly diverse geological features and ecological patterns such as mountain canyons and the like. In recent years, the rapid development of technologies such as remote sensing, GIS and space statistics provides a new way for acquiring and processing ecological geological information, but how to scientifically integrate multi-source geological and ecological information, realize automatic partition and risk evaluation of space continuous blocks, and still face technical barriers such as high data fusion difficulty, non-uniform evaluation index system, insufficient modeling of high-dimensional attribute relationship and the like. In addition, the existing partitioning results often lack a targeted field verification and model feedback optimization link, so that scientificity and universality of the partitioning results are to be improved. Therefore, a new technology for high-efficiency ecological geological background comprehensive investigation and evaluation with the capabilities of automatic partition, model dynamic optimization and field verification is needed to be constructed, so as to meet the actual demands of regional ecological protection, geological disaster prevention and control, environmental management and the like. Disclosure of Invention The invention aims to solve the technical problems of realizing efficient fusion and automatic partitioning of multidimensional geological-ecological information such as landforms, lithology, vegetation and the like in highly complex areas of geological landforms such as mountain canyons and the like and ecological environment, overcoming the problems of fuzzy boundary identification, insufficient modeling of attribute relationship and lack of scientific field verification and model optimization in the traditional method, thereby improving the scientificity, the accuracy and the universality of ecological geological background partition evaluation. In order to achieve the purpose, the invention is realized by adopting the following technical scheme that the method comprises the following steps: Acquiring and preprocessing remote sensing and geological multisource data, and acquiring multisource remote sensing data covering a target mountain canyon area and the information of the existing geology, landform, soil and vegetation databases; The method comprises the steps of extracting the feature factors and the feature structural features of the landforms, introducing topological relations of the landform structures as constraint conditions on the basis of automatically extracting gradient and slope-direction basic feature parameters based on DEM data, and finely dividing landform units by combining fractal dimension and morphological indexes; The geological and ecological information is fused, and a geological-ecological coupling self-supervision graph neural network model is adopted to construct an isomerism map containing geological units, lithology and ecological factor multidimensional node properties; The ecological geological background is evaluated in a rapid partition mode, ecological sensibility and geological stability indexes of various landforms and lithology-vegetation combinations are automatically quantized by using a space statistical analysis method, and a space continuous block is divided into a plurality of types of ecological geological background units by using a multi-criterion clustering fusion algorithm; And (3) field sample plot checking and model optimization, preferably carrying out field checking on a representative sample plot, collecting and carrying out field investigation, carrying out field verification on key landforms, geological and ecological features,