Method for managing and controlling ecological environment partition based on multi-source data fusion of geographic information system

Abstract

The present application relates to the technical field of geographic information system, and particularly relates to a multi-source data ecological environment zoning management and control method fusing a geographic information system. The method constructs a multi-level discrete global grid system as a unified spatial reference, maps remote sensing images, ground sensor monitoring and point source pollution data to corresponding grid cells to generate a multi-modal grid feature matrix. Terrain elevation and water system flow direction data are extracted, terrain elevation differences between grid cells are calculated, and a geographic barrier adjacency matrix is constructed. The above matrix is input into a graph convolutional neural network, the weight distribution of the graph convolution kernel is constrained by the geographic barrier adjacency matrix, and spatial feature aggregation is performed on the grid topology structure. The present application avoids the damage of uniform resampling to the spatial distribution characteristics of data, introduces terrain physical constraints into the feature aggregation process, overcomes the defects of step-shaped jagged zoning boundaries and crossing natural geographic barriers, and makes the zoning boundaries fit the natural transition zone of ecological elements.