A land ecosystem degradation evaluation method based on a random forest model

Abstract

The application discloses a kind of land ecosystem degradation evaluation methods based on random forest model, comprising: constructing net primary productivity sample library;With net primary productivity value as dependent variable, with the optimal subset of environmental factors determined by characteristic screening as independent variable, construct random forest model, simulate land ecosystem net primary productivity benchmark year by year by using trained model;Based on the theory of climatic climax, the maximum value in long time series net primary productivity benchmark is extracted as the reachable upper limit of net primary productivity per pixel, and the complete loss state of productivity is taken as the degradation endpoint, combined with the actual observation net primary productivity and net primary productivity benchmark, construct land ecosystem degradation index per year per pixel;Based on the spatial and temporal distribution of land ecosystem degradation index, identify and evaluate the spatiotemporal differentiation characteristics and degradation dynamic trend of different ecosystem types.The present application provides a high-fidelity, generalizable natural reference standard for degradation assessment.

Images