APPLICATION OF GIS TECHNOLOGIES TO BUILD SPATIAL PREDICTORS FOR MAPPING FOREST ECOSYSTEM FUNCTIONS AT THE LOCAL LEVEL

Abstract

The article presents the experience of using GIS technologies to prepare predictors of regression models of carbon stocks created using the random forest machine learning method. The study was conducted on the territory of the Dankovsky district forestry, located in the south of the Moscow region. GIS analysis of spatial data containing information about the relief and hydrographic network of the study area was performed. As a result, morphometric values describing the surface runoff and altitude zonality of the study area have been created, which will be considered as predictors of carbon stock modeling. The article describes GIS tools that allow you to create thematic geospatial products: exposure, slope steepness and curvature; direction, distance and length of the flow line, total flow; average altitude above sea level and distance to the river. In addition, the boundaries of river catchment basins have been identified by means of GIS analysis, within which it is also planned to perform carbon stock modeling.