Remote sensing inversion of forest and grassland aboveground biomass in Inner Mongolia,China
Multi-source remote sensing data were selected as the primary data sources,including MODIS,Landsat,and land use cover change production dataset.From these data,a total of 21 characteristic variables were extracted encompassing reflectance properties,vegetation indices,climatic factors,as well as soil texture and nutrient content.Following the optimization of these characteristic variables,5 distinct machine learning methods were employed to assess their individual advantages across different climate types.Based on the mod-el with the highest accuracy The spatiotemporal characteristics of forest and grassland aboveground biomass during the growing season in Inner Mongolia from 2000 to 2020 were analyzed.The results show that:1)The number of variables after feature selection varies from 4 to 21,among which reflectance,vegetation index and climatic factors are sensitive characteristic variables of aboveground biomass in all ecosystems and climatic zones;2)Random forest is the model with the highest inversion accuracy,and the precision is significantly bet-ter after zoning;3)The annual average aboveground biomass during the forest growing season fluctuates mod-estly around the average value(3.68 kg/m2)with a slight increase over 21 years,while grassland shows a trend of first decreasing and then increasing,with an overall increase,with the annual average value increasing from 46.36 g/m2 in 2000 to 56.19 g/m2 in 2020;4)The spatial distribution of aboveground biomass during the forest growing season shows a"low-high-low-high"trend from north to south,while grassland gradually increases from west to east.Over 21 years,the area of low biomass decreased and the area of high biomass increased.This study helps to understand the dynamics of local natural resources and provides ideas for large-scale multi-ecosystem biomass inversion.
万方数据
维普
