Leaf Area Index Inversion Of Larix Gemlinii Forest Based On GF-1 Image

Leaf area index(LAI)is an important structural parameter of ecosystems,which can reflect plant canopy structure,plant community vitality,and its environmental effects.It is widely used in agriculture and forestry,and has important research significance in the construction of ecosystem and crop growth models,and forest environmental monitoring.In recent years,remote sensing technology has provided a fast and reliable data source for rapid extraction of vegetation LAI.Based on GF-1 satellite data,using empirical models and BP neural network models,this study selected the best remote sensing inversion model for Larix gmelinii forest LAI,and analyzed its spatial distribution status.The following research results were obtained:(1)The five LAI empirical models based on 9 vegetation indices(RVI,NDVI,TVI,SAVI,OSAVI,DVI,GNDVI,RDVI,and WDRVI)have good fitting effects on the LAI in the study area.Among them,the univariate linear and quadratic curve models have good fitting effects,and the OSAVI linear regression model has the best fitting degree.(2)Compared with empirical models,the overall R~2of the BP neural network model is slightly higher,and the RMSE is smaller,enabling more accurate inversion values and higher overall prediction accuracy;The LAI inversion results based on the BP neural network model show that the growth of Larix gmelinii in the study area is good,and its LAI is basically above 3.0,which is basically consistent with the actual growth situation of Larix gmelinii forest.The BP neural network model established by using a hidden layer in this study can provide methodological support for the application and popularization of GF-1 satellite in the phenological phase inversion of plant LAI.