Inversion Of Vegetation Biomass Using Remote Sensing In The Yongding River Basin

In this paper, the inversion model of vegetation biomass in the Yongding river basin has been built and analyzed based on multiple level remote sensing data, which can be meaningful to the management and development of environment in the Yongding river basin.Landsat MSS and TM images of five years selected in them from1978to2009have used as the data source to do vegetation classification of the study area. Integrated with the sample survey data in2009, the method of multiple regression analysis has been used to establish the inversion model to estimate the whole vegetation biomass all over the basin. The model was set up based on the relationship among grey scale values of GIS information like terrain. Finally, the vegetation biomass in the Yongding river basin has been estimated with rather ideal estimation accuracy, the result also has been displayed in virtual expression as well. There are some important result as follow.Based on the distribution map of vegetation cover in the Yongding river basin with normalized difference vegetation index, the method of typical sampling and principal components analysis had been used to estimate the vegetation in this region. The results from statistics of vegetation coverage, there have been some fluctuation in the change curve of vegetation coverage. It had been dropping from1978to1987, with a peak in1996, and reached the lowest point in2000. After that, it has been rising slowly.Combined with the survey data and TM images in2009, the model of unitary linear, polynomial, logarithmic/exponential and multiple linear has been established respectively. With precision and error analysis, some useful conclusion was drawn. In the above models, the prediction accuracy of the logarithmic/exponential model with RV1as independent variable is the lowest, and has reached73.85%. The model of logarithmic/exponential and multiple linear with NDVI as independent variable have reached the precision of81.07%and84.56%respectively. The estimation results of multiple linear equation based on multiple factors are ideal and better for performance of the coefficients between vegetation biomass and remote sensing factor.According to the model, vegetation classification map and distribution of vegetation cover with the same relative radiation level, the inversion results of vegetation biomass in the Yongding river basin has been got. The normal values have been filtered from the results with random sample points, and then calculate the average biomass of the unit pixel and draw the change of total biomass in annual. The change is a process of decreased and then increased, which has been decreasing from1987to1996and then growing until2009.