Study On Biomass Inversion Of Desert Grassland In Junggar Basin

As the largest renewable resource in the global terrestrial ecosystem,grassland plays an important role in maintaining regional ecological environment and sustainable development of animal husbandry.Grassland biomass is the material source to support the operation of grassland ecosystem.Therefore,it can provide theoretical basis for the improvement of ecological environment and the rational allocation of resources for the study area to accurately obtain the temporal and spatial distribution characteristics of grassland biomass,even to carry out dynamic ethnology,GIS and statistical theory and related methods,the above-ground biomass of desert grassland vegetation in Junggar basin is studied in this paper,to select the best remote sensing retrieval model of aboveground biomass for typical desert grassland types in Junggar basin based on vegetation index,and then the spatial distribution characteristics of the desert grassland biomass in the study area are analyzed by the inversion results,so as to realize the transformation from sample plot investigation to large area remote sensing invonitoring.Based on the remote sensing image of Landsat 8 OLI satellite,combined with Remote Sensing Tecersion.The main results are as follows:（1）The correlation analysis between vegetation index（NDVI,GNDVI,ARVI,DVI,SAVI and RVI）extracted from lanndsat-8 OLI image and the measured data of aboveground biomass in the study area shows that the six vegetation indexes have significant correlation with the aboveground biomass of grassland,the best single factor modeling parameter for small trees,shrubs and semi shrubs is NDVI,and the best modeling vegetation index for herbs is SAVI.（2）The stepwise regression method is used to construct the multivariate linear regression model of vegetation indexs and test samples which are significantly related to the aboveground biomass of grassland.The inversion accuracy of the biomass model of arbor,shrub and grassland is improved in different degrees compared with that of the single curve regression model.（3）To validate the ability of the single regression model and the multiple stepwise linear regression model for estimating the aboveground biomass of four grassland types,the results show that the best model for the selection of small arbor grassland is Y=103.436+987.372NDVI-707.169GNDVI+26.497RVI-211.619ARVI+129.592SAVI,the determination coefficient of the inversion model is 0.716,and the root mean square error is 44.8 g/m2,the optimal model of shrub grassland is Y=44.883+236.484NDVI+205.558GNDVI,R2=0.699,RMSE is 79.8 g/m2,the optimal model of semi shrubby grassland is Y=63.539+407.950 NDVI-13,131RVI-106.414 SAVI,R2 is 0.826,RMSE is 53.8 g/m2,the optimal model of herbaceous grassland is Y=60.414+312.134 SAVI+12.608ARVI,R2 is 0.796,RMSE is 57.8 g/m2.（4）The above ground biomass of four types of grassland was retrieved by using the selected optimal above ground biomass inversion model.The spatial distribution map of above ground biomass of four types of grassland was obtained,and the above ground biomass information of each type of grassland was statistically analyzed.Among them,the average value of above ground biomass retrieved by small trees model was 160.5 g/m2;the average value of aboveground biomass inversion of shrub grassland is 168.6 g/m2,the average value of aboveground biomass inversion of semi shrub grassland is 110.1 g/m2,for herb grassland,the average value of aboveground biomass inversion of model is 78.2 g/m2.