UAV and satellite remote sensing images based aboveground biomass inversion in the meadows of Lake Shengjin

The aboveground biomass of wetland vegetation, as an essential indicator of the wetland ecosystem health, is of great significance for the overwintering reproduction, global carbon cycle and ecological purification of rare waterfowl. It is one of the research hotspots in ecology and remote sensing interpretation. The advantage of satellite remote sensing data lies in its wide coverage, but its spatial resolution is low. UAV remote sensing data have high spatial resolution but small acquisition range. At the same time, because of the influence of wetland area, observation system and external environment, it is more complicated and difficult to retrieve the aboveground biomass from remote sensing images. This research studies a kind of inversion method of aboveground biomass based on UAV and GF-1 data. Firstly, UAV visible images of four sample areas and the ground measured sample data are used to establish linear, power function, polynomial, and logarithmic regression model of biomass, visible light band, and a variety of visible light vegetation index. The accuracy of this method was evaluated by the coefficient of determination (R2), mean absolute error (MAE) and root mean square error (RMSE). The optimal model was selected for biomass inversion of UAV images. Then the biomass data inverted from the visible light band and the GF-1 WFV normalized difference vegetation index (NDVI) image are used to establish a regression model to obtain the aboveground biomass distribution map of the vegetation in Lake Shengjin meadows. The results show that the polynomial equation was determined using the red band has higher simulation accuracy for biomass inversion, R2=0.86, MAE=111.33 g/m2, RMSE=145.42 g/m2,and the inversion results obtained by the red band biomass inversion method is highly consistent with the actual biomass distribution. The polynomial model, constructed by GF-1 WFV and biomass inversed by UAV, is the optimal model, and R2 reached 0.91. This study uses UAV and GF-1 data to conduct biomass inversion research. It integrates the advantages of each data and can obtain richer and more accurate information. It could improve inversion accuracy and provide data and technical support for wetland monitoring and wetland restoration management. Thus this work has important research significance and application value. © 2019 by Journal of Lake Sciences.