基于无人机航拍的高寒草地生物量升尺度估算研究OA

The combination of ground surveys and multi-source satellite remote sensing inversion is an effective method to estimate grasslands aboveground biomass(AGB).However,the mismatch of vegetation parameters between quadrat and pixel scales restricts its accurate estimation.In the present study,aerial photography with unmanned aerial vehicle(UAV)was carried out across 65 long-term fixed monitoring plots in the Qilian Mountains during the peak growing season from 2023 to 2024.Within the same plots,AGB surveys were conducted at 390 quadrats.Based on Landsat pixels,aerial images,and measured AGB,the Random Forest algorithm was applied to construct an upscaling estimation model of AGB.Model estimation accuracy was verified at different scales and the spatial distribution of AGB in the Qilian Mountains was predicted.Our results showed that vegetation height could correct the saturation of fitting AGB.The upscaling estimation model of AGB had high fitting accuracy both at quadrat scale(R2=0.72,RMSE=38.77 g·m-2)and aerial photo scale(R2=0.78,RMSE=25.41 g·m-2).The predicted AGB(0-250 g·m-2)and its spatial distribution were consistent with the measured AGB.High and low AGB were predominantly distributed in the southeast and northwest of the Qilian Mountains,respectively.Our results suggested that aerial images served as the bridge between the quadrat and pixel scales for upscaling estimation model of AGB.Therefore,we concluded that UAVs could fill the spatial scale mismatch between ground samples and satellite pixels,which was conducive to improve AGB estimation accuracy.