基于Sentinel-2和Landsat 8数据的天祝县草地地上生物量遥感反演

为精确获取区域草地地上生物量（Above-ground biomass，AGB），本研究利用Sentinel-2和Landsat 8数据，计算5种植被指数，与野外实测AGB建立草地AGB遥感估算模型，并用均方根误差、决定系数和平均相对误差等指标综合比较不同估算模型的反演精度。结果表明：5种植被指数与草地AGB均显著相关；基于Sentinel-2数据建立的AGB估算模型总体上优于Landsat 8的估算结果；7月最优反演模型为基于差值植被指数（Difference vegetation index，DVI）的二次多项式模型，精度达86％；8月最优反演模型为基于绿色归一化植被指数（Green normalized difference vegetative index，GNDVI）的指数模型，精度达84％；天祝县草地AGB的空间差异明显，不同草地类型平均AGB顺序为：山地草甸>高寒草甸>温性草原>温性荒漠草原。以上研究结果可为研究区草地AGB合理估算和放牧管理提供科学依据。

Remote Sensing Retrieval of grassland Above-Ground Biomass in Tianzhu County based on Sentinel-2 and Landsat 8 Data

Accurately obtain the above-ground biomass of grassland (Above-ground biomass, AGB) in the region is important to grassland management. In this paper, Sentinel-2 and Landsat 8 data were used to calculate 5 vegetation indices, and then the grassland AGB remote sensing estimation models were built with the measured AGB in the field. Statistical indicators, such as the root mean square error (RMSE), the R-squared (R²) and the mean relative error (MRE), were used to comprehensively compare the accuracy of different estimation model. The results showed that:the 5 vegetation indices and the grassland AGB were significantly correlated. The optimal inversion model for the grassland AGB in Tianzhu County in July was a quadratic polynomial model based on DVI (Difference vegetation index) with the accuracy of 86%;the best inversion model in August was an exponential model based on GNDVI (Green normalized difference vegetation index) with the accuracy of 84%. The spatial difference of grassland AGB in Tianzhu County was obvious, and the average AGB order of different grassland types was:mountain meadow > alpine meadow > temperate grassland > temperate desert grassland. The above research results can provide a scientific basis for reasonable estimation of grassland AGB and grazing management in study area.