Renewed Estimates of Grassland Aboveground Biomass Showing Drought Impacts

Abstract

Variations in ecosystem composition and structure within a cover type of the same biome can be responsible for the large amounts of uncertainty in remote sensing modeling of aboveground biomass (AGB). Based on 820 ground measurements of AGB and the pure vegetation index models, we proposed a spatially variable remote sensing scalar (i.e., M scalar) in estimating AGB by addressing within-type variations of vegetation. We tested this new modeling approach in Inner Mongolia grasslands and achieved a higher R2 (>0.88) and lower RMSE value (<28.17 g/m2) than those in the conventional remote sensing models. The annual average (STD) of the grassland AGB for 2001–2016 was ~87 (±9) Tg, with an AGB density of 104 ± 74 g/m2. The interannual variation of AGB appeared insignificant over the study period, except in the western region where it increased significantly. The dynamics of AGB were highly correlated with drought severity during the growing season (R2 = 0.691, P < 0.001). AGB under normal climate (i.e., without drought) was ~90.88 Tg, but it increased by ~19% (17.11 Tg) in a wet year (2012) and decreased by 20% (18.16 Tg) in a dry year (2001). More importantly, the high grazing intensity played a less significant role in directing grassland AGB dynamics (R2 = 0.024, P = 0.599). Future efforts are needed to explore how within-type variation of vegetation may change temporally and spatially in accurate estimate of AGB across heterogeneous landscapes, as well as the potential underlying mechanisms from biological, physical, and human perspectives.