Improving the accuracy of models to map alpine grassland above‐ground biomass using Google earth engine

Accurate modelling and mapping of alpine grassland aboveground biomass (AGB) are crucial for pastoral agriculture planning management on the Qinghai Tibet Plateau (QTP). This study assessed effectiveness four popular models (traditional multiple linear regression (MLR), support vector machine (SVM), artificial neural network (ANN), deep (DNN)) with various input combinations (geospatial variables [GV], vegetation types [VT], field measurements [FM], meteorological [MV] observation time [OT]) AGB estimation based a new framework using Google Earth Engine. The results showed that feature GV had poor performance in (0.121 < R2 0.591). FM improved accuracy most when incorporated (0.815 0.833). Although MV, VT OT (R2) only by 0.112–0.216 an importance rank order MV > learning models, their outputs could be used to map AGB. Grass was less accurately predicted than shrub AGB, but pooling both VTs 0.171–0.269. followed ranked DNN ANN SVM MLR. highest (R2 = 0.818) all non-field measured (excluding FM) as inputs, it successfully applied dataset (not associated data training testing) 0.676. presents effective operational Accordingly, provides scientific foundations determine sustainable grazing carrying capacity grasslands.