Parameter optimizations and future projections of pasture and toxic weeds with the Community Land Model (CLM5) over the Three River Source Region, Tibetan Plateau

78% of the Three River Source Region (TRSR) is covered by grassland, understanding future variations in grassland growth is fundamental to ecological barrier security. Many advanced land surface models have incorporated vegetation growth modules, but rarely have current land surface models considered the differential growth of pasture and toxic weeds, which have quite different roles in altering surface energy and water cycles. To represent the different growth for pasture and toxic weeds, we performed a global parameter sensitivity analysis for the Community Land Model (CLM5) based on the eFAST algorithm and calibrated two sets of parameters representing pasture and toxic weeds growth. The previous overestimation of Leaf Area Index and above ground biomass was largely reduced after the parameter optimization. Then we performed future simulations for four Shared Social-economic Pathways (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5) during 2015-2100, considering only meteorological impacts and ignoring other future changes (e.g. CO2 or nitrogen deposition). Pasture and toxic weeds biomass (fresh weight) showed a statistically significant increasing trend in all SSPs. This trend was higher for pasture (5.58-22.76 kg·Ha^-1·yr^-1) than for toxic weeds (2.12-7.44 kg·Ha^-1·yr^-1), while toxic weeds showed greater interannual variability. Radiation and soil mineral nitrogen became the two main constraints on future grassland greening rather than warming and moisture. The strongest biomass increases in SSP5-8.5 were mainly due to increases in pasture and toxic weeds biomass in the western TRSR. Such spatial differences between indicated that the western TRSR had much greater uncertainties in the future.