Improving evapotranspiration partitioning by integrating satellite vegetation parameters into a land surface model

Land evapotranspiration (ET) primarily involves vegetation transpiration, canopy interception loss, and soil evaporation. Previous studies have made significant progress in total ET estimation; however, substantial challenges remain in partitioning ET on a regional scale, largely due to the intricate water and energy balance that is disrupted by vegetation cover changes. In particular, the use of land surface models to interpret biophysical processes may be susceptible to uncertainties derived from the estimation of vegetation dynamics. In this study, we integrate satellite leaf area index (LAI) and fraction of vegetation coverage (FVC) into the variable infiltration capacity model (VIC) to improve ET partitioning in the Loess Plateau of China. This region has experienced substantial vegetation greening as evidenced by increased LAI and FVC. The results showed that satellite dynamic vegetation parameters in modeling are effective in improving the estimation of ET components compared with the default vegetation parameters. Specifically, the dynamic parameter of LAI in the model altered the inter- and intra-annual variations in vegetation transpiration and canopy interception loss, supporting the application of dynamic FVC in VIC as being reasonable for allocating transpiration to soil evaporation to capture evaporation from forest gaps. This effect is particularly relevant in arid and semiarid regions. Among the ET components, transpiration was the most sensitive to the two dynamic vegetation parameters, followed by canopy interception loss and soil evaporation. In the Loess Plateau, VIC modeling with dynamic vegetation parameters revealed that the effect of soil evaporation was twice that of transpiration, which is appropriate for this semi-arid region with relatively sparse vegetation coverage. Our study offers valuable insights regarding the use of vegetation coverage for partitioning ET and highlights the advantages of integrating satellite vegetation products into land surface models.