Impact of leaf area index assimilation and gauge-corrected precipitation on land surface variables in LDAS-Monde: a case study over China

A global land data assimilation system (LDAS-Monde) forced by the European Centre for Medium-Range Weather Forecasts ERA5 reanalysis is used to simulate land surface variables (LSVs) over China from 1979 to 2019 at a spatial resolution of 0.25 degrees. LDAS-Monde is coupled with the CNRM version of the Total Runoff Integrating Pathways (CTRIP) to convert runoff into streamflow simulations. Four experiments are conducted, with and without assimilating satellite derived leaf area index (LAI) observations, with and without gauge-corrected ERA5 precipitation. Four independent reference datasets are used to assess the impact of different model setups over contrasting climate zones and land cover types. LAI assimilation tends to reduce simulated LAI, evapotranspiration (ET) and gross primary production (GPP), and increase soil moisture (SM) and streamflow. Over semi-arid areas, the corrected precipitation is generally larger than the original ERA5, leading to increased ET, SM and streamflow. Meanwhile, the overestimation of precipitation in relatively humid regions is significantly reduced, leading to a decrease in ET, SM and streamflow. Overall, LAI assimilation alone shows a general improvement for all LSVs, including GPP and ET fluxes, over regions with dense vegetation cover, but degrades streamflow. Precipitation correction shows a general improvement for all LSVs, especially for water-related LSVs (SM and river discharge), but shows little improvement for ET. The impact of LAI assimilation and precipitation correction is more pronounced over agricultural areas in southeastern China, where a wet bias of ERA5 is observed. Except for ET, the combination of LAI assimilation and precipitation correction performs best among all experiments.