Exploring the impact of time-varying land cover and data assimilation in ecLand

Vegetation and land cover information play an important role in land-atmosphere interactions for both Numerical Weather Prediction and reanalysis systems. In the ECMWF land surface model (ecLand), a fixed monthly climatology is currently employed for land cover,  leaf area index (LAI) and lake cover. Whilst this information accounts for the seasonal cycle, it lacks inter-annual variability. As part of the Copernicus Climate Change Evolution (CERISE) project, monthly varying maps of LAI, land cover and lake information have been developed from 1925 onwards. These maps are based on a combination of observation data sets, machine learning and back-filling methods. These maps are being tested in ecLand forced by ERA5, together with an offline land data assimilation system (LDAS), to produce the ERA-Land-CERISE reanalysis (1939-2019). The LDAS is based on the operational ECMWF land DA system, and consists of a soil moisture, snow depth, 2 m temperature/humidity and lake temperature analysis. Here we briefly describe the time-varying vegetation, land cover and LDAS methods. Case studies for ERA-Land-CERISE are presented, including the warm European summer of 2003. Furthermore, an evaluation of the soil moisture, snow, lake temperature and heat fluxes is performed.