eCLM simulation outputs for GRACE data assimilation experiments on state vector and ensemble design

Description This dataset contains output from data assimilation experiments using the encore Community Land Model (eCLM), a streamlined fork of CLM5, conducted over Europe at 12.5 km spatial resolution (EUR-11 grid). The simulations assimilate monthly terrestrial water storage anomalies (TWSA) derived from GRACE observations using ensemble-based Kalman filtering techniques. The archived data correspond to a representative one-year subset of the full seven-year simulation period (2003–2009) analyzed in the manuscript “GRACE TWSA data assimilation in land surface models – the role of state vector and model uncertainty representation”. The subset includes all variables and diagnostics required to reproduce the analyses presented in the study. Method The simulations were performed with eCLM driven by meteorological forcing and coupled to the Parallel Data Assimilation Framework (PDAF). Monthly GRACE TWSA observations were assimilated using both a global Ensemble Kalman Filter (EnKF) and the Local Ensemble Subspace Transform Kalman Filter (LESTKF). Multiple data assimilation configurations were tested, differing in(i) the definition of the model state vector (ranging from fully disaggregated soil layers to partially or fully aggregated representations), and(ii) the ensemble generation strategy used to represent model uncertainty (including perturbations of atmospheric forcing and soil hydraulic properties). The dataset includes both model state outputs and assimilation increments, allowing analysis of how vertically integrated GRACE TWSA information is redistributed across soil layers and hydrologically active storage components. Data format and content The data are provided as a compressed tar.gz archive containing NetCDF files. Monthly files include ensemble means of daily outputs of the following variables: H2OSOI – volumetric soil water content TWS_hactive – terrestrial water storage of hydrologically active grid cells In addition, the archive contains: Mean assimilation increment files for each analysis time step A coordinate file defining the EUR-11 model grid Notes Due to the large data volume of the complete experiment suite, only a representative one-year subset of the simulations is archived here. This subset contains the full variable set and temporal structure used in the analyses. The complete simulation output is available from the corresponding author upon reasonable request.