Systematic Land Model Evaluation in the NLDAS Science Testbed

This study demonstrates the utility of the NLDAS (North American Land Data Assimilation System) Science Testbed for the comprehensive evaluation of land-surface models (LSMs). The NLDAS Science Testbed is an environment that enables standardarized evaluation of LSMs available in the NASA Land Information System (LIS) against land and hydrology observational and reference data products, using the NASA Land surface Verification Toolkit (LVT). Recent efforts have employed the NLDAS science testbed for the evaluation of current opertational NLDAS Phase 2 LSMs. Building on these efforts, we present the evaluations of additional LSMs in the same configuration. One such “new” model in LIS is the JULES (Joint UK Land Environment Simulator) LSM. JULES is a community land surface model (LSM) that is used both as a standalone model and as the land surface component in the UK Met Office Unified Model (UM).

To support land data assimilation at the United States Air Force’s 557^thWeather Wing, JULES was integrated into LIS to leverage LIS’s capacities of high performance computation, data assimilation, and flow routing. Two LIS-JULES simulations were conducted in the NLDAS framework, one in an aggregated surface mode and the other in a tiled surface mode. In the JULES aggregated surface mode, it is assumed there is only one surface type in a grid box and aggregated parameter values are used to solve a single energy for the grid. In the JULES tiled surface mode, a separate energy balance is calculated for each surface type in the grid, and grid averages are calculated using the fractional percentage of each surface type. Comprehensive evaluation of the simulated water, energy, and carbon fluxes and states (e.g. snow depth, soil moisture contents, latent heat flux, sensible heat flux, and streamflow) from LIS-JULES, against reference datasets and other LSMs are performed in the NLDAS science testbed. The presentation represents the efforts toward establishing standardized land surface model evaluation procedures and highlights the utility of a comprehensive evaluation/benchmarking environment for the systematic assessment of advances in land surface modeling.