Code 614.3 LIS Seminar Series: Dr. Benjamin Zaitchik

Hydrological Sciences Branch
The Global Land Data Assimilation System (GLDAS) uses sophisticated land surface models (LSMs) to synthesize satellite and ground-based observations and produce optimal fields of land surface states and fluxes. In this study, we present an evaluation of GLDAS performed using river discharge data catalogued by the Global Runoff Data Center. The evaluation is accomplished by implementing a Source to Sink (STS) routing scheme that estimates simulated river discharge at a gauge location on the basis of GLDAS runoff fields and a set of geographically-derived routing parameters. The STS algorithm is computationally efficient and allows for flexible parameterization. In this study STS is applied to GLDAS outputs for the Noah, Mosaic, Common Land Model (CLM), and Variable Infiltration Capacity (VIC) LSMs. Results are useful in evaluating the simulation of hydrological fluxes in the GLDAS suite of land surface models. Over the length of the GLDAS record (1979-2007) it was found that the Noah Land Surface Model (LSM) provides realistic simulations of runoff throughout the tropics and mid-latitudes, but fails to capture the timing of meltwater discharge at high latitudes. The Mosaic LSM performs better at high latitudes but underestimates discharge for many tropical basins. The Community Land Model (CLM) generally provides consistent results across the globe, though results are poor for certain hydrologically important basins. For all models, the STS routing code improved the simulation of inter-annual peak flow variability in most rivers analyzed. It should be noted that all results are sensitive to the choice of precipitation forcing data, and should not be interpreted as a judgment on model quality. Higher resolution GLDAS simulations (0.25˚) provided little improvement in simulated discharge for the large river basins considered in this study.