This powerful new land surface modeling system integrates data from advanced observing systems to support improved forecast model initialization and hydrometeorological investigations. Land surface temperature and wetness conditions affect and are affected by numerous climatological, meteorological, ecological, and geophysical phenomena. Therefore, accurate, high-resolution estimates of terrestrial water and energy storages are valuable for predicting climate change, weather, biological and agricultural productivity, and flooding, and for performing a wide array of studies in the broader biogeosciences. In particular, terrestrial stores of energy and water modulate fluxes between the land and atmosphere and exhibit persistence on diurnal, seasonal, and interannual time scales. Furthermore, because soil moisture, temperature, and snow are integrated states, biases in land surface forcing data and parameterizations accumulate as errors in the representations of these states in operational numerical weather forecast and climate models and their associated coupled data assimilation systems. That leads to incorrect surface water and energy partitioning, and, hence, inaccurate predictions. Reinitialization of land surface states would mollify this problem if the land surface fields were reliable and available globally, at high spatial resolution, and in near-real time.A Global Land Data Assimilation System (GLDAS) has been developed jointly by scientists at the National Aeronautics and Space Administration (NASA) Goddard Space Flight Center (GSFC) and the National Oceanic and Atmospheric Administration (NOAA) National Centers for Environmental Prediction (NCEP) in order to produce such fields. GLDAS makes use of the new generation of groundand space-based observation systems, which provide data to constrain the modeled land surface states. Constraints are applied in two ways. First, by forcing the land surface models (LSMs) with observationbased meteorological fields, biases in atmospheric model-based forcing can be avoided. Second, by employing data assimilation techniques, observations of land surface states can be used to curb unrealistic model states. Through innovation and an ever-improving conceptualization of the physics underlying earth system processes, LSMs have continued to evolve and to display an improved ability to simulate complex phenomena. Concurrently, increases in computing power and affordability are allowing global simulations to be run more routinely and with less processing time, at spatial resolutions that could only be simulated using supercomputers five years ago. GLDAS harnesses this low-cost computing power to integrate observationbased data products from multiple sources within a sophisticated, global, high-resolution land surface modeling framework.What makes GLDAS unique is the union of all of these qualities: it is a global, high-resolution, offline (uncoupled to the atmosphere) terrestrial modeling system that incorporates satellite-and ground-based observations in order to produce opt...

Forcing a land surface model (LSM) offline with realistic global fields of precipitation, radiation, and near-surface meteorology produces realistic fields (within the context of the LSM) of soil moisture, temperature, and other land surface states. These fields can be used as initial conditions for precipitation and temperature forecasts with an atmospheric general circulation model (AGCM). Their usefulness is tested in this regard by performing retrospective 1-month forecasts (for May through September, 1979–93) with the NASA Global Modeling and Assimilation Office (GMAO) seasonal prediction system. The 75 separate forecasts provide an adequate statistical basis for quantifying improvements in forecast skill associated with land initialization. Evaluation of skill is focused on the Great Plains of North America, a region with both a reliable land initialization and an ability of soil moisture conditions to overwhelm atmospheric chaos in the evolution of the meteorological fields. The land initialization does cause a small but statistically significant improvement in precipitation and air temperature forecasts in this region. For precipitation, the increases in forecast skill appear strongest in May through July, whereas for air temperature, they are largest in August and September. The joint initialization of land and atmospheric variables is considered in a supplemental series of ensemble monthly forecasts. Potential predictability from atmospheric initialization dominates over that from land initialization during the first 2 weeks of the forecast, whereas during the final 2 weeks, the relative contributions from the two sources are of the same order. Both land and atmospheric initialization contribute independently to the actual skill of the monthly temperature forecast, with the greatest skill derived from the initialization of both. Land initialization appears to contribute the most to monthly precipitation forecast skill.

Off-line land surface modeling simulations require accurate meteorological forcing with consistent spatial and temporal resolutions. Although reanalysis products present an attractive data source for these types of applications, bias to many of the reanalysis fields limits their use for hydrological modeling. In this study, we develop a global 0.5°forcing data sets for the time period 1979-1993 on a 6-hourly time step through application of a bias correction scheme to reanalysis products. We then use this forcing data to drive a land surface model for global estimation of soil moisture and other hydrological states and fluxes. The simulated soil moisture estimates are compared to in situ measurements, satellite observations and to a modeled data set of root zone soil moisture produced within a separate land surface model, using a different data set of hydrometeorological forcing. In general, there is good agreement between anomalies in modeled and observed (in situ) root zone soil moisture. Similarly, for the surface soil wetness state, modeled estimates and satellite observations are in general statistical agreement; however, correlations decline with increasing vegetation amount. Comparisons to a modeled data set of soil moisture also demonstrates that both simulations present estimates that are well correlated for the soil moisture in the anomaly time series, despite being derived from different land surface models, using different data sources for meteorological forcing, and with different specifications of the land surfaces properties.