The National Centers for Environmental Prediction (NCEP) completed the North American Regional Reanalysis (NARR) in 2003 as an improvement over the NCEP/National Centers for Atmospheric Research Global Reanalysis (GR). Some of the improvements in the NARR over the GR include: higher spatial and temporal resolution, use of recent versions of the Eta model and the Noah land surface model (LSM), and direct assimilation of precipitation. The initial reanalysis spanned a 25-year period from 1979-2003, with continued products delivered in near-real time as the Regional Climate Data Assimilation System (R-CDAS).

A team of federal agencies and universities which includes the NCEP Environmental Modeling Center, NASA Goddard Space Flight Center, NWS/OHD, NESDIS/ORA, Princeton University, Rutgers University, the University of Washington, University of Maryland, and the University of Oklahoma developed the North American Land Data Assimilation System (NLDAS) for use over the continental United States. The NLDAS infrastructure consists of uncoupled land surface models (LSMs) forced with precipitation observations, output from the Eta model data assimilation system (EDAS), solar radiation from the GOES satellites, and radar precipitation estimates.

The Oklahoma Mesonet is an automated network of over 100 remote, hydrometeorological stations across Oklahoma. In 1999, the Oklahoma Atmospheric Surface-layer Instrumentation System (OASIS) Project upgraded 10 sites, known as OASIS Super Sites, with instrumentation capable of directly measuring the components of the surface energy balance. For this study, the in-situ observations collected at OASIS Super Sites were rigorously quality assured over the period spanning 2002-2003 to ensure they were research quality. Modeled and observed surface-layer variables from three regions of Oklahoma were then compared to determine the relative performance of NARR and the NLDAS LSMs, with an emphasis placed on soil moisture and the turbulent heat fluxes. Modeled soil moisture from NARR was found to decrease more quickly during dry periods in the warm season when compared to soil moisture values from the NLDAS LSMs and observations. The dry anomalies in soil moisture produced relatively large biases in NARR's modeled turbulent heat flux values.