Operational and LIS-Based North American Land Data Assimilation Systems at National Centers for Environmental Prediction: Capability in Simulating Water and Energy Budget over the Western United States

NLDAS is a multi-institutional collaborative project sponsored by NOAA's Climate Program Office and NASA's Terrestrial Hydrological Program. NLDAS has a long successful history of producing soil moisture, snow cover, total runoff and streamflow products via application of surface meteorology and precipitation datasets to drive four land-surface models (i.e., Noah, Mosaic, SAC, VIC). The purpose of the NLDAS system is to support numerous research and operational applications in the land modeling and water resources management communities. Since the operational NLDAS version was successfully implemented at NCEP in August 2014, NLDAS products are being used by over 5000 users annually worldwide, including academia, governmental agencies, and private enterprises. Over 71 million files and 144 Tb of data were downloaded in 2015. As we endeavor to increase the quality and breadth of NLDAS products, a joint effort between NASA and NCEP is underway to enable the assimilation of hydrology-relevant remote sensing datasets within NLDAS through the NASA Land Information System (LIS). The use of LIS will also enable easier transition of newly upgraded land surface models into NCEP NLDAS operations. Cold season processes significantly affect water and energy cycles, and their partitioning. As such, in the evaluation of NLDAS systems it is important to assess water and energy exchanges and/or partitioning processes over high-elevations. The Rocky Mountain region of the western U. S. is chosen as such a region to analyze and compare snow water equivalent (SWE), snow cover, snow melt, snow sublimation, total runoff, and sensible heat and latent heat flux. Reference data sets (observation-based and reanalysis) of monthly SWE, streamflow, evapotranspiration, GRACE-based total water storage change, and energy fluxes are used to evaluate model-simulated results. The results show several key factors that affect model simulations: (1) forcing errors such as precipitation partitioning into snowfall and rainfall, (2) snow albedo, (3) refreezing of melted snow, (4) boundary layer stability, and (5) freezing and thawing of soil. Though the anomaly correlations indicate good agreement with the observations or reanalysis products, large quantitative differences are evident in certain cases.