Off-line computer simulations of continental- and global-scale water balances are valuable for studying climate variability/change and the hydrological implications thereof. We describe the creation of a global, multi-decade, daily, 0.5 degree, terrestrial meteorological forcing dataset to drive land surface model simulations of the global water and energy balance. These simulations will provide a long-term, globally consistent and validated set of land surface water and energy fluxes and states at a high temporal and spatial resolution. The dataset will facilitate the study of seasonal and inter-annual variability studies to an extent not possible with currently available datasets and will be suitable for evaluating the ability of coupled models and other land surface prediction schemes to reproduce observed variability of surface fluxes and state variables in space, and temporally for time scales up to decadal.

The forcing dataset is constructed from a combination of existing global datasets. The precipitation component is derived from the Climatic Research Unit (CRU) monthly precipitation product and is downscaled to a daily time step using a combination of the following products: 15-year gauge-based dataset developed by the surface water modeling group at the University of Washington (UW), (2) the NCEP/NCAR Reanalysis product and (3) the 1997-present GPCP gauge/satellite-based dataset. Use of the NCEP/NCAR product for downscaling is advantageous because other forcing variables (e.g., temperature, wind, vapor pressure, radiation) can also be derived from the reanalysis. Comparison of the 3 daily datasets indicated a spurious wave-like pattern in the NCEP/NCAR monthly wet day frequency and precipitation totals in the winter of the Northern Hemisphere high-latitudes. This is corrected using inter-monthly statistics from the UW dataset and monthly wet day frequencies from the CRU dataset. Results from the recent World Meteorological Organization (WMO) Solid Precipitation Measurement Intercomparison are used to correct for the wind-induced undercatch of solid precipitation. Corrections are also made for liquid precipitation wind-induced undercatch and wetting losses. Underestimation of precipitation in mountainous regions is corrected using a hydrologic water balance approach based on watershed runoff ratios and historical discharge data. Corrections are determined by comparing the long-term annual mean of precipitation within the basin to the annual mean precipitation calculated by the annual runoff from that basin by the runoff ratio.

We present summary results of the simulated land surface energy and water fluxes and states. Seasonal soil moisture patterns exhibit expected behaviour with variation in the tropics following the movement of the ITCZ and monsoon systems, regional wetting in the winter of the Northern Hemisphere and low moisture in the desert. Evaporation also shows expected regional patterns and seasonality. Snow cover has a major influence on the global climate and comparisons are made over large spatial scales between predicted snow cover extent and satellite based observations. Trends in the major components of the water cycle are indicators of the effect of and possible feedback to changes in climate. We therefore also calculate trends in evaporation, runoff and soil moisture on a global and continental basis and indicate whether these trends are significant.