Reconstruction of Historical Atmospheric Data by a Hydroclimate Model for the Mekong River Basin

Abstract

It is necessary to have historical atmospheric data at fine time and spatial grid resolutions as input to watershed-scale models of the Mekong River Basin to perform water balance studies over the basin. However, such historical atmospheric data are not available at the desired fine time-space scales for various reasons in this region. Therefore, it is necessary to reconstruct the atmospheric data over the Mekong River Basin at a fine spatial resolution (approximately 20 km) by means of a regional coupled atmospheric-hydrologic model from the available historical global atmospheric data that have a very coarse resolution (approximately 285 km). The coarse resolution (approximately 285 km) reanalysis of atmospheric data for 1956–2006 from the U.S. National Center for Atmospheric Research (NCAR) and the U.S. National Center for Environmental Prediction (NCEP) were used for the reconstruction of the fine resolution (approximately 20 km) historical atmospheric data over the basin. To incorporate the complex topographical and land surface features of this basin, a regional, fully coupled atmospheric-hydrologic hydro-climate model of the Mekong River Basin, called “RegHCM-Mekong,” was developed for the dynamic downscaling. The coarse time-space resolution NCEP-NCAR historical atmospheric data were then downscaled by RegHCM-Mekong to produce meteorological information on rainfall, air temperature, specific humidity, wind velocity, solar radiation, and net radiation at 0.2 degrees spatial resolution (approximately 20 km) and at hourly intervals over the whole Mekong River Basin during a historical period. After a bias correction of the RegHCM-Mekong-downscaled data (by means of Global Precipitation Climatology Project (GPCP) data), the RegHCM-Mekong-reconstructed historical rainfall fields were successfully validated through comparisons against available point-location observations. The comparisons against the interpolated spatial maps of historical precipitation showed satisfactory results.