Developing Temperature Forcing for Snow and Ice Melt Runoff Models in High Mountain Regions

Thursday, 18 December 2014
Andrew P Barrett1, Richard L Armstrong2, Mary J. Brodzik3, Siri-Jodha S Khalsa3, Bruce H Raup3 and Karl Rittger1, (1)National Snow and Ice Data Center, Boulder, CO, United States, (2)Univ Colorado, Boulder, CO, United States, (3)University of Colorado at Boulder, Boulder, CO, United States
Glaciers and snow cover are natural storage reservoirs that delay runoff on seasonal and longer time-scales. Glacier wastage and reduced snow packs will impact the volume and timing of runoff from mountain basins. Estimates of the contributions of glacier and snow melt to runoff in river systems draining mountain regions are critical for water resources planning. The USAID funded CHARIS project aims to estimate the contributions of glacier and snow melt to streamflow in the Ganges, Indus, Brahmaputra, Amu Darya and Syr Darya rivers. Most efforts to estimate glacier and snow melt contributions use temperature-index or degree-day approaches. Near-surface air temperature is a key forcing variable for such models. As with all mountain regions, meteorological stations are sparse and may have short records. Few stations exist at high elevations, with most stations located in valleys below the elevations of glaciers and seasonal snow cover. Reanalyses offer an alternative source of temperature data. However, reanalyses have coarse resolution and simplified topography, especially in the Himalaya. Surface fields are often biased. Any reanalysis product must be both bias-corrected and “downscaled” to the resolution of the melt-runoff model. We present a combined empirically-based bias-correction and downscaling procedure that uses near-surface air temperature from global atmospheric reanalyses to generate near-surface temperature forcing fields for the five river basins in the CHARIS study area. We focus on three 3rd Generation reanalyses; NASA MERRA, NCEP CFSR and ECMWF ERA-Interim. Evaluation of reanalysis temperature fields reveals differences between seasonal means of 500 hPa air temperatures for the three products are of the order of 1 °C, indicating choice of reanalysis can impact model results. The procedure accounts for these seasonal variations in biases of the reanalysis products and in lapse rates.