Validation of ECMWF (re)analysis surface climate data, 1979-1998, for Greenland and implications for mass balance modelling of the Ice Sheet

Climate (re)analysis products are potentially valuable tools, when properly verified, for helping to constrain the surface mass balance of the Greenla nd Ice Sheet (GIS). Monthly surface fields from European Centre for Medium- Range Weather Forecasts (ECMWF) operational- and re-analyses spanning 1979- 1998 were validated using in situ data (surface air pressure and temperatur e, precipitation, cloud cover, short-/all-wave radiation, and wind speed/di rection). These validation data are from coastal or near-coastal Danish Met eorological Institute (DMI) synoptic stations, inland Greenland Climate Net work (CC-Net) and University of Wisconsin Automatic Weather Stations (AWSs) , and two energy balance stations near the southern ice margin. The ECMWF a nalyses closely reproduce the seasonal patterns and interannual variations of much of the in situ data. Differences in the mean values of surface air pressure and temperature can mainly be ascribed to orography errors in the analyses' schemes, compared with the latest available accurate digital elev ation model. Much of the GIS margin as modelled by ECMWF was too cold, on a verage by 4 degreesC, and ECMWF precipitation averaged some 136% of the DMI station values. The misrepresentation of the (relatively) steep ice-sheet margin, which tends to be broadened and systematically over-elevated by sev eral hundred metres, orographically reduced temperature and enhanced precip itation there in the ECMWF models. The cloud-cover comparison revealed not dissimilar annual mean cloud covers (ECMWF -8%) but the ECMWF analyses had too little cloud and were too 'sunny' during the critical summer melt seaso n. ECMWF-modelled surface albedo in summer was similar to 11% lower than CC -Net values, which was mainly responsible for the disagreement of modelled surface short-wave radiation fluxes with observations. Model albedo and clo ud errors need to be rectified if the analyses are to be used effectively t o drive energy balance models of Greenland snowmelt. ECMWF wind speed avera ged 66% (62%) of the DMI station (AWS) values. The validation results provi de useful insights into how one can best improve the ECMWF Greenland climat e data for use in glaciological and climatological studies. Copyright (C) 2 001 Royal Meteorological Society.