Abstract Clouds have an important controlling influence on the radiation balance, and hence surface melting, over the Greenland ice sheet and need to be classified to derive reliable albedo estimates from visible imagery. Little is known, however, about the true cloud cover characteristics for the largest island on Earth, Greenland. Here, an attempt is made to address this knowledge gap by examining cloud characteristics, as determined by three complementary satellites sensors: the Advanced Very High Resolution Radiometer (AVHRR), the Along Track Scanning Radiometer-2 (ATSR-2), and the Moderate Resolution Imaging Spectroradiometer (MODIS). The first provides a multidecadal time series of clouds, albedo, and surface temperature, and is available, in the form of the extended AVHRR Polar Pathfinder dataset (APP-x), as a homogeneous, consistent dataset from 1982 until 2004. APP-x data, however, are also the most challenging to cloud classify over snow-covered terrain, due to the limited spectral capabilities of the instrument. ATSR-2 permits identification and classification using stereophotogrammetric techniques and MODIS has enhanced spectral sampling in the visible and thermal infrared but over more limited time periods. The spatial cloud fractions from the three sensors are compared and show good agreement in terms of both magnitude and spatial pattern. The cloud fractions, and inferred patterns of accumulation, are then assessed from three commonly used reanalysis datasets: NCEP–NCAR, the second NCEP–Department of Energy (DOE) Atmospheric Model Intercomparison Project (AMIP-II), and the 40-yr ECMWF Re-Analysis (ERA-40). Poor agreement between the reanalysis datasets is found. NCEP–DOE AMIP-II produces a cloud fraction similar to that observed by the satellites. NCEP–NCAR and ERA-40, however, bear little similarity to the cloud fractions derived from the satellite observations. This suggests that they may produce poor accumulation estimates over the ice sheet and poor estimates of radiation balance. Using these reanalysis data to force a mass balance model of the ice sheet, without appropriate downscaling and correction for the substantial biases present, may, therefore, produce substantial errors in surface melt rate estimates.
Read full abstract