Radiative forcing bias of surface albedo modifications linked to simulated forest cover changes at northern latitudes
Abstract. Simulated land use/land cover change (LULCC) radiative forcings (RF) from changes in surface albedo (Δα) predicted by land surface schemes of six leading climate models were compared to those based on daily MODIS retrievals for three regions in Norway and for three winter–spring seasons. As expected, the magnitude and sign of the albedo biases varied considerably for forests; unexpectedly, however, biases of equal magnitude were evident in predictions at open area sites. The latter were mostly positive and exacerbated the strength of vegetation masking effects and hence the simulated LULCC Δα RF. RF bias was considerably small across models (-0.08 ± 0.04 W m-2; 21 ± 11%); 4 of 6 models had normalized mean absolute errors less than 20% (3-year regional mean). Identifying systematic sources of the albedo prediction biases proved challenging, although for some schemes clear sources were identified. Our study should provide some reassurance that model improvement efforts of recent years are leading to enhanced LULCC climate predictions.