Stratospheric Wind Errors, Initial States and Forecast Skill in the GLAS General Circulation Model

Abstract

Relations between stratospheric wind errors, initial states and 500 mb skill are investigated using the GLAS general circulation model initialized with FGGE data. Erroneous stratospheric winds are seen in all current general circulation models, appearing also as weak shear above the subtropical jet and as cold polar stratospheres, In this study we show that the more anticyclonic large flows are correlated with large forecast stratospheric winds. In addition, we find that for North America the resulting errors are correlated with initial state jet stream accelerations while for East Asia the forecast winds are correlated with initial rate jet strength. Using 500 mb skill scores over Europe at day 5 to measure forecast performance, we find that both poor forecast skill and excessive stratospheric winds are correlated with more anticyclonic large-scale flows over North America. We hypothesize that the resulting erroneous kinetic energy contributes to the poor forecast skill, and that the problem is caused by a failure in the modeling of the stratospheric energy cycle in current general circulation models independent of vertical resolution.