Statistical prediction of continental planetary albedo

Abstract

Regression equations have been derived for albedo in the United States, summer and winter, as a function of conventional meteorological variables of the type that might be available from a . numerical model. For monthly mean patterns, the variance reduction was from 60% in summer to 84% in winter on test samples. On a day to day basis, the reduction of variance was from 45% in summer to 55% in winter. Still, errors of estimate are smaller in summer than in winter.The principal predictors were the mean relative humidity below 500 mb, and the pole-ward meridional wind component aloft. In addition, a snow-cover variable was needed in winter. Dynamic predictors, such as vorticity advection or vertical velocity, had relatively little skill and did not enter the equations consistently. Also, stability parameters played an unimportant part.