Atmospheric Model Intercomparison Project (AMIP) simulations of snow cover are compared to observed values on hemispheric and continental scales for the time period 1980 through 1988. Monthly means and variability, as well as interannual correlations, of snow cover areal extent are evaluated.
Analyses of January snow cover results from sixteen models reveal that mean snow cover extent tends to be underestimated in North America and western Eurasia (west of 75°E), and overestimated in eastern Eurasia (east of 75°E). In North America three quarters of the models show mean January snow cover extent between 65% and 80%, generally underestimating the observed mean of 78%; the remaining quarter lie below 65%. Over western Eurasia the models also tend to underestimate mean snow cover, with three quarters of the mean values between 40% and 55% compared to the observed 51%; the remaining quarter lie below 40%. Models tend to overestimate mean snow cover over eastern Eurasia, where three quarters of the simulated mean values lie between 55% and 75% compared to the observed 63% cover; the remaining quarter have mean snow cover exceeding 75%.
The dispersion of simulated January snow cover extent around central values are close to the observed dispersion in North America and eastern Eurasia, but are lower than observed in western Eurasia. Standard deviations and ranges of simulated January snow cover in North America and eastern Eurasia are within 3% and 8% of observed values, respectively. Over western Eurasia, however, models tend to underestimate the observed standard deviation and range by as much as 5% and 20%, respectively.
Interannual fluctuations for 1980-1988 are poorly correlated to observed values on both continents. Simulated snow cover time series are compared to observations using the Spearman Rank Correlation method. Out of 48 correlations (16 models, 3 continental areas) only two had significant correlations (n=9, p=.05, 1-tailed).
Snow cover plays a substantial role in modulating the earth's surface
energy balance, thereby constituting an important climate feedback. Accurate
simulation of snow cover is therefore important for accurate climate modeling,
and for the prediction and detection of climate fluctuations.