Influences of Snow Cover and Soil Moisture on Monthly Air Temperature

A series of objective specification experiments are performed with monthly 700 mb heights and surface station temperatures for the United States during 1947–80. The errors in these specifications are used in conjunction with observed snow cover and a computed soil moisture index to assess the impacts of a variable surface state on monthly surface air temperature. Over the eastern and central United States, the mean errors of the temperature specifications for the winter months vary by 1–2°C according to the sign of the anomaly of snow cover. Lag results suggest that snow cover can make a modest contribution to the skill of temperature predictions near the snow boundary. The summer specifications are evaluated in terms of a soil moisture index computed from monthly temperatures and precipitation amounts using a modified Thornthwaite/Nappo parameterization scheme. This index varies seasonally in a realistic manner, while the corresponding mean annual runoff is shown to agree well with runoff amounts derived from observed streamflow data. The soil moisture index shows coherent and physically plausible associations with temperature in the central and western United States. In these regions, the mean errors of the temperature specifications vary by ∼0.5°C according to the sign of the soil moisture anomaly. Impacts of this magnitude are smaller than those obtained in recent general circulation model experiments, but are statistically significant at most stations in the west‐central portion of the country.