A new approach to quantifying soil temperature responses to changing air temperature and snow cover

Abstract

Seasonal snow cover provides an effective insulating barrier, separating shallow soil (0.25 m) from direct localized meteorological conditions. The effectiveness of this barrier is evident in a lag in the soil temperature response to changing air temperature. The causal relationship between air and soil temperatures is largely because of the presence or absence of snow cover, and is frequently characterized using linear regression analysis. However, the magnitude of the dampening effect of snow cover on the temperature response in shallow soils is obscured in linear regressions. In this study the author used multiple linear regression (MLR) with dummy predictor variables to quantify the degree of dampening between air and shallow soil temperatures in the presence and absence of snow cover at four Greenland sites. The dummy variables defining snow cover conditions were z = 0 for the absence of snow and z = 1 for the presence of snow cover. The MLR was reduced to two simple linear equations that were analyzed relative to z = 0 and z = 1 to enable validation of the selected equations. Compared with ordinary linear regression of the datasets, the MLR analysis yielded stronger coefficients of multiple determination and less variation in the estimated regression variables.