Predicting spatial and temporal patterns of soil temperature based on topography, surface cover and air temperature

Abstract

Soil temperature is a variable that links surface structure to soil processes and yet its spatial prediction across landscapes with variable surface structure is poorly understood. In this study, a hybrid soil temperature model was developed to predict daily spatial patterns of soil temperature in a forested landscape by incorporating the effects of topography, canopy and ground litter. The model is based on both heat transfer physics and empirical relationship between air and soil temperature, and uses input variables that are extracted from a digital elevation model (DEM), satellite imagery, and standard weather records. Model-predicted soil temperatures fitted well with data measured at 10 cm soil depth at three sites: two hardwood forests and a bare soil area. A sensitivity analysis showed that the model was highly sensitive to leaf area index (LAI) and air temperature. When the spatial pattern of soil temperature in a forested watershed was simulated by the model, different responses of bare and canopy-closed ground to air temperature were identified. Spatial distribution of daily air temperature was geostatistically interpolated from the data of weather stations adjacent to the simulated area. Spatial distribution of LAI was obtained from Landsat Thematic Mapper images. The hybrid model describes spatial variability of soil temperature across landscapes and different sensitivity to rising air temperature depending on site-specific surface structures, such as LAI and ground litter stores. In addition, the model may be beneficially incorporated into other ecosystem models requiring soil temperature as one of the input variables.