Soil temperature under forests: a simple model for predicting soil temperature under a range of forest types

A new model (soil temperature under forests-STUF) is described for predicting the seasonal variation of daily average temperature within surface soil layers under a range of forest types from the time of establishment through to harvesting. Daily air temperature for the period of simulation (greater than 1 year) was used to estimate the average annual air temperature, its amplitude, and daily fluctuations. Daily soil temperatures were then predicted from this information using empirical modifications to allow for the effects of canopies of trees, understorey and weeds, mass of the litter layer, and depth at which soil temperature is predicted. Measurements of soil temperature under a range of forest species, ages and management systems were available for 51 sites across southern Australia. The model was parameterised to 31 of these data sets, and the remainder used for validation. The model explained 85% of the variation in observed average daily soil temperature in the latter. Despite the wide range in observed canopy and litter cover, model performance was maintained. Sensitivity analysis indicated that the most important input data required were air temperature, leaf area index and soil depth. For less important inputs such as mass of litter and fraction of ground area covered by weeds or understorey, it may be adequate to use either default values or a scoring procedure for ground cover fractions. The model provides a simple and accurate means to predict daily temperatures in soil in any chosen depth increment. It can be used where accurate daily predictions of average soil temperature are required in a soil profile, for example to generate data to drive a model of organic matter decomposition and mineralisation of nutrients. (C) 2003 Published by Elsevier B.V.