A simple model is proposed to describe transient heat and moisture transfer in the soil under moderate climates to predict near surface ground temperatures using a minimum set of variables and easily accessible weather data. The model is computationally efficient enough to allow for multi-year simulations of shallow ground heat exchangers. It uses a realistic representation of the interactions between the main processes occurring at the soil surface and the heat and moisture dynamics in the soil including the influence of water content on soil thermal properties. The model has been tested against soil temperature measurements taken at different depths (from 0.06 to 1.5 m) on a grass-covered site. Measurements, including meteorological data, were recorded with a time step of 10 min for one year. It is shown that the agreement between soil temperatures predicted by the proposed model and measurements is relatively good for either dry or rainy conditions. Average errors are between +0.47 and + 1.63 degrees C. Furthermore, this study shows that a proper account of the soil surface cover and site specific soil properties is necessary to obtain accurate soil temperature predictions. (C) 2016 Elsevier Ltd. All rights reserved.