Monthly Soil Temperature Modeling Using Gene Expression Programming

Abstract

Soil temperature is a critical variable controlling below-ground processes for global and continental carbon budgets. However, there are an insufficient number of climatic stations monitoring soil temperature. In this study, GEP model was used for estimation of monthly soil temperature using air temperature, depth, relative humidity and solar radiation data for the Antalya, Isparta, and Burdur in Turkey. This model was tested using measured meteorological data. The values of R2 between observed and predicted soil temperatures ranged from 0.95 to 0.97. Predictions with GEP model show good agreement with actual soil temperature measurements. New equations are presented for calculation of soil temperatures at different depths. The GEP-based formulations are very practical to predict soil temperature. Soil temperature prediction with GEP model is helpful in various processes, including agricultural decision, heating or cooling of buildings and ground-source heat pump applications.