Evaluation of Linear and Quadratic Models for Estimating Soil Surface Temperature Using Air Temperature in Four Climate Zones of Iran

Soil temperature is an important soil parameter that affects physical، chemical، and biological processes. In this research، some relationships between air temperature (independent variable) and soil temperature at 5 cm depth (dependent variable) were developed using linear and quadratic models. Data from four synoptic stations in Bandar Abbas، Shahrekord، Rasht، and Yazd were used as the representative stations of Iran''s main climates. Daily air and soil temperature data for 10 years (1998-2007) were processed on monthly basis. Linear and quadratic relations were fitted to the data and then evaluated. For this purpose، the coefficient of determination (R2)، Root Mean Square Error (RMSE) and Mean Bias Error (MBE) were calculated. The results showed that the relations between soil and air temperature in different months and climates were different. Comparison of the frequency distribution of the coefficients of linear equations showed that 6. 2%، 31. 3%، 50%، and 12. 5% of the linear model coefficients were weak، moderate، strong، and very strong، respectively. In the case of quadratic model، the coefficient of determination was 6. 3%، 20%، 57%، and 16. 7% as weak، moderate، strong، and very strong، respectively. Therefore، the quadratic model may be more accurate in estimating soil temperature. Comparison of the RMSE values showed that the quadratic model had higher accuracy than the linear model for estimation of soil temperature using air temperature. In conclusion، practical nomographs for determining the coefficients of linear and quadratic equations are presented for different months، which provide rapid estimation of soil temperature (5 cm depth) from air temperature data.