Predicting Multiple Depths of Bare Soil Temperatures: A Model Evaluation and Modifications for Sandy Soils

Soil temperature (ST) is one of the important soil properties that distinctly influence many physical, biological, and chemical processes taking place in the soil. Only a few models exist that can predict ST as a function of easily obtainable variables. This study evaluates the effectiveness of one of the existing equations (developed for fine-textured soils [clay, silty loam, and sandy loam]) to estimate maximum (max) and minimum (min) STs at multiple depths (0.05, 0.1, and 0.2 m) for bare sandy soils (>900 g kg^-1 sand) of Florida. We calibrated the existing equation parameters for Gainesville, FL, using 7 yr of measured dataset and validated for the same location for 3 yr of independent measured dataset. The performance of the calibrated equations were evaluated for the seven other locations, representing both coastal and inland locations (Bradenton, Lake Alfred, Ft. Pierce, Immokalee, Alachua, Belle Glade, and Ft. Lauderdale). The existing equations did not estimate max and min STs well and provided poorer estimates for the min ST. The absolute error of the min ST estimates were as high as 13.3°C (root mean square error, RMSE = 5.7°C), and 11.8°C (RMSE = 4.4°C) for the 0.05 and 0.1-m soil depths, respectively. When the parameters of the existing models were optimized for a sandy soil (Eustis fine sand, Psammentic Paleudults, sandy, siliceous, hyperthermic) for Gainesville, the performance of both max and min ST estimations were significantly improved. For instance, RMSEs were 1.31, 1.35, and 1.30°C for the 0.05, 0.1, and 0.2 m, respectively, for estimating max ST while they were 1.61, 1.68, and 1.60°C for the min ST for the same depths, respectively. Although equations were modified for a sandy soil (Eustis) some deviations between the estimated and measured ST were observed in the validation for other sandy soils. These deviations were attributed to the variations between the soils, locations, and to the other variables that have influence on ST such as solar radiation, precipitation, etc., that are not included neither in existing equations nor in our equations. The equations to estimate max and min ST are depth-specific. Overall results suggested that the modified equations can be reliably used to estimate max and min ST at multiple depths in sandy soils of Florida.