Spatial and multi-depth temporal soil temperature assessment by assimilating satellite imagery, artificial intelligence and regression based models in arid area

Soil temperature is a noteworthy factor, particularly in farming and land treatment of organic wastes, since development of organic system is firmly controlled by soil temperature. Furthermore, soil temperature has impacts on the physical, substance, and microbiological forms that occur in soil. In this study, for estimation of multi-depth soil temperature, the most reliable combination of input variables for model development was achieved using gamma test technique. The most suitable training and testing data length i.e. training data set 11,529 (01 January 1975 to 25 July 2006; 81%) and testing data set 2716 (26 July 2006 to 31 December 2013; 19%) of the most efficient input variables were determined by means of M-test technique. Multi-depth soil temperature model was achieved by means of multiple linear regression (MLR), multilayer perceptron (MLP) and adaptive neuro-fuzzy inference system (ANFIS) techniques. After comparing results of all models, it was found that the ANFIS technique with generalized bell membership function had the best performing results when compared to MLR and MLP for all cases i.e. 5 cm, 10 cm and 20 cm depths of soil. Potential of gamma test technique was evaluated by means of sensitivity analysis. The spatial variation of land surface temperature (LST) was assessed by remote sensing (RS) and GIS technique and also the qualitative and quantitative performances were evaluated for determining the reliable and accurate spatial variation of LST. In qualitative evaluation, it was observed that the split-window (SW) approach overestimated the LST with coefficient of determination 0.8654. According to the quantitative evaluation based on root mean square error, correlation coefficient and coefficient of efficiency values between observed and estimated LST, the spatial variation of LST was found to be accurate over the Udaipur City, Rajasthan.