Geo-Statistical Assessment of Spatial Variability of Soil Water-Holding Capacity for Optimal Irrigation Under Semi-Arid Vertisols in South India

Based on a field survey conducted in Lalgudi block, Tiruchirapalli district, Tamil Nadu, India, details pertaining to the crops grown, cropping pattern, land utilization and soil samples were collected from 20 locations under 7 different soil textures of semi-arid vertisols. Using the data, efficient geostatistical models have been explored to study the spatial variation of irrigation water requirement of paddy, sugarcane and banana grown in the study area. Field capacity and wilting point are two major soil physical properties that would influence the available plant water. These parameters were estimated with the pressure plate apparatus using soil samples collected in the block. Soil texture-wise assessment was made with regard to the field capacity, wilting point and soil water holding capacity based on the observations. The locations of Lalgudi block were grouped into three categories based on the mean and standard deviation of available water holding capacity, field capacity and wilting point parameters and were statistically assessed. The groups were made by considering the locations with values of a parameter lying under (i) (Mean + SD) limits. Geostatistics were applied for identifying the best interpolation method in order to acquire the spatial map of the available water holding capacity. An initial data exploration indicated about non-normal distribution of the available water holding capacity. Accordingly, log-transformation was made before kriging technique was used for the data. Ordinary kriging and Disjunctive kriging were explored with six models viz., Circular, Spherical, Exponential, Gaussian, Penta-spherical and Sine-Hole effect models and were used for spatial prediction. An estimate of Root Mean Square Error (RMSE) based onthe six models ranged from 2.1 to 2.5 for available water holding capacity parameter. The cross validation statistics indicated that Sine-Hole effect model with disjunctive kriging of available water holding capacity was superior for interpolation with minimum value of RMSE and moderate spatial dependency. The Sine-Hole effect model was found to overestimate the soil properties when interpolation was carried out and gave mean prediction error of -0.07 which indicated that the model was highly efficient. With the available water holding capacity map of Lalgudi block, total available water, readily available water and irrigation interval of paddy, sugarcane and banana were estimated. The irrigation interval of paddy, sugarcane and banana were found to vary between 1-2, 3-5 and 2-4 days respectively within Lalgudi block. Adoption of spatial algorithms for estimating the crop water requirement would greatly help the irrigation planners and water policy makers to create efficient regional plans for making precision irrigation under semi-arid vertisols.