Parameter Variability and Drought Models: A Study Using the Agricultural Reference Index for Drought (ARID)

Water stress is one of the crucial factors that determine crop yield losses. The Agricultural Reference Index for Drought (ARID) is a generic plant water stress index that estimates the level of water stress as the ratio of actual evapotranspiration to reference evapotranspiration, on a daily time step, considering a hypothetical grass reference crop. The objective of this study was to determine the effect of soil inter- and intraclass variability, plant characteristics, and climate locations on ARID output using global sensitivity and uncertainty analysis. Root zone depth, maximum uptake factor, the wilting point (theta(wp)), field capacity (theta(fc)), drainage coefficient, and runoff curve number were the ARID model parameters, while inputs consisted of rainfall and reference evapotranspiration. Global sensitivity analysis was performed for five locations in the southeastern United States and four soil types, sandy loam, silt loam, sandy clay loam, and silty clay, using three methods, the method of Sobol', partial correlation coefficients, and partial rank correlation coefficients. Sensitivity analyses results indicated the monotonic and linear nature of ARID. The effect of location on the results of this study was minimal. Root zone depth was found to be the most influential model parameter, followed by soil hydraulic properties (i.e., theta(fc) and theta(wp)). Uncertainty in ARID outputs varied from soil to soil (intersoil variability), with soils having similar sand content showing similar results. Considerable uncertainty was seen in the sandier soils, indicating the need for accurate estimation of soil hydraulic properties in such soils.