Drought assessment modelling using biophysical parameters and remote sensing data

Abstract

This study considers the advancement in technical development of a few disciplines as an infrastructure for developing a suitable model and methodology for agricultural drought assessment in semi-arid area. It evaluates capabilities of multisource remote sensing data in developing raster-based biophysical drought assessment models. The capability for expressing the spatial and inter-annual variation of evapotranspiration (ET) over a study area by the proposed models has made it efficient. The base model, Mapping EvapoTranspiration at high Resolution with Internal Calibration (METRIC) has been evaluated for its performance in estimating ET over the pistachio plantation in a semi-arid region. The result proved that the base model gives good accuracy and is suitable for the selected study area. The base model, METRIC, is found sensitive to a number of meteorological parameters. Two-factor analysis for the primary inputs of the base model shows that the surface albedo and surface temperature pairs is the most effective while other tested pairs are found to be least effective. The study suggests that improving the equations of the effective pair should increase the accuracy. In this case, the multilayer perceptron Artificial Neural Network (ANN) technique is used for estimating spatial and temporal distribution of actual ET from satellite based biophysical parameters. The result shows that a strong correlation exist between ET values computed using METRIC and those generated using ANN. ANN sensitivity analysis shows that surface temperature, soil heat flux and surface albedo are the most significant parameters. Exploratory factor analysis using Principal Component Analysis (PCA) was performed to select the most significant biophysical parameters to be used as input to a newly developed BioPhysical Water Stress Index (BPWSI). The BPWSI is a new model for estimating water stress index using the selected biophysical parameters. The results of BPWSI are found to be significant and can be used for predicting the pistachio water status which represents the indication of agricultural drought.