Evaluating the Performance of Self-Organizing Maps to Estimate Well-Watered Canopy Temperature for Calculating Crop Water Stress Index in Indian Mustard (<i>Brassica juncea</i>)

Abstract

The crop water stress index (CWSI) is a reliable indicator of water status in plants and has been utilized for stress monitoring, yield prediction, and irrigation scheduling. Despite this, however, its use is limited because its estimation requires baseline temperatures under similar environmental conditions, which can be problematic. In this study, field crop experiments were performed to monitor the canopy temperature of Indian mustard (Brassica juncea) from crop development through harvest under different irrigation treatment levels during the 2017 and 2018 growing seasons. Kohonen self-organizing map (KSOM), feed-forward neural network (FFNN), and multiple linear regression (MLR) models were developed for estimating the well-watered canopy temperature (Tc-ww) using air temperature and relative humidity as input predictor variables. Comparisons were performed between model-estimated and measured Tc-ww values. The findings indicate that the KSOM-modeled values presented a better agreement with the measured values in comparison to MLR- and FFNN-based estimates, with R2 values of 0.978, 0.924, and 0.923 for KSOM, MLR, and FFNN, respectively, during model validation. The dry canopy temperature was estimated to be air temperature plus 2°C. The CWSI computed using KSOM-based estimates of Tc-ww was compared with the CWSI obtained from measured values of Tc-ww. The results suggest a significant potential of KSOM for reliable estimation of the Tc-ww for calculating the CWSI that can be automated for developing precision irrigation systems.