Internet of Things (IoT) and Machine Learning (ML) Assisted Reference Evapotranspiration (ETO) Estimations

Abstract

Reference Evapotranspiration (ETO) is the amount of irrigation water required by a model crop to grow at its optimal level under prevailing environmental conditions. Freshwater is a scarce resource that needs to be used judiciously according to the ETO rate for the conservation of irrigation water. Existing methods of ETO rate are complex to be applied at the farmer level. There is a need for a solution to determine the ETO rate from the available minimal set of environmental conditions. Internet of Things (IoT) and Machine Learning (ML) assisted monthly ETO rate estimation from the directly sensed temperature of the crop field is proposed. IoT-assisted directly sensed temperature from the crop field is more effective in the determination of actual crop field ETO rate. The proposed solution for ETO rate determination can be very useful in Precision Agriculture (PA) applications. The study also compares the performance of Na¨ıve Bayes, Support Vector Classifier (SVC), and K-Nearest Neighbors (KNN) in the determination of ETO rate from local crop field temperature. The performance of the ML models is evaluated based on accuracy, F-measure, and recall for ETO estimation. The proposed ETO estimations are also compared against the Penman-Monteith method of ETO measurements to benchmark the performance of the proposed solution.