Deep Machine Learning for Forecasting Daily Potential Evapotranspiration in Arid Regions, Case: Atacama Desert Header

Accurately estimating and forecasting evapotranspiration is one of the most important tasks to strengthen water resource management, especially in desert areas such as La Yarada, Tacna, Peru, a region located at head Atacama Desert. In this study, we used temperature, humidity, wind speed, air pressure, solar radiation from local weather station forecast potential (ETo) using machine learning. The Feedforward Neural Network (Multi-Layered Perceptron) algorithm for prediction was under two approaches: “direct” “indirect”. first one, ETo predicted based on historical records, second predicts climate variables upon which calculation depends, Penman-Monteith, Hargreaves-Samani, Ritchie, Turc equations were used. results evaluated statistical criteria calculate errors, showing remarkable precision, predicting up 300 days ETo. Comparing performance approaches learning used, obtained indicate that, despite similar proposed approaches, indirect approach provides better capabilities longer time intervals than direct approach, whose values corresponding metrics are MAE = 0.033, MSE 0.002, RMSE 0.043 RAE 0.016.