Application of spatiotemporal techniques to estimate evapotranspiration in the Paraíba do Sul river watershed

Abstract

Nowadays, any study related to water resources and its usage, such as irrigation, water consumption and energy production is a central issue due to the climate change scenario we are currently living. The efficient use of such resources is a must and involves several factors, among them, the estimate of some hydrologic cycle-related variables, highlighting evapotranspiration, among them. About thirty years ago, the FAO recommended Penman-Monteith equation as the most trustworthy and representative methodology to estimate reference crop evapotranspiration. Since then, it has been applied successfully over different regions and under diverse weather conditions. However, this approach has some cons, such as its dependency on ground measurements of most common climatological parameters: temperature, relative air humidity, wind speed or solar radiation. In addition, these measurements are punctual on the weather station’s location and may not fully represent surrounding environments’ conditions. Lately, thanks to technological advances, the development of powerful programming data analysis-oriented languages, the rising of artificial intelligence, as well as big data, we have a wide variety of tools to improve the way we analyse natural phenomena, making it more efficient and productive. Therefore, the main objective of the present study is the use of such technologies aiming to estimate reliable evapotranspiration values, as a central parameter on water resources management at watershed basis, or even as an indicator of crop water loss, by farmers. Mainly, two different technology-based approaches have been applied along this dissertation pursuing the objective previously mentioned: artificial neural networks (ANN’s) and METRIC algorithm. The former is an artificial intelligence-based tool, capable of “recording” specific variables behaviour and succeed in “mimicking” them at a high resemblance level, favouring the possibility of short-half term forecasts to help watershed committees and other responsible bodies manage water resources. On the other hand, METRIC algorithm uses satellite imagery in order to estimate evapotranspiration hourly and able thus to catch the disrupting changes some of the climatological variables suffer along the day, turning into a vital piece of information for farmers, since they can design irrigation schedule more precisely. Results obtained after both procedures’ application, which compose the methodological approach throughout this study, fully satisfied our expectations and showed a high correlation to those results estimated by the methodology of reference. To sum up, we conclude that both approaches are reliable at estimating reference crop evapotranspiration and can be transferred to the agricultural management assuring a steady improvement due to the quick and unstoppable evolution in technology on the “agriculture of precision” field.