A framework based on high-resolution imagery datasets and MCS for forecasting evaporation loss from small reservoirs in groundwater-based agriculture

Abstract

In arid and semi-arid regions, evaporation from small irrigation reservoirs can be a significant source of water loss. Since groundwater is a major source of water supply for irrigation, evaporation losses from irrigation water reservoirs represent a challenging aspect in aquifers governance in such limited-water areas. Estimating these losses is crucial for water resource managers to regulate irrigation reservoirs development and to implement appropriate mitigation measures. Several practical challenges make the individual inventory and monitoring of small irrigation reservoirs unfeasible, especially in large irrigation perimeters with dynamic irrigated surfaces. Thus, significant uncertainty is generally associated with the determinist estimation of evaporation from small irrigation reservoirs. This study is an attempt to develop remote sensing and Monte Carlo Simulation (MCS)-based framework to estimate the evaporation loss from small irrigation water reservoirs used for storing groundwater pumped from the Berrechid aquifer, in Morocco. To that end, remote sensing datasets were validated using data of 49 known reservoirs to identify small reservoirs and their surface area over the growing season. An Exploratory Data Analysis (EDA) of the remotely sensed results was conducted to process the outlier values. Meanwhile, MCS was implemented using 20000 iterations for developing a probabilistic model to estimate the annual evaporation loss associated with exceedance probabilities. The results showed that for an exceedance probability of 90% the associated annual evaporation loss is about 1.50 Mm3·yr−1 with a median of about 1.84 Mm3·yr−1. A sensitivity analysis (SA) of the model was conducted which revealed that the model is more sensitive to the pan coefficient (Cp) followed by reservoir area, and pan evaporation (EVP) for dry months. As for wet months, the SA showed that the model is more sensitive to the daily EVP. Overall, the study provides a new insight for forecasting evaporation loss from small reservoirs and, therefore, will help decision-makers to consider the uncertainty in evaluating the economic viability of mitigation measures. Furthermore, the methodology developed could be valuable in estimating the evaporation loss from the reservoirs in poorly monitored zones.