1,- 1Data Science for Sustainable Earth Laboratory (Data 4 Sustainable Earth), Sultan Moulay Slimane University, Beni Mellal, Morocco
- 2Director SMART CONTROL COMMAND ENERGY - S2Cenergy
Water evaporation represents a major source of water loss in agricultural systems, particularly in arid and semi-arid regions. Developing innovative, data-driven approaches to quantify and manage evaporation is therefore essential for sustainable water resource management. This study proposes an original experimental protocol combined with machine learning techniques to support smart evaporation management at the plot scale.
The experimental setup consists of two identical artificial basins exposed to the same climatic conditions: one partially covered by a photovoltaic panel, while the other remains uncovered. Continuous high-resolution measurements of water level variations, along with other climatic parameters (humidity, air temperature, water temperature, TDS, etc.), are collected using sensors, enabling a precise characterization of evaporation dynamics under contrasting surface conditions.
The acquired experimental data constitute a dedicated database for training machine learning models, including Support Vector Machines (SVM), Gradient Boosting, and Random Forest, aimed at predicting evaporation rates and identifying the main controlling factors. According to the results, the Gradient Boosting model performed best, achieving an R² of 0.993 and RMSE of 0.245 for the open basin, and an R² of 0.996 and RMSE of 0.158 for the covered basin, indicating highly accurate predictions. Random Forest and SVM were also tested, showing good and poor predictive performance, respectively.These findings demonstrate the reliability of ensemble models, particularly Gradient Boosting, for modeling evaporation from measured climatic parameters. The models support adaptive irrigation strategies and contribute to the development of an intelligent agricultural plot, where water losses can be anticipated and minimized.
This work highlights the potential of coupling experimental hydrological observations with machine learning to reduce evaporation losses while promoting the integration of renewable energy solutions in agricultural water management.
key words: Water evaporation,Climate Change,Machine Learning,Experimental protocol,Photovoltaic covering, Smart agricultural plot
How to cite: ouchkir, I., Arioua, A., Arioua, B., Karaoui, I., Nait-taleb, O., El Kamouni, F. E., and Bimouhen, M.: An Experimental Protocol Combined with machine mearning for Smart Evaporation Management under Climate Change, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4527, https://doi.org/10.5194/egusphere-egu26-4527, 2026.
