Artificial Intelligence Techniques in Pan Evaporation Modeling

Pan evaporation, a key indicator of hydrological and agricultural water loss, has traditionally been estimated by physical pan measurements or empirical formulae. Recent advances in artificial intelligence (AI) have transformed this field by offering data-driven alternatives capable of capturing complex nonlinear interactions among meteorological variables. Machine learning approaches such as support vector machines, random forests and Gaussian process regression have demonstrated robust performance across diverse climates. More sophisticated architectures—including recurrent neural networks with long short-term memory units, temporal attention mechanisms and convolutional neural networks—have further improved forecasting by learning temporal patterns and seasonality. Hybrid frameworks that couple deep learning with meta-heuristic optimisation or wavelet preprocessing have yielded additional gains in accuracy and generalisability. These developments are of global significance for water-scarce regions, enabling more reliable irrigation scheduling, reservoir management and climate-change impact assessments.

Research from Nature Portfolio

One study assessed the feasibility of ensemble and deep learning models to predict monthly pan evaporation at four tropical stations over two decades. A random forest model was benchmarked against a convolutional neural network and a deep neural network trained on local temperature, humidity and solar radiation data. All machine learning approaches outperformed classical empirical equations, with the convolutional neural network delivering the highest accuracy in capturing seasonal and inter-annual variability. Model performance was evaluated using root-mean-square error, correlation coefficient and Willmott’s index, demonstrating that convolutional architectures can robustly model nonlinear evaporation processes even with standard meteorological inputs.

Artificial Intelligence Techniques in Pan Evaporation Modeling publication trend

The graph below shows the total number of articles in artificial intelligence techniques in pan evaporation modeling across all publications each year (not limited to Nature Index journals).

Technical terms

Pan evaporation: The rate at which water evaporates from a standardised open-pan evaporimeter under given climatic conditions.

Long short-term memory (LSTM): A gated recurrent neural network architecture designed to capture long-term temporal dependencies in sequential data.

Attention mechanism: A neural network component that dynamically weights input features or time steps according to their relevance to the prediction task.

Convolutional neural network (CNN): A deep learning model employing convolutional layers to extract hierarchical spatial or temporal features from data.

Random forest (RF): An ensemble learning method combining multiple decision trees to reduce variance and improve predictive performance.

Meta-heuristic algorithm: A stochastic search strategy that guides optimisation processes to near-optimal solutions without requiring gradient information.

Relevance vector machine (RVM): A sparse Bayesian model similar to support vector machines that yields probabilistic regression predictions with fewer basis functions.

References

1. Prediction of pan evaporation across diverse climates and scenarios using temporal attention clockwork recurrent neural networks coupled with long short-term memory. Water Cycle (2025).
2. Modeling Pan Evaporation Using Gaussian Process Regression K-Nearest Neighbors Random Forest and Support Vector Machines; Comparative Analysis. Atmosphere (2020).
3. Pan evaporation estimation by relevance vector machine tuned with new metaheuristic algorithms using limited climatic data. Engineering Applications of Computational Fluid Mechanics (2023).
4. Modelling monthly pan evaporation utilising Random Forest and deep learning algorithms. Scientific Reports (2022).