Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Enhancing Pan evaporation predictions : Accuracy and uncertainty in hybrid machine learning models

Khosravi, Khabat ; Farooque, Aitazaz A. ; Naghibi, Amir LU ; Heddam, Salim ; Sharafati, Ahmad ; Hatamiafkoueieh, Javad and Abolfathi, Soroush (2025) In Ecological Informatics 85.
Abstract

Pan Evaporation (Ep) plays a pivotal role in water resource management, particularly in arid and semi-arid regions. This study assesses the predictive performance of a comprehensive range of advanced machine learning (ML) and deep learning (DL) algorithms for Ep prediction using readily available environmental sensing data. The models investigated include M5 Prime (M5P), M5Rule (M5R), Kstar, as well as their hybridized versions employing Bagging (BA), the adaptive neuro-fuzzy inference system (ANFIS), ANFIS-GA (genetic algorithm), and long short-term memory (LSTM) networks. A 30-year dataset of monthly meteorological observations (1988–2018) from the Kermanshah synoptic station in Iran served as the basis for this... (More)

Pan Evaporation (Ep) plays a pivotal role in water resource management, particularly in arid and semi-arid regions. This study assesses the predictive performance of a comprehensive range of advanced machine learning (ML) and deep learning (DL) algorithms for Ep prediction using readily available environmental sensing data. The models investigated include M5 Prime (M5P), M5Rule (M5R), Kstar, as well as their hybridized versions employing Bagging (BA), the adaptive neuro-fuzzy inference system (ANFIS), ANFIS-GA (genetic algorithm), and long short-term memory (LSTM) networks. A 30-year dataset of monthly meteorological observations (1988–2018) from the Kermanshah synoptic station in Iran served as the basis for this analysis, incorporating variables such as temperature, relative humidity, solar exposure, wind speed, and rainfall. Eight input scenarios were developed using both manual and automated feature selection techniques, including correlation-based subset selection evaluation (CfsSubsetEval or CSE), Principal Component Analysis (PCA), and the Relief Attribute Evaluator (RAE). The results demonstrate that the BA-Kstar ensemble model achieved superior performance (R2 = 0.91, RMSE = 1.60, NSE = 0.91, and RSR = 0.30). Notably, manually constructed input scenarios outperformed automated feature selection methods, with maximum temperature emerging as the most significant predictor of Ep variability. This study underscores the reliability and efficacy of hybrid ML models for Ep forecasting, with significant implications for their broader application in diverse climates and geographical regions.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; and
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
BA-Kstar, Deep learning, Evaporation, Kermanshah, Machine learning, Uncertainty analysis
in
Ecological Informatics
volume
85
article number
102933
publisher
Elsevier
external identifiers
  • scopus:85211971758
ISSN
1574-9541
DOI
10.1016/j.ecoinf.2024.102933
language
English
LU publication?
yes
id
cb0443a8-e379-4ee0-a64d-f65bf9d563f5
date added to LUP
2025-03-03 12:11:41
date last changed
2025-04-04 14:38:11
@article{cb0443a8-e379-4ee0-a64d-f65bf9d563f5,
  abstract     = {{<p>Pan Evaporation (E<sub>p</sub>) plays a pivotal role in water resource management, particularly in arid and semi-arid regions. This study assesses the predictive performance of a comprehensive range of advanced machine learning (ML) and deep learning (DL) algorithms for E<sub>p</sub> prediction using readily available environmental sensing data. The models investigated include M5 Prime (M5P), M5Rule (M5R), Kstar, as well as their hybridized versions employing Bagging (BA), the adaptive neuro-fuzzy inference system (ANFIS), ANFIS-GA (genetic algorithm), and long short-term memory (LSTM) networks. A 30-year dataset of monthly meteorological observations (1988–2018) from the Kermanshah synoptic station in Iran served as the basis for this analysis, incorporating variables such as temperature, relative humidity, solar exposure, wind speed, and rainfall. Eight input scenarios were developed using both manual and automated feature selection techniques, including correlation-based subset selection evaluation (CfsSubsetEval or CSE), Principal Component Analysis (PCA), and the Relief Attribute Evaluator (RAE). The results demonstrate that the BA-Kstar ensemble model achieved superior performance (R<sup>2</sup> = 0.91, RMSE = 1.60, NSE = 0.91, and RSR = 0.30). Notably, manually constructed input scenarios outperformed automated feature selection methods, with maximum temperature emerging as the most significant predictor of E<sub>p</sub> variability. This study underscores the reliability and efficacy of hybrid ML models for E<sub>p</sub> forecasting, with significant implications for their broader application in diverse climates and geographical regions.</p>}},
  author       = {{Khosravi, Khabat and Farooque, Aitazaz A. and Naghibi, Amir and Heddam, Salim and Sharafati, Ahmad and Hatamiafkoueieh, Javad and Abolfathi, Soroush}},
  issn         = {{1574-9541}},
  keywords     = {{BA-Kstar; Deep learning; Evaporation; Kermanshah; Machine learning; Uncertainty analysis}},
  language     = {{eng}},
  publisher    = {{Elsevier}},
  series       = {{Ecological Informatics}},
  title        = {{Enhancing Pan evaporation predictions : Accuracy and uncertainty in hybrid machine learning models}},
  url          = {{http://dx.doi.org/10.1016/j.ecoinf.2024.102933}},
  doi          = {{10.1016/j.ecoinf.2024.102933}},
  volume       = {{85}},
  year         = {{2025}},
}