Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Integration of the reptile search algorithm and the adaptive neuro-fuzzy inference system enhances standardized precipitation evapotranspiration index forecasting

Kayhomayoon, Zahra ; Bahmani, Mohammad Javad ; Ghordoyee Milan, Sami ; Bazrafshan, Ommolbanin and Berndtsson, Ronny LU orcid (2025) In Scientific Reports 15. p.14647-14647
Abstract

A novel metaheuristic algorithm called the reptile search algorithm (RSA) was introduced in conjunction with artificial neural fuzzy inference system (ANFIS) for the estimation of standardized precipitation evapotranspiration index (SPEI). The model was tested in three different climates: arid and super-cold, semi-arid and cold, and semi-arid and moderate climate across Iran by combining meteorological indices (minimum temperature, maximum temperature, average temperature, precipitation, and potential evapotranspiration) and large-scale climate signals (North Atlantic Oscillation, Arctic Oscillation, Pacific Decadal Oscillation, and Southern Oscillation Index). The results of the ANFIS + RSA model were compared with those of the ANFIS +... (More)

A novel metaheuristic algorithm called the reptile search algorithm (RSA) was introduced in conjunction with artificial neural fuzzy inference system (ANFIS) for the estimation of standardized precipitation evapotranspiration index (SPEI). The model was tested in three different climates: arid and super-cold, semi-arid and cold, and semi-arid and moderate climate across Iran by combining meteorological indices (minimum temperature, maximum temperature, average temperature, precipitation, and potential evapotranspiration) and large-scale climate signals (North Atlantic Oscillation, Arctic Oscillation, Pacific Decadal Oscillation, and Southern Oscillation Index). The results of the ANFIS + RSA model were compared with those of the ANFIS + WOA and ANFIS + GWO models for evaluation. Based on the estimation results and error evaluation criteria, the performance of the ANFIS + RSA model is considered appropriate, showing a higher relative accuracy compared to ANFIS, ANFIS + GWO, and ANFIS + WOA. In semi-arid and moderate climates, the ANFIS + RSA model exhibited the highest prediction accuracy, with RMSE = 0.28, MAE = 0.20, CA = 0.19, and NASH = 0.91. In semi-arid and cold climates, the model's accuracy was slightly lower, with RMSE = 0.33, MAE = 0.23, CA = 0.23, and NASH = 0.85. In arid and super-cold climates, the model's accuracy remained relatively consistent, with RMSE = 0.24, MAE = 0.18, CA = 0.19, and NASH = 0.84. Furthermore, the promising results of the hybrid ANFIS + RSA model can be further evaluated in other regions and climates to assess its overall effectiveness.

(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
in
Scientific Reports
volume
15
pages
14647 - 14647
publisher
Nature Publishing Group
external identifiers
  • pmid:40287468
  • scopus:105003691904
ISSN
2045-2322
DOI
10.1038/s41598-025-98772-9
language
English
LU publication?
yes
additional info
© 2025. The Author(s).
id
3c988827-8bc0-4d99-9fce-676948764513
date added to LUP
2025-06-16 22:18:24
date last changed
2025-07-15 06:49:56
@article{3c988827-8bc0-4d99-9fce-676948764513,
  abstract     = {{<p>A novel metaheuristic algorithm called the reptile search algorithm (RSA) was introduced in conjunction with artificial neural fuzzy inference system (ANFIS) for the estimation of standardized precipitation evapotranspiration index (SPEI). The model was tested in three different climates: arid and super-cold, semi-arid and cold, and semi-arid and moderate climate across Iran by combining meteorological indices (minimum temperature, maximum temperature, average temperature, precipitation, and potential evapotranspiration) and large-scale climate signals (North Atlantic Oscillation, Arctic Oscillation, Pacific Decadal Oscillation, and Southern Oscillation Index). The results of the ANFIS + RSA model were compared with those of the ANFIS + WOA and ANFIS + GWO models for evaluation. Based on the estimation results and error evaluation criteria, the performance of the ANFIS + RSA model is considered appropriate, showing a higher relative accuracy compared to ANFIS, ANFIS + GWO, and ANFIS + WOA. In semi-arid and moderate climates, the ANFIS + RSA model exhibited the highest prediction accuracy, with RMSE = 0.28, MAE = 0.20, CA = 0.19, and NASH = 0.91. In semi-arid and cold climates, the model's accuracy was slightly lower, with RMSE = 0.33, MAE = 0.23, CA = 0.23, and NASH = 0.85. In arid and super-cold climates, the model's accuracy remained relatively consistent, with RMSE = 0.24, MAE = 0.18, CA = 0.19, and NASH = 0.84. Furthermore, the promising results of the hybrid ANFIS + RSA model can be further evaluated in other regions and climates to assess its overall effectiveness.</p>}},
  author       = {{Kayhomayoon, Zahra and Bahmani, Mohammad Javad and Ghordoyee Milan, Sami and Bazrafshan, Ommolbanin and Berndtsson, Ronny}},
  issn         = {{2045-2322}},
  language     = {{eng}},
  month        = {{04}},
  pages        = {{14647--14647}},
  publisher    = {{Nature Publishing Group}},
  series       = {{Scientific Reports}},
  title        = {{Integration of the reptile search algorithm and the adaptive neuro-fuzzy inference system enhances standardized precipitation evapotranspiration index forecasting}},
  url          = {{http://dx.doi.org/10.1038/s41598-025-98772-9}},
  doi          = {{10.1038/s41598-025-98772-9}},
  volume       = {{15}},
  year         = {{2025}},
}