Application of ERA-Interim, empirical models, and an artificial intelligence-based model for estimating daily solar radiation
Mohammadi, Babak LU
; Moazenzadeh, Roozbeh
; Bao Pham, Quoc
; Al-Ansari, Nadhir
; Ur Rahman, Khalil
; Tran Anh, Duong
and Duan, Zheng
LU
(2022)
In Ain Shams Engineering Journal
13(1).
- Abstract
Solar radiation plays a pivotal role in the energy balance at the Earth's surface, evaporation, snow melting, water requirements of plants, and hydrological control of catchments. In this work, performance of ERA-Interim (a reanalysis dataset) was examined to estimate solar radiation at Ahvaz, BandarAbbas, and Kermanshah weather stations representing the even spatial distribution over Iran using eight empirical models and an artificial intelligence-based model (SVM: Support Vector Machine). In the calibration set, SVM exhibited the best performance with RMSEs of 249, 299 and 437 J.cm−2.day−1 at the aforementioned stations, respectively. In validation set, SVM reduced the errors in the estimates of solar radiation... (More)
Solar radiation plays a pivotal role in the energy balance at the Earth's surface, evaporation, snow melting, water requirements of plants, and hydrological control of catchments. In this work, performance of ERA-Interim (a reanalysis dataset) was examined to estimate solar radiation at Ahvaz, BandarAbbas, and Kermanshah weather stations representing the even spatial distribution over Iran using eight empirical models and an artificial intelligence-based model (SVM: Support Vector Machine). In the calibration set, SVM exhibited the best performance with RMSEs of 249, 299 and 437 J.cm−2.day−1 at the aforementioned stations, respectively. In validation set, SVM reduced the errors in the estimates of solar radiation by 2.5 and 7.3 percent compared to the best empirical model at Ahvaz station (Abdallah model, RMSE = 242 J.cm−2.day−1) and Kermanshah station (Angstrom-Prescott model, RMSE = 315 J.cm−2.day−1), respectively. During the validation at BandarAbbas station, Bahel and Abdallah model (RMSE = 309 J.cm−2.day−1), Angstrom-Prescott model (RMSE = 310 J.cm−2.day−1) and SVM (RMSE = 312 J.cm−2.day−1) showed a relatively similar performance. The results also showed that the ERA-Interim dataset can be a comparatively suitable alternative to some of the empirical models, where radiation or the input parameters of empirical models are not directly measured, with RMSEs of 382.81, 320.82 and 414.1 J.cm−2.day−1 at Ahvaz, BandarAbbas, and Kermanshah stations, respectively (in validation phase); although its error rates are significant compared with the SVM model, and substituting it for artificial intelligence-based models is not recommended.
(Less)
- author
- Mohammadi, Babak
LU
; Moazenzadeh, Roozbeh
; Bao Pham, Quoc
; Al-Ansari, Nadhir
; Ur Rahman, Khalil
; Tran Anh, Duong
and Duan, Zheng
LU
- organization
- publishing date
- 2022-01
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Iran, Meteorological variables, Renewable energy, Solar radiation
- in
- Ain Shams Engineering Journal
- volume
- 13
- issue
- 1
- article number
- 101498
- publisher
- Ain Shams University
- external identifiers
-
- scopus:85108184760
- ISSN
- 2090-4479
- DOI
- 10.1016/j.asej.2021.05.012
- language
- English
- LU publication?
- yes
- additional info
- Publisher Copyright: © 2021 THE AUTHORS Copyright: Copyright 2021 Elsevier B.V., All rights reserved.
- id
- 0b9729df-f61d-4e08-a576-dd3504e7452b
- date added to LUP
- 2021-06-28 09:49:53
- date last changed
- 2023-05-16 13:22:38
@article{0b9729df-f61d-4e08-a576-dd3504e7452b,
abstract = {{<p>Solar radiation plays a pivotal role in the energy balance at the Earth's surface, evaporation, snow melting, water requirements of plants, and hydrological control of catchments. In this work, performance of ERA-Interim (a reanalysis dataset) was examined to estimate solar radiation at Ahvaz, BandarAbbas, and Kermanshah weather stations representing the even spatial distribution over Iran using eight empirical models and an artificial intelligence-based model (SVM: Support Vector Machine). In the calibration set, SVM exhibited the best performance with RMSEs of 249, 299 and 437 J.cm<sup>−2</sup>.day<sup>−1</sup> at the aforementioned stations, respectively. In validation set, SVM reduced the errors in the estimates of solar radiation by 2.5 and 7.3 percent compared to the best empirical model at Ahvaz station (Abdallah model, RMSE = 242 J.cm<sup>−2</sup>.day<sup>−1</sup>) and Kermanshah station (Angstrom-Prescott model, RMSE = 315 J.cm<sup>−2</sup>.day<sup>−1</sup>), respectively. During the validation at BandarAbbas station, Bahel and Abdallah model (RMSE = 309 J.cm<sup>−2</sup>.day<sup>−1</sup>), Angstrom-Prescott model (RMSE = 310 J.cm<sup>−2</sup>.day<sup>−1</sup>) and SVM (RMSE = 312 J.cm<sup>−2</sup>.day<sup>−1</sup>) showed a relatively similar performance. The results also showed that the ERA-Interim dataset can be a comparatively suitable alternative to some of the empirical models, where radiation or the input parameters of empirical models are not directly measured, with RMSEs of 382.81, 320.82 and 414.1 J.cm<sup>−2</sup>.day<sup>−1</sup> at Ahvaz, BandarAbbas, and Kermanshah stations, respectively (in validation phase); although its error rates are significant compared with the SVM model, and substituting it for artificial intelligence-based models is not recommended.</p>}},
author = {{Mohammadi, Babak and Moazenzadeh, Roozbeh and Bao Pham, Quoc and Al-Ansari, Nadhir and Ur Rahman, Khalil and Tran Anh, Duong and Duan, Zheng}},
issn = {{2090-4479}},
keywords = {{Iran; Meteorological variables; Renewable energy; Solar radiation}},
language = {{eng}},
number = {{1}},
publisher = {{Ain Shams University}},
series = {{Ain Shams Engineering Journal}},
title = {{Application of ERA-Interim, empirical models, and an artificial intelligence-based model for estimating daily solar radiation}},
url = {{http://dx.doi.org/10.1016/j.asej.2021.05.012}},
doi = {{10.1016/j.asej.2021.05.012}},
volume = {{13}},
year = {{2022}},
}