Using a linear discriminant analysis (LDA)-based nomenclature system and self-organizing maps (SOM) for spatiotemporal assessment of groundwater quality in a coastal aquifer
Amiri, Vahab and Nakagawa, Kei LU
(2021)
In Journal of Hydrology
603.
- Abstract
In this study, a linear discriminant analysis (LDA) - based nomenclature system have been used for the classification of groundwater samples in a coastal aquifer. The capability of three models (7 hlr, 7 M conc, and Greater molar conc) in determining the water types has been investigated. The results show that Ca-HCO3, Na-Cl, and Na-HCO3; and Ca-HCO3, Na-HCO3, and Na-Cl are the three dominant water types for the wet and dry seasons, respectively. The results of 7 hlr (basic + hybrid) model as the best option show that 42 and 27 water types can be identified for wet and dry seasons, respectively. Twenty-four physicochemical components have been used for the hydrogeochemical study of selected... (More)
In this study, a linear discriminant analysis (LDA) - based nomenclature system have been used for the classification of groundwater samples in a coastal aquifer. The capability of three models (7 hlr, 7 M conc, and Greater molar conc) in determining the water types has been investigated. The results show that Ca-HCO3, Na-Cl, and Na-HCO3; and Ca-HCO3, Na-HCO3, and Na-Cl are the three dominant water types for the wet and dry seasons, respectively. The results of 7 hlr (basic + hybrid) model as the best option show that 42 and 27 water types can be identified for wet and dry seasons, respectively. Twenty-four physicochemical components have been used for the hydrogeochemical study of selected water resources based on Kohonen's self-organizing map (SOM) method. The k-means clustering tool detected 7 and 6 clusters using the Davies-Bouldin index (DBI) for the wet and dry seasons, respectively. Hydrogeochemical assessment of duplicated samples showed that the seasonal changes can not cause significant changes in the chemical composition of groundwater resources of this aquifer. Radar diagrams of physicochemical variables showed some changes in the chemical composition of groundwater in some locations due to water–rock interaction as well as pollutants produced by human activities. Comparison of the results of classification of water samples by SOM and principal component analysis (PCA) shows that both methods can perform clustering of samples correctly. However, the SOM method provides more accurate clustering without any overlap between different clusters.
(Less)
- author
- Amiri, Vahab
and Nakagawa, Kei
LU
- publishing date
- 2021-12
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Coastal aquifer, Groundwater classification, K-means clustering, Linear discriminant analysis, Principal component analysis, Self-organizing maps
- in
- Journal of Hydrology
- volume
- 603
- article number
- 127082
- publisher
- Elsevier
- external identifiers
-
- scopus:85117562746
- ISSN
- 0022-1694
- DOI
- 10.1016/j.jhydrol.2021.127082
- language
- English
- LU publication?
- no
- additional info
- Funding Information: This work is financially supported by the Geological Survey of Iran (GSI). We would like to thank all of the members of the GSI, especially Dr. Razyeh Lak, for their kind cooperation that made this research possible. Publisher Copyright: © 2021 Elsevier B.V.
- id
- 72ab4d52-50cd-4605-bd14-45ba23f2effa
- date added to LUP
- 2021-12-12 07:27:13
- date last changed
- 2022-04-27 06:33:17
@article{72ab4d52-50cd-4605-bd14-45ba23f2effa,
abstract = {{<p>In this study, a linear discriminant analysis (LDA) - based nomenclature system have been used for the classification of groundwater samples in a coastal aquifer. The capability of three models (7 hlr, 7 M conc, and Greater molar conc) in determining the water types has been investigated. The results show that Ca-HCO<sub>3</sub>, Na-Cl, and Na-HCO<sub>3</sub>; and Ca-HCO<sub>3</sub>, Na-HCO<sub>3</sub>, and Na-Cl are the three dominant water types for the wet and dry seasons, respectively. The results of 7 hlr (basic + hybrid) model as the best option show that 42 and 27 water types can be identified for wet and dry seasons, respectively. Twenty-four physicochemical components have been used for the hydrogeochemical study of selected water resources based on Kohonen's self-organizing map (SOM) method. The k-means clustering tool detected 7 and 6 clusters using the Davies-Bouldin index (DBI) for the wet and dry seasons, respectively. Hydrogeochemical assessment of duplicated samples showed that the seasonal changes can not cause significant changes in the chemical composition of groundwater resources of this aquifer. Radar diagrams of physicochemical variables showed some changes in the chemical composition of groundwater in some locations due to water–rock interaction as well as pollutants produced by human activities. Comparison of the results of classification of water samples by SOM and principal component analysis (PCA) shows that both methods can perform clustering of samples correctly. However, the SOM method provides more accurate clustering without any overlap between different clusters.</p>}},
author = {{Amiri, Vahab and Nakagawa, Kei}},
issn = {{0022-1694}},
keywords = {{Coastal aquifer; Groundwater classification; K-means clustering; Linear discriminant analysis; Principal component analysis; Self-organizing maps}},
language = {{eng}},
publisher = {{Elsevier}},
series = {{Journal of Hydrology}},
title = {{Using a linear discriminant analysis (LDA)-based nomenclature system and self-organizing maps (SOM) for spatiotemporal assessment of groundwater quality in a coastal aquifer}},
url = {{http://dx.doi.org/10.1016/j.jhydrol.2021.127082}},
doi = {{10.1016/j.jhydrol.2021.127082}},
volume = {{603}},
year = {{2021}},
}