Seasonal rationalization of river water quality sampling locations: a comparative study of the modified Sanders and multivariate statistical approaches

Abstract

The design of surface water quality sampling location is a crucial decision-making process for rationalization of monitoring network. The quantity, quality, and types of available dataset (watershed characteristics and water quality data) may affect the selection of appropriate design methodology. The modified Sanders approach and multivariate statistical techniques [particularly factor analysis (FA)/principal component analysis (PCA)] are well-accepted and widely used techniques for design of sampling locations. However, their performance may vary significantly with quantity, quality, and types of available dataset. In this paper, an attempt has been made to evaluate performance of these techniques by accounting the effect of seasonal variation, under a situation of limited water quality data but extensive watershed characteristics information, as continuous and consistent river water quality data is usually difficult to obtain, whereas watershed information may be made available through application of geospatial techniques. A case study of Kali River, Western Uttar Pradesh, India, is selected for the analysis. The monitoring was carried out at 16 sampling locations. The discrete and diffuse pollution loads at different sampling sites were estimated and accounted using modified Sanders approach, whereas the monitored physical and chemical water quality parameters were utilized as inputs for FA/PCA. The designed optimum number of sampling locations for monsoon and non-monsoon seasons by modified Sanders approach are eight and seven while that for FA/PCA are eleven and nine, respectively. Less variation in the number and locations of designed sampling sites were obtained by both techniques, which shows stability of results. A geospatial analysis has also been carried out to check the significance of designed sampling location with respect to river basin characteristics and land use of the study area. Both methods are equally efficient; however, modified Sanders approach outperforms FA/PCA when limited water quality and extensive watershed information is available. The available water quality dataset is limited and FA/PCA-based approach fails to identify monitoring locations with higher variation, as these multivariate statistical approaches are data-driven. The priority/hierarchy and number of sampling sites designed by modified Sanders approach are well justified by the land use practices and observed river basin characteristics of the study area.

Evaluation of physicochemical and heavy metals characteristics in surface water under anthropogenic activities using multivariate statistical methods, Garra River, Ganges Basin, India

“…This is a consequence of the evaluation being centered on the data of water quality already acquired, ignoring the attributes of the river basin under investigation [13]. Despite this, PCA and HCA have been greatly recognized and widely used in monitoring plans, optimizing spatial sampling points and determining the most appropriate water quality factors [15][16][17][18][19][20][21][22][23][24].…”

Indicador da qualidade da água através de análise fatorial: um estudo de caso na zona portuária de Belém, Amazônia, Brasil

“…Essa característica de redução e simplificação da informação, fez com que essas análises multivariadas de dados fossem amplamente aceitas e utilizadas em diversos estudos de avaliação da qualidade das águas, que contempla parâmetros físicos, químicos e biológicos que por vezes se relacionam entre sí (ALVES et al, 2018;BONANSEA et al, 2015;BU et al, 2014;VAREKAR et al, 2016). Guedes et al (2012), Araújo Neto et al (2014), Ferreira et al (2015, Rocha et al (2016) e Alves et al (2018) também utilizaram a análise multivariada aplicada ao estudo da qualidade das águas no Rio Pomba (MG), bacia metropolitana do Ceará, semiárido brasileiro, bacia do rio Sergipe e manancial de Juiz de Fora respectivamente.…”

Monitoring water quality of the Sergipe River basin: an evaluation using multivariate data analysis

“…This is a result of the assessment being focused on water quality data already obtained, disregarding the characteristics of the river basin under study (KHALIL;OUARDA, 2009). Despite this, principal component analysis and hierarchical cluster analysis have been greatly accepted and they are being very used in monitoring programs, for spatial optimization of sampling locations and selection of the most suitable water quality variables (SINGH et al, 2004;OUYANG et al, 2006;SHRESTHA;KAZAMA, 2007;RAZMKHAH;ABRISHAMCHI;TORKIAN, 2010;ZHAO et al, 2011;WANG et al, 2012;CHEN et al, 2012;BONANSEA et al, 2015;BU et al, 2014;TANOS et al, 2015;PHUNG et al, 2015;VAREKAR et al, 2016).…”