Multivariate and seasonal optimization of surface water quality monitoring: a case study in an arid-semiarid watershed

ABSTRACT

Graphical Abstract: A three-part infographic summarizing the optimized river water quality monitoring network design for the Dehloran watershed. (Left) A flowchart illustrating the research methodology steps, including field sampling, multivariate statistical analysis, and optimization. (Center) A DEM map of the Dehloran watershed showing the river network and the 62 sampling stations, with Station 38 highlighted as a salinity hotspot. (Right) A conceptual illustration showing that optimizing the monitoring network (reducing stations from 62 to 18) leads to a 75% reduction in analytical costs.

This study, aiming to design an efficient river water quality monitoring network in the Dehloran watershed using multivariate statistics, applied an integrated approach with a seasonal framework. Measurements encompassing 34 water quality parameters were conducted at 62 stations during both the dry (September 2020) and wet (March 2021) seasons. Hierarchical cluster analysis using the UPGMA algorithm and squared Euclidean distance was conducted in Minitab to identify station similarities. Among the measured parameters, 24 showed statistically significant seasonal variation; in addition, 11 parameters (EC, SO4, Ca, Mg, Cr, Cu, As, Al, DO2, total hardness, and chlorophyll) differed significantly across both spatial and temporal dimensions. Given the presence of evaporitic formations and saline/sulfur springs, EC was further analyzed, revealing Station 38 as a persistent salinity hotspot, independent of seasonal change. Based on the cluster analysis dendrogram, 18 representative stations were selected from the original 62, optimizing network coverage while reducing redundancy. The targeted selection of Station 38, located at the confluence of several polluted flows, underscores its importance as a key monitoring point. Furthermore, parameters with minimal seasonal variation were recommended for annual instead of biannual monitoring, leading to reduced laboratory costs – with an estimated 75% decrease in analytical expenditures.