Using one-way clustering and co-clustering methods to reveal spatio-temporal patterns and controlling factors of groundwater geochemistry

Abstract

Effectively and efficiently protecting groundwater resources requires a better understanding of spatial and temporal patterns of groundwater geochemistry and their controlling factors (natural and anthropogenic). In this study, to reveal the spatial and temporal patterns of groundwater geochemistry, we explored two clustering methods, i.e., the one-way clustering implemented by using the hierarchical cluster analysis (HCA) and the co-clustering implemented by using the Bregman block average co-clustering algorithm with I-divergence (BBAC_I). The one-way clustering identifies the spatial and temporal patterns separately, whereas the co-clustering identifies the spatial and temporal patterns simultaneously. The two clustering methods were applied to 9 geochemical parameters of 329 groundwater samples collected over the period of 2000 – 2017 from 19 monitoring wells located in the unconfined aquifer of the Dagu River Basin, China. As a first attempt of applying BBAC_I co-clustering to groundwater geochemical data, results show that it is feasible to use the BBAC_I co-clustering method for simultaneously identifying location and timestamp clusters. The clusters identified by the HCA one-way clustering and the BBAC_I co-clustering are similar. Based on the clustering results, two geochemical zones and three geochemical periods of groundwater geochemistry were delineated. The controlling factors of the two geochemical zones are residual seawater freshening, elevated NO3− concentrations, and silicate weathering. The temporal changes between the three geochemical periods are caused by a heavy rainfall in July 2011 after a one-year drought and by ceased flow of the Dagu River after 2013 due to drought conditions. The impacts of drought conditions and anthropogenic activities in the Dagu River Basin on groundwater geochemistry should be considered for managing groundwater resources of the basin.