Use of multivariate geostatistics to investigate time-dependent saltwater intrusion and perform groundwater quality zonation in a Nile Delta area (northern Egypt)

Freshwater salinization is a global problem that impacts water resources as well as ecosystems worldwide and is increasingly worsening. Possible causes multiple, whereas one of the most important is the saltwater intrusion, which is fostered by overexploitation of groundwater resources, in urbanized and cultivated coastal areas. This is the case of deltaic regions of large river systems, where porous multi-layer aquifers are present. Regular shapes and considerable amounts of water resources make them areas suitable for human activities. This leads to an increasing water demand, which provokes groundwater overexploitation and consequently freshwater salinization. Since deltaic aquifers are generally characterized by clear surface water/groundwater relationships, anthropogenic modification of surface water bodies (e.g., canals, and lakes) for irrigation and fish livestock purposes can contribute to salinization, by moving the freshwater/saltwater interface and altering the aquifer recharge. The eastern portion of the Nile Delta hosts a multi-layer aquifer, whose recharge inflow is represented by the Damietta branch and Ismailia canal. At NE, this area is bounded by Manzala Lake and Suez Canal, both saline water bodies. In the last decades, this area has experienced increasing groundwater exploitation for irrigation and drinking purposes, canal and lake modifications, and progressive groundwater salinization. This work aims at assessing the groundwater quality by partitioning the aquifer into different zones. Data refer to three surveys of chemical analyses on water samples, collected in the 1996-2018 period. The methodology is based on multivariate geostatistics using Multi-Collocated Co-Kriging (MCCK) and Multi-Collocated Factorial Kriging (MCFK) and considering distances from the different surface water bodies as exhaustive auxiliary covariates. The approach consists of four steps: 1) Gaussian Anamorphosis to transform raw variables into Gaussian transformed variables; 2) fitting a Linear Model of Co-regionalization on both experimental direct and cross-variograms; 3) estimation of the spatial distribution of the studied variables with MCCK; 4) extraction and interpolation of the sets of scale-dependent regionalized factors with MCFK. Maps of individual variables showed a clear decreasing trend in the N-W direction, due to a manifest process of saltwater intrusion. The saline front advanced southwards, over the years. Three zones were defined: the northern zone was characterized by the highest salinity values, the middle zone by the lowest values, and the southern zone by intermediate values as well as more pronounced variability. This partitioning does not follow a linear trend and is indicative of the complexity of processes and interactions occurring in groundwater