Modelling spatial and temporal variability of water quality from different monitoring stations using mixed effects model theory

Floodplains and polder-type agricultural catchments, arespecific land formations which management is highly demanding fromseveral aspects. Water pumping system affects and modifies naturalhydrology, influencing thus surface and groundwater quality. The closecontact with the coastal sea may affect the quality of adjacent marineenvironment as well. The study uses the case of the Lower Neretva Valley(LNV) to test the efficiency of applying Linear Mixed Effect (LME) theoryin modelling spatial and temporal variations of surface and groundwaterquality. Sea water intrusion and nutrients entry from polders' drainagesystem into the adjacent water system of the study area exhibitpronounced variability in space and time. Therefore, the methodologycombined big spatial-temporal data from the long-term water qualitymonitoring program and specific (geo)statistical methods to provide aglobal model by effectively fusing spatial and temporal information whichcould be used for site-specific environmental management. The objectivewas to assess the impact of natural processes and human activities onwater quality. A dataset of physicochemical properties of surface andgroundwater quality of the LNV, recorded monthly in the period 2009-2017,was used to model the spatial and temporal variations of water salinityand nitrate concentrations. The network of water quality monitoring sitescovers four polders on five thousand hectares of agricultural land,including four classes of water bodies in each: river streams, lateralcanals, pumping stations, drainage canals and groundwater. The method ofdata analysis used is based on LME theory with correlated spatial andtemporal residuals, which takes also into account the heterogeneity ofthe variance associated with each type of water quality monitoringstation. The results show that the main source of variation are sea waterintrusion and agricultural activities and variations are significantlydepending on the seasons. LME model could be effectively used for sitespecificenvironmental management.