Evaluation of two linear kriging methods for piezometric levels interpolation and a framework for upgrading groundwater level monitoring network in Ghiss-Nekor plain, north-eastern Morocco

Groundwater levels serve as a monitoring parameter of changes in groundwater storage and vulnerability status of coastal aquifers to seawater intrusion. Secondary piezometric levels datasets obtained in dry and wet seasons of 2017 from 12 dedicated observation wells in Ghiss-Nekor coastal plain were used to conduct this study. The measured data are reliable; however, investigated sampling locations are irregularly distributed and clustered along the northern half of the plain. Our research examined the performance of two linear kriging methods: empirical Bayesian kriging (EBK) and ordinary kriging (OK), regarding interpolation of this scattered dataset. Cross-validation results for assessing the prediction accuracy approved the selection of EBK as the best-fit method. By adopting interpolated EBK-based estimates, accurate groundwater levels distribution maps were developed enabling the delineation of zones at the highest risk of seawater intrusion or occurrence of slight piezometric recovery. Besides yielding satisfactory outcome for small number of observations, EBK technique generates standard errors associated with the predicted values; the latter were included in our study as a criterion for selecting locations of high uncertainty of estimates and requiring an increase in the number of wells. By considering the results of the Densify Sampling Network tool and the highest priority indices, 26 new wells are proposed to be added to the investigated observation network, with the possibility of excluding wells of least priority. Usage of DSN tool and priority index, relying respectively on the standard error of prediction surface and cross-validation residuals, is deemed satisfactory as the mean standard error diminished by over 60% after placing the additional monitoring sites.