Best applicable geostatistical model for interpolating groundwater-levels in El-Obour city, Egypt

Abstract

The lack of data, mainly in the field of hydrogeology and groundwater especially, makes it necessary to implement a special estimation technique to overcome this shortage. Nowadays, geostatistics have become a popular mean to describe spatial patterns and to predict the value of unmeasured locations (Kriging). Many types of Kriging could be used such as ordinary, simple, universal Kriging. Each type has a different underlying assumption. The aim of this paper is to put a methodology that could be used for identifying the most suitable geostatistical model for interpolating groundwater-level data collected from District VI, El-Obour city using geographical information systems (GIS) to generate a prediction error map with good accuracy that could be used further on designing and/or optimizing the existing monitoring network. The procedure has to 1) build all available geostatistical models depending on Hierarchy Stepwise criteria, 2) compare them using cross-validation statistical parameters, and 3) choose the best suitable model for the study area. It was found that 1) the ordinary Kriging method, 2) fitted with a Gaussian function, 3) of groundwater-level skewed data normalized using Box-Cox with a power parameter equal to 2, and 4) no trend removal, all of them together compose the best matched model for the studied area.