Kriging with an external drift versus collocated cokriging for water table mapping

Abstract

In aquifers driven by gravity, the water table resembles topographic features. Topographic data are easily obtainable at a low cost and geostatistics provides an adequate framework for incorporating exhaustive topographic secondary information to improve the estimate of a primary variable such as water table level. This incorporation results in models that better coincide with the natural phenomena being analysed. The present paper introduces collocated cokriging (multicollocated cokriging) and kriging with an external drift to incorporate topography as secondary information for mapping the water table level. The case study presented comes from a groundwater monitoring programme carried out in an underground coal mine. Incorporation of topography is aimed at diminishing the number of monitoring piezometers without deteriorating the maps of the estimates. The results of the present study show that both methodologies used improved the quality of the water table elevation maps when compared with ordinary kriging that does not introduce the topographic data. A slightly better performance obtained through collocated cokriging can be attributed to the spatial correlation taken into account by this methodology.