On the kriging of water table elevations using collateral information from a digital elevation model

Abstract

In unconfined aquifers flowing under topographic gradients, the water table is a subdued replica of the ground surface above. This principle is the basis for using detailed collateral or secondary information from digital elevation models to supplement sparse observations from water wells in the mapping of phreatic surfaces. Data from DEM-derived secondary variables are incorporated into the estimation of water table elevations using the geostatistical method known as kriging with an external drift (KED). Two different KED models are proposed based on the choice of secondary variable. In the first, water table elevation is expressed as the sum of a deterministic trend given by topographic elevation and a residual random component representing depth to water table. In the second, depth to water table is expressed as a linear function of a deterministic trend, given by the TOPMODEL topographic index, and a residual random error. The relationship between water table depth and topographic index is derived from simplified groundwater dynamics and forms the basis of TOPMODEL-type rainfall-runoff models. The two KED models are applied to the mapping of water table elevations in the Oak Ridges Moraine (ORM), an unconfined aquifer near Toronto, Canada. Results show that KED with topographic elevation as external drift is the more robust of the two models. Despite its strong theoretical basis, the second model yields kriged water table elevations that are not always physically plausible. In part, this is because field observations of water table depth do not verify the predicated relationship with topographic index in large parts of the study area. However, this relationship may be valid in other cases and at other spatial scales. In such cases, the second model would provide a very powerful approach for mapping water table elevations and for calibrating distributed parameters of the TOPMODEL equations on water well observations.