Practical considerations for kriging groundwater surfaces

Abstract

Kriging is an accepted method of characterizing the groundwater elevation surface at sites where the water level data are available but where there may be insufficient additional data necessary for groundwater flow modeling. Groundwater surface interpolation via kriging is readily performed using commercial, state of the practice software, but some practitioners may not be able to justify such efforts because the process is not validated within studies documented in the peer‐reviewed literature. This paper describes the available kriging software and literature studies and then uses a case study to compare practical groundwater surface modeling to the studies available in the scientific literature. The literature review shows that the state of the practice as represented by the commercial software approach is consistent with the literature. Specifically, cokriging with groundwater elevation as the primary variable with trend removal and ground‐surface elevation as the secondary variable is an appropriate point of departure in practice. The literature review‐based summary of variogram model parameters (model type, nugget, sill, range, and trend model) was not useful as a quality‐control step to assess the reasonableness of variogram parameters identified by the standard practice of software‐assisted iteration when applied to the case‐study data set. The literature review indicated that groundwater elevation kriging has been performed using as few as 10 data points but a comparison of the case‐study simulated groundwater elevations and groundwater gradient magnitudes and directions indicated that the 30‐well threshold more commonly found in the literature was an appropriate minimum at the study site.