Efficient and Transparent Construction of 3D Hydro-stratigraphy from Geophysical and Geological Data

Thursday, 27 July 2017: 10:30 AM
Paul Brest West (Munger Conference Center)
Anders Vest Christiansen1, Troels N Vilhelmsen1, Esben Auken1, Nikolaj Foged1 and Pernille Aabye Marker2, (1)Aarhus University, Department of Geoscience, Aarhus C, Denmark, (2)Technical University of Denmark, Department of Environmental Engineering, Lyngby, Denmark
We present an automatic method for delineation of a 3D structure model, integrating lithological information from boreholes with resistivity models. The objective is to create direct input to groundwater models for areas, where the sand/clay distribution governs groundwater flow and the sand/clay have distinguishable resistivities. The coupling between hydrological and geophysical parameters is made with a spatially variable translation calibrated against observed lithological data. The translator “interprets” the geophysical resistivities into clay fractions (CF) at sounding sites. Using k-means clustering, the resistivities and CF are grouped into a subset of units/clusters (typically between 3 and 5). Once subdivided, the clusters can be treated as indicator variables in indicator kriging, and the probability of each cluster can be calculated everywhere in the subsurface. From this probability distribution, the most likely 3D zonation of the subsurface can be determined. By combining this hydro-stratigraphical zonation with the indicator probabilities several realizations of the full hydro-stratigraphical model domain are created with a multiple-point statistic approach (SNESIM). Lastly these realizations are calibrated in a groundwater model.

We present the methodology by show-casing a study from Denmark were a groundwater model is constructed, and structural realizatinos are generated by including SkyTEM data and lithological information from 621 boreholes covering an 75 sqkm area.

The main benefits of the proposed methodology are 1) the transparent and objective workflow, 2) the possibility to make uncertainty estimates on e.g well catchment areas, and 3) the reduced time for the hydrological model development. The latter is reduced to a few weeks.

Figure 1: Model realizations: a) shows a model realization with three hydrological units based on a manual interpretation of the geophysical and borehole data; b) is most likely representation using the automatic approach described without any manual interpretation of geology. The two images are very much alike.