Data Sources


For the CPRAS study, mapped surficial geology and well data sources that have a classification scheme sufficient to identify the major geologic model units were used. Rather than selecting a subset of the “best” data for an exact match, all data were used in the investigative process, and data were removed from the dataset only when the data point was shown to be nonrepresentative to such an extent that it resulted in significant bias that skewed the results. If only a few data are substantially different from a larger set of proximal data and no reasonable geologic explanation, such as a fault, explains the difference, then the outlier data are removed. All other data, including suspect data, are retained, and uncertainty and error are quantified. 


For the work by Kahle and others (2009), a well dataset was prepared, a simplified geologic map was compiled, and a generalized map of hydrogeologic units and generalized hydrogeologic sections was constructed (Kahle and others, 2009). These data have been supplemented and reinterpreted for the current work. The well data were supplemented as described in section, “Well Data.” Compilation and further simplification of the geologic map and the extraction of geologic information from the map for interpolation is described in section, “Geologic Map.”


Well Data


All relevant sources of data should be considered when formulating or updating a geologic model. For this reason, the 2,523 wells used by Kahle and others (2009) were supplemented with data published by others, resulting in a significantly larger set of 13,226 wells with relevant lithologic picks (fig. 1B). For non-USGS studies (Leek, 2006; Lindsey and others, 2007; and Tolan and others, 2007), data were used as published. All USGS data were reviewed, and the datasets were rectified wherever studies had conflicting picks for stratigraphy. Lithologic picks for wells made in the more recent study were assumed to be correct where they differed from picks from earlier studies, subject to the assumption that newer USGS interpretations took the older data into account when reinterpreting the data. USGS study wells [RASA wells updated by Kahle and others (2009), Yakima study wells (Jones and others, 2006; Jones and Vaccaro, 2008), and gas resources assessment wells (Wilson and others, 2008)] were supplemented with any additional wells from the USGS NWIS database (U.S. Geological Survey, 2009b) for which the thickness of the Overburden unit was estimated.


Geologic Map


If a geologic contact shown on a surficial geology map is representative of the top of a map unit, then the topographic elevation along that line represents the elevation of the top of the geologic unit. Furthermore, in areas where the geologic unit is exposed and not significantly eroded, those elevations also represent samples of the pre-erosional top of unit that may be used to interpret relational geometries of geologic units. The map-view relationships of geologic units used for this work are shown in figure 1A. These units are simplified (fig. 2) even further than was done by Kahle and others (2009).


The foundation for the model is a simplified geologic map of the Columbia Plateau (Kahle and others, 2009), which was further simplified by grouping all CRBG units older than the Wanapum Basalt into the Grande Ronde Basalt unit (fig. 2). These units include the Grande Ronde, Picture Gorge, Prineville, and Imnaha Basalts. On the basis of location, most undifferentiated Miocene age basalts appear to be one of these older units, so these undifferentiated units also were grouped into the Grande Ronde Basalt unit. All sedimentary overburden units also were grouped for the geologic model. 


Given these simplifications, the only sizable areas not defined lithologically for the model (fig. 1A) are mapped as Quaternary volcanic deposits. These data were not used in the model because of the limited areal extent, and because the modeling process will assign reasonable geologic and hydrogeologic properties to these locations during modeling of the sedimentary Overburden unit. If these deposits occur in lowlands near wells with thick sedimentary deposits, then the deposits likely overlie the sediments and will be simulated as having subsurface flow properties of sedimentary deposits. Otherwise, the unit will be represented as one of the Columbia River Basalts and flow properties will be those consistent with lava.


Mapped surficial geology contact lines that represent tops of units were converted to point data by sampling the land-surface elevation at closely spaced points along the line. Because the line is a smooth line traversing undulating terrain, the elevation along the contact varies. It was assumed that on average the line is correct, so a local median value was extracted as the best estimator of elevation along the line. Sample frequency is arbitrary, because the geologic contact line may be sampled as often as desired. The key consideration when selecting surficial geology sample frequency is that the final trend model should be influenced by all available data. If surficial geology is sampled too frequently, then the influence of well points is diminished because there are fewer well points. This bias is partly offset by the tri-cube weight function used by loess when well samples are far from surficial geology points, but is a larger problem when wells are near mapped geologic contacts. A trial-and-error approach was followed, and a final ratio of about 1:10 of wells to surficial geology points was selected, allowing a good fit to both wells and surficial geology. This estimate of the ratio does not hold in every interpolation group for the piecewise interpolation. Some groups have little well data, whereas others have little or no surficial geologic outcrop data for the unit being interpolated.


Other Interpreted Data


With an estimated maximum thickness of approximately 16,500 ft near the center of the basin, few wells penetrate the entire thickness of the Columbia River Basalts over much of the Columbia Plateau. In order to estimate the top of the Older Bedrock unit for the entire model area, well data were supplemented with thickness estimates made by Reidel and others (2002) (see section, “Using Thickness Maps for Quality Assurance and Model Revision”). 


First posted February 25, 2011