The application of predictive analytics for hydrocarbon exploration in the Denver-Julesburg Basin

The Denver-Julesburg (D-J) Basin lies on the eastern side of the Rocky Mountains and extends from south of Denver northward into southeast Wyoming, and eastward into western Nebraska and western Kansas, blanketing nearly 30,000 square miles along the Front Range. A decades-old USGS study estimated that nearly two billion BOE have been produced from more than 20,000 D-J wells. Major oil-producing areas in the D-J’s Niobrara Formation include the Hereford field in the northwest, the East Pony field in the northeast, and the Wattenberg field in the center.

We were engaged to conduct a multi-physics integrated study over a 3,000 square-mile portion of the D-J Basin. The objective of the study, which included acquisition of new airborne geophysical data, was to provide a regional view into the subsurface from basement through the target reservoir intervals and to highlight potential sweet spots within the Niobrara. To do this, we applied multi-physics interpretation and integration techniques to map key geologic features, including mapping basement topography, along with igneous intrusions and deep basement faults, both of which are believed to act as conduits for hydrothermal fluids that impact the thermal maturity of the Niobrara and the sweet spots in the play.

In order to build a sweet spot map of the area of interest through predictive analytics, we gathered and analyzed the historical production data from several thousand wells. We isolated a subset of these wells and selected a measure of early oil production to be used as training data for our machine learning algorithms. The predictor variables for this study consisted of a set of 2-D regional grids derived from selected geophysical attributes (e.g., gravity, magnetic, well log resistivity) and interpreted grids (e.g., isopach, structure, total organic carbon). To model the complex relationships between the predictor-variable grids and the training production data, we selected the Alternating Conditional Expectations (ACE) approach. ACE allows a prediction of production to be driven not only by the well production data but also the combination of all available information for the full area of interest.

The result is a map predicting the early oil production over the full area of interest. Whereas the prediction in densely drilled regions is mostly driven by the well information, away from wells, the prediction takes full advantage of the predictor variables to draw an estimate of the production. In particular toward the southern end of the studied polygon, where only a few wells have been drilled, a high predicted production is shown. To the south of the Wattenberg field is an area with low predicted production.


Denver-Julesburg Basin, Colorado

CUSTOMER Large E&P Operator

FOCUS Regional Mapping

TYPE Unconventional







Predictive Analytics


  • Regional resistivity voxels down to 10,000 ft.
  • Basement-to-surface maps
  • Maps of the key reservoir interval of interest
  • Regional sweet spot maps
  • Map of early oil production


Utilized predictive analytics and production data to provide a regional view into the subsurface, highlighting potential sweet spots within the Niobrara.


The Leading EdgeThe use of predictive analytics for hydrocarbon exploration in the Denver-Julesburg Basin, March 2017