Geothermal resource favorability: select features and predictions for the western United States curated for DOI 10.1016/j.geothermics.2023.102662

The data contained herein are five input features (i.e., heat flow, distance to the nearest quaternary fault, distance to the nearest quaternary magma body, seismic event density, maximum horizontal stress) and labels (i.e., where known geothermal systems have been identified) from Williams and DeAngelo (2008) and nine favorability maps from Mordensky et al. (2023). The favorability maps are the untransformed predictions from models resulting from the features and labels used with either the methods presented in Williams and DeAngelo (2008) or the machine learning approaches presented in Mordensky et al. (2023). Each favorability map depicts an estimate of relative favorability with respect to the other locations (i.e., cells), allowing for a comparison of the influence the different methods and machine learning approaches produced when predicting geothermal favorability. The machine learning approaches sought to minimize the influence of expert bias imparted by the methods from Williams and DeAngelo (2008). The favorability maps presented from the models that used the methods from Williams and DeAngelo (2008) are provided for comparative purposes.