Uncertainty Analysis of Subsurface Temperature Estimates in the Snake River Plain, Idaho

Uncertainty of Snake River Plain (SRP) subsurface temperatures related to heat flow and a radiogenic heat production model is assessed. Heat flow data vary in abundance, spatial distribution, depth distribution, and quality, which make assessing the subsurface thermal regime difficult. We examined data attributes including quality, depth, location, fault proximity, and quantified heat flow interpolation variability. Results show initial data processing and selection of regionally representative data is most important for reducing interpolation uncertainty. Data removal using a 2.5 km buffer around faults is appropriate to inhibit potential advective heat flux while a 5 km buffer is too much, such that it produces random changes in the heat flow grid. The fault buffers change mapped heat flow by less than 20 percent in most areas of the SRP. Additional temperature-at-depth calculation uncertainty is associated with the radiogenic heat production model, which follows the heat flow – radiogenic heat production (Q-A) relationship. Model sensitivity to the radiogenic heat production model is tested by varying heat production layer thickness from 5 to 10 km and examining temperature change at 4 km. Temperatures vary by 10 percent (up to 27 °C), while most temperature is less than 5 percent different. Examination of heat flow, upper crustal thickness, and whole rock geochemistry revealed the Q-A relationship may be an oversimplification of the SRP thermal regime. Radiogenic heat production is approximately 12 to 40 mW/m² for this region. Mantle heat flux is set at 60 mW/m², which means measured heat flow above 100 mW/m² is unaccounted for in the Q-A relationship and must derive from an additional heat source. This anomalous heat flux may be heat refraction, advection, additional mantle heat flux, or a combination. Many data sites along the SRP margins and in the Eastern SRP contain anomalous heat flow, whereas the Western SRP is generally lower than the 100 mW/m² cutoff.