Rates and patterns of groundwater flow in the Cascade Range Volcanic Arc, and the effect on subsurface temperatures

Abstract

The central Oregon section of the Cascade Range volcanic arc is characterized by relatively high Quaternary volcanic extrusion rates and hot‐spring discharge rates, and by high conductive heat flow. However, a large area of near‐ zero near‐surface conductive heat flow occurs in the younger volcanic rocks, due to downward and lateral flow of cold groundwater. Alternate models for the high heat flow observed in older rocks on the flanks of the Cascade Range involve (1) a laterally extensive midcrustal heat source or (2) a narrower, spottier deep heat source that is confined to the Quaternary arc and is flanked by relatively shallow conductive heat flow anomalies caused by regional groundwater flow. We simulated groundwater flow and heat transport through two cross sections west of the Cascade Range crest: one in the Breitenbush area, where there is no major arc‐parallel normal faulting, and one in the McKenzie River drainage, where major graben‐bounding faults exist. Measured temperature profiles, hot‐spring discharge rates, and geochemical inferences constrain the results. The numerical simulations provide some estimates of regional‐scale permeabilities; simulated bulk permeabilities of ∼10−14 m2 in the youngest (0–2.3 Ma) rocks and ∼10−17 m2 in the oldest (18–25 Ma) rocks allow the thermal observations to be matched. In general, permeability decreases downsection, but for rocks of any age, permeability at very shallow (<50 m) depths is probably much higher than the bulk permeability values required by the thermal observations: this is indicated by high recharge rates in 0–7 Ma rocks (>1 m yr−1) and well‐test data from domestic wells in rocks older than 7 Ma (which indicate permeability values of about 10−14 to 10−12 m2). In the simulations, the alternate conceptual models for the deep thermal structure were represented as wide or localized deep heat sources. We found that either model can satisfy the observations. Thermal observations in the Breitenbush area seem to require significant advective heat transfer, whereas the sparser observations in the McKenzie River area can be satisfied with either advection‐ or conduction‐dominated simulations. Available regional gravity, magnetic, and electrical geophysical data do not clearly favor either of the two alternate models. Deep drilling in areas of high heat flow in the older rocks would be the most definitive test. The actual thermal structure is probably more complex than either of the models considered here.