Subaqueous geology and a filling model for Crater Lake, Oregon

Abstract

Results of a detailed bathymetric survey of Crater Lake conducted in 2000, combined with previous results of submersible and dredge sampling, form the basis for a geologic map of the lake floor and a model for the filling of Crater Lake with water. The most prominent landforms beneath the surface of Crater Lake are andesite volcanoes that were active as the lake was filling with water, following caldera collapse during the climactic eruption of Mount Mazama 7700 cal. yr B.P. The Wizard Island volcano is the largest and probably was active longest, ceasing eruptions when the lake was 80 m lower than present. East of Wizard Island is the central platform volcano and related lava flow fields on the caldera floor. Merriam Cone is a symmetrical andesitic volcano that apparently was constructed subaqueously during the same period as the Wizard Island and central platform volcanoes. The youngest postcaldera volcanic feature is a small rhyodacite dome on the east flank of the Wizard Island edifice that dates from 4800 cal. yr B.P. The bathymetry also yields information on bedrock outcrops and talus/debris slopes of the caldera walls. Gravity flows transport sediment from wall sources to the deep basins of the lake. Several debris-avalanche deposits, containing blocks up to 280 m long, are present on the caldera floor and occur below major embayments in the caldera walls. Geothermal phenomena on the lake floor are bacterial mats, pools of solute-rich warm water, and fossil subaqueous hot spring deposits. Lake level is maintained by a balance between precipitation and inflow versus evaporation and leakage. High-resolution bathymetry reveals a series of up to nine drowned beaches in the upper 30 m of the lake that we propose reflect stillstands subsequent to filling of Crater Lake. A prominent wave-cut platform between 4 m depth and present lake level that commonly is up to 40 m wide suggests that the surface of Crater Lake has been at this elevation for a very long time. Lake level apparently is limited by leakage through a permeable layer in the northeast caldera wall. The deepest drowned beach approximately corresponds to the base of the permeable layer. Among a group of lake filling models, our preferred one is constrained by the drowned beaches, the permeable layer in the caldera wall, and paleoclimatic data. We used a precipitation rate 70% of modern as a limiting case. Satisfactory models require leakage to be proportional to elevation and the best fit model has a linear combination of 45% leakage proportional to elevation and 55% of leakage proportional to elevation above the base of the permeable layer. At modern precipitation rates, the lake would have taken 420 yr to fill, or a maximum of 740 yr if precipitation was 70% of the modern value. The filling model provides a chronology for prehistoric passage zones on postcaldera volcanoes that ceased erupting before the lake was filled.