This dissertation seeks to characterize the cave atmospheres and dynamics of fumarolic ice caves. The introduction presents a broad framework for understanding the caves and describes the historical and conservation context into which the work fits. This framework provides the motivation for five investigations which are presented as Chapters 2 through 6. Chapter 2 details a fiber-optic distributed temperature sensing (FODTS) experiment in which 438m of fiber-optic cable was deployed along the main passages of Warren Cave on Erebus Volcano, Antarctica. Point sources of warm gas flowing into the cave manifested as multi-degree C temperature anomalies and persisted throughout the weeklong experiment. Observed temperatures were anti-correlated with local atmospheric pressure, indicating barometric pumping of the gas vents. Chapter 3 extends the FODTS technique used in Chapter 2 to three dimensions for volumetric imaging of the temperature field inside a fumarolic ice cave chamber. Using terrestrial laser scanning (TLS) and automatic pointcloud classification techniques, I precisely located each virtual temperature sensor along the fiber optic cable. Interpolation and analysis of spatial patterns revealed a strong, upward-positive temperature gradient which averaged 0.265C m-1 over the 7 day experiment. I used satellite data and a permafrost model to assess potential Holocene volcano ice interaction globally, finding that 19.8% of known Holocene volcanic centers host glaciers or areas of permanent snow. The results, presented in Chapter 4, suggest that fumarolic ice caves are globally widespread and largely undiscovered. Fumarolic ice caves are expected to form when degassing begins beneath any volcano with moderate ice overburden. In Chapter 5, I present six years of morphological observations using TLS, structure from motion (SfM), and traditional cave survey, revealing that fumarolic ice caves change on the scale of tens of centimeters annually, and that the topography above the caves responds to enlargement of chambers through melting. I find that the cave wall icehas passed the pore-closeoff density, and conclude that densification is accelerated by heat from the cave. The rapid passage enlargement observed means that fresh rock substrate regularly becomes available to the cave microbial communities. For theoretical context, I developed two “toy” models. A computational fluid dynamics (CFD) simulation of cave melt is presented which represents a cave during initiation of growth.A simple flow model based on Glen's flow law, gives a first estimate of expected passage closure rates due to ice creep. Chapter 6 represents a collaborative effort to characterize the isotopic and chemical composition (δ2H and δ18O) of Erebus' snow and ice mantle which hosts the fumarolic ice caves. We found that snow samples from the entire summit caldera area, including ice cores collected through fumarolic ice tower walls, fall far outside an Antarctic Meteoric Water Field which encompasses all other available Antarctic snow isotope data. This suggests a magmatic component in the snow, which may be supplied by the plume emanating from Erebus' main crater. Several cross-cutting themes are addressed in multiple chapters. I discuss how fumarolic ice caves provide important indicators of volcanic unrest, analogues of extraterrestrial systems, and critical habitats for microbes. Going forward, this dissertation should be a foundation on which to plan the further exploration of fumarolic ice caves on Earth and elsewhere in the solar system. Keywords: distributed temperature sensing, LiDAR, isotopes, glaciovolcanism, flank degassing, Erebus, Antarctica.

Raw project data is available by contacting ctemps@unr.edu