The Little Grand Wash fault and the Salt Wash Graben in the Colorado Plateau of southeastern Utah emit CO2 gas from abandoned drillholes, springs, and a hydrocarbon seep. Similar CO2-charged water has also been emitted in the past, as shown by large localized travertine deposits and veins along and near the fault traces. The faults cut natural CO2 reservoirs and provide an excellent analog for geologic CO2 sequestration. The faults cut a north-plunging anticline of rocks consisting of siltstones, shales, and sandstones from the Permian Cutler Formation through the Cretaceous Mancos Shale. The Little Grand Wash fault has 260 m of throw and the stratigraphic separation across the Salt Wash Graben is 50 m. The fault rocks in the damage zone show hundreds of fractures, which decrease in density farther away from the faults. In specific areas, fractures with the presence of calcite mineralization indicate fluid migration and bleach zones from a few millimeters to 30 cm. This is evidence of past fluid migration directly associated with the fault zone. Calcite mineralization fills these fractures and is also deposited in a variety of other bed forms. Foliated fault gouge, 5 to 20 cm thick, forms clay smear structures with a scaly shear fabric in a zone l0 to 15 cm thick is seen in the fault core. The leakage is constrained to the footwalls of the northernmost faults throughout the area. Clay-rich gouge structures should be effective barriers to cross-fault flow . Well log, surface geologic, and geochemical data indicate that the CO2 reservoirs have been cut by the faults at depth, providing a conduit for the vertical migration of CO2 to the surface, but not for horizontal flow across the fault plane. Even though lateral cross-fault migration may be impeded, this study clearly indicates that there are possible migration pathways for the escape of CO2 from faulted subsurface aquifers, including aquifers faulted by "low-permeability" faults with clay gouge. Three-dimensional flow models show how the fault's maximum permeability in the damage zone is parallel to the faults, and the leakage though the damage zone is localized near the fold axis of the regional anticline. Direct dating of the clay in the fault gouge was done by ExxonMobil with 40Ar/39Ar methods, indicating that fault movement occurred between the middle Eocene and the end of the Miocene. During this time, the Colorado Plateau is interpreted to have been experiencing rapid uplift. The middle Jurassic, upper Jurassic, and Cretaceous rocks at the surface have been uplifted approximately 1.8 km since the end of the Eocene. This uplift may have influenced fault movement in the Colorado Plateau and along the Little Grand Wash fault, and Salt Wash and Ten Mile Graben. In evaluating these deep aquifers for CO2 sequestration, careful design and monitoring of the geological structure and stress regimes must be considered to avoid leakage.

The Little Grand Wash normal fault near Green River, eastern Utah, hosts a series of naturally occurring CO₂ seeps in the form of active and extinct CO₂-charged springs distributed along the fault zone. I have studied the association of fault structure with CO₂-related alteration as an analogue for the long-term (1,000- to 10,000-year) effects of leakage through faults in CO₂ sequestration reservoirs. Structure and alteration in a portion of the Little Grand Wash fault zone were mapped at a 1:700 scale in order to determine the association of faulting with CO₂-related diagenesis. I combined structural and diagenetic mapping were combined with laboratory analyses of mineralogical, isotopic and textural changes in order to assess controls on the migration of CO₂ traveling up the fault and its effects on the fault itself. The fault zone is 200 m wide at its widest and contains 4-5 major subparallel fault segments that form multiple soft- and hard-linked relay ramps. The area includes a travertine deposit and related sandstone alteration: outcrop-visible coloration, porosity-occluding calcite cement and veins occasionally so abundant that they obliterate the rock fabric. Structural mapping shows that the travertine is located at an intersection of major fault segments constituting the hard link of a 450-meter-long relay ramp. Sandstone alteration is confirmed to be related to the CO₂ seep by mapping its distribution, which shows a decrease in concentration away from the travertine, and by the unique isotopic signature of calcite cement near the travertine. At distances greater than 25 m from the travertine intense alteration disappears, though scattered fault-subparallel veins and patchy, burial-related calcite cement remain. Intense alteration is limited to major fault overlaps and does not permeate the fault zone along its entire length, nor does it extend outside the zone. This indicates that rising CO₂-laden fluids do not flow uniformly through the entire fault zone, but that vertical flow is channeled at fault intersections. In thin section, porosity near the travertine has been extensively or completely occluded by calcite cement. Permeability in some conduit samples is less than 1 mD, three or four orders of magnitude lower than sandstone away from the travertine. In active CO₂ conduits, such reduction in porosity and permeability would occlude the preferred flow conduit and ultimately restrict upward flow of CO₂-charged water. X-ray diffraction detects small amounts of goethite and hematite and a decrease in chlorite-smectite in altered conduit sandstones. Calcite is abundant, but many authigenic minerals predicted by geochemical models of CO₂ influx into sandstone reservoirs are not observed, including kaolinite, aragonite, dolomite, siderite, ankerite or dawsonite. This difference between observed and predicted mineral occurrence likely results from differences in mineral kinetics between natural and laboratory systems. Prediction of leakage risk based on fault geometry improves the ability to assess the suitability of potential carbon sequestration reservoirs, many of which will be faulted. The point seep nature of leakage through a fault zone limits the amount of CO₂ that can escape over time and also enables targeted surface monitoring for CO₂ escape into the atmosphere--both critical for ensuring the effectiveness of injection projects and earning the trust necessary for carbon sequestration to gain public acceptance. The point seep nature of leakage also accelerates the rate at which conduits may seal through mineralization, since precipitation from a large volume of fluid is focused in a narrow conduit. The presence of multiple fossil and active seep locations along the Little Grand Wash fault, active at different times in the geologic past, indicates that cementation may be effective in sealing single conduits but that fault systems with complex geometry such as Little Grand Wash may continue to leak for a long period of time. / text