Quantification of Flow Architecture and Recharge Processes at the Watershed Scale using Joint Hydrogeophysical Inversion Approaches
We present joint inversion approaches for integrating hydrological, geochemical, and geophysical datasets over scales relevant to watersheds and plumes. Our two primary objectives are to develop inversion approaches that can be used with multi-scale datasets to: (1) quantify subsurface architecture that may influence flow at the regional scale, and (2) to monitor processes associated with recharge in contaminated subsurface environments, such as changes in contaminant concentrations. We focus our studies on datasets collected within a uranium-contaminated aquifer underlying the DOE Integrated Field Research Center located at the Oak Ridge National Laboratory in Tennessee, where large seasonal precipitation may play a significant role in contaminant transport. Quantification of aquifer architecture was met through development of a Bayesian framework that permits simultaneous inversion of surface geophysical and wellbore datasets, thereby jointly honoring all available (multi-scale) datasets and minimizing errors associated with conventional inversion of surface-based geophysical datasets. The developed framework was tested using synthetic surface seismic refraction travel time and wellbore-based lithofacies information as well as with real datasets collected at the Oak Ridge site. The analysis indicated that the incorporation of local-scale depth constraints provided by wellbore datasets in the geophysical inversion procedure significantly reduced the uncertainty architecture obtained using surface seismic refraction datasets. The analysis also delineated the location of a seismic low velocity zone that may serve as a preferential subsurface flowpath. To meet the second objective, we are using coupled numerical models to explore the sensitivity of time-lapse electrical tomographic (ERT) methods for monitoring recharge-related processes, such as changes in moisture due to infiltration or changes in total dissolved solids due to dilution. We have developed a coupled numerical framework that co-simulates hydrological and geophysical phenomena and that permits incorporation of hydrological, geochemical, and geophysical measurements collected at or between wellbores. The modeling is allowing us to distinguish the impact of different recharge-related processes on the geophysical signature and to gain confidence in the use of surface-based ERT measurements for monitoring recharge-related processes across the plume. Ultimately, we will superimpose our seismic-based architecture quantification with the electrical-based recharge monitoring estimates to gain insights about processes associated with natural episodic, seasonal, and annual recharge across the extent of the plume.
SUBGLACIAL RECHARGE INTO THE WESTERN CANADA SEDIMENTARY BASIN - IMPACT OF PLEISTOCENE GLACIATION ON BASIN HYDRODYNAMICS
Brine springs discharging from Devonian carbonates of the Western Canada Sedimentary Basin have distinct water chemistry from brines in laterally equivalent units deeper in the basin. Stable isotope data suggests that the brine springs originated as Pleistocene melt water. These waters are interpreted to originate as an influx of subglacial meltwater related to a reversal of the basin-scale flow-system, caused by the overriding ice sheet. Esker distribution shows a notable relationship between shield and sedimentary rocks. An integrated sedimentary basin/ice sheet model supports the interpretation that high permeability carbonate units acted as preferential subglacial drains that in turn affected esker development. The fluid flow history of the Western Canada Sedimentary Basin is characterized by back-and-forth movement through geological time in response to changing boundary conditions. Modern day flow systems may not then be indicative of historic movement of economic fluids through the basin.
On the Existence of Oscillatory-Convective Thermohaline Flow in Sedimentary Basins
In the Earth's crust, both groundwater temperature and salinity increase with depth. As a consequence, water density is variable, thereby creating density-driven thermohaline groundwater flow. While prior steady-state studies of thermohaline flow in porous media identified conductive, oscillatory and convective thermohaline flow modes, the present study numerically analyzes thermohaline flow using a transient approach. We discovered the existence of an oscillatory-convective flow mode within a specific range of thermal and haline Raleigh numbers. Oscillatory-convective thermohaline flow only exists when water temperature and salinity increase with depth (positive RaT, negative RaS). Candidate sedimentary basins of oscillatory-convective thermohaline flow may be found in Western Canada (Alberta), in the Gulf of Mexico, in Northern Germany, or in Australia.
Near Surface Thermal Disequilibrium, Implications on Detection of Ground Water Flow in Fractured Rock with Temperature Logs
The presence of an anomaly on a temperature log that relates to ground water flow requires both a source or depletion of thermal energy (usually at surface) and a driving mechanism for moving the water that is in thermal disequilibrium to the depth of interest before the energy contrast is dissipated. The utility of temperature logs for the interpretation of groundwater flow in fractured rock under ambient conditions has dramatically increased with improvements in instrumentation and the recent development of techniques for temperature logging in removable (FLUTe) liners. With this improved sensitivity, the maximum depth of applicability has also increased, particularly in areas of downward hydraulic gradient. Whereas historically the near surface heterothermic zone was typically presented as extending to a few tens of metres, there is abundant evidence of thermal variability under ambient flow conditions that extends to over a 100 m at many locations. We observe that the frequency of temperature anomalies indicative of flow through fractures generally decreases with depth. This trend can be inconsistent with other indirect lines of evidence for the potential of flow such as rock core analysis, packer testing and other geophysical data. These observations are supported with numerical modeling results. We conclude that although the usefulness of temperature logging for interpreting ground water flow in fractured rock has increased dramatically, the fundamental need for thermal disequilibrium biases the interpretation of passive temperature data towards the identification of shallow flow zones. Although generally undesirable, in contaminated site investigations this bias can better align the cause and effect conceptualization between surface temperature variations and sources of chemical contamination.
Using Acoustic Televiewer Logs to Spatially Resolve Matrix Porosity and Bulk Density at the Centimeter Scale to Better Interpret Rock Core VOC Profiles
At contaminated sites on fractured sedimentary rock, it is commonly found that most of the contaminant mass resides in the rock matrix rather than in the fractures, because of diffusion driven mass transfer from the fractures over many decades. In the saturated groundwater zone, volatile organic contaminants (VOCs) such as trichloroethene (TCE) occur in the matrix as dissolved and sorbed mass. A method has been developed to determine the contaminant mass distribution in the matrix by conducting total contaminant mass analyses on rock samples collected from continuous core through zones of interest. However, to determine the contaminant distribution in the porewater, a conversion using bulk rock density, porosity, and fraction of organic carbon (foc) is required. This conversion is important to understanding contaminant plume behavior. These rock core VOC analyses are spaced on average along the core at 0.3 m, and for each of these samples physical property values are assigned to convert to porewater concentrations. The cost and nature of sampling makes sample specific measurements for VOC analyses impractical. High resolution porosity and bulk density values were determined from geophysical logs correlated with a representative number of core derived measurements. A total of 246 dolostone core samples taken from seven coreholes were collected to represent the variety of site rock matrix fabrics. Three types of geophysical logs were calibrated to core derived porosity and bulk density measurements: the neutron log, the gamma- gamma log, and the acoustic televiewer amplitude log (ACTV). The ACTV and core sample derived values showed the best correlation compared to the neutron and gamma-gamma logs. This is advantageous because at many sites permits for using neutron and gamma-gamma logs are an obstacle. It is expected that the ACTV will give the best correlation due to its high resolution and shallow depth of investigation but will not correlate well in holes having the borehole wall smeared with sediment, frequent breaking out around fractures, or excessive salinity variations. There are good prospects for this method to be applicable at other sites, however, sufficient laboratory derived data from lithologically representative core samples is necessary.
Quantification of Non-Darcian Flow Encountered During Packer Testing in Fractured Rock
High precision straddle packer tests were conducted using constant rate injection (Q) in fractured dolostone boreholes to identify the conditions of change from Darcian (linear) to non-Darcian (non-linear) flow based on the Q vs dP relationship where dP is the applied pressure above ambient. In each interval (1.5 m packer spacing) as many as 18 separate tests were completed over a large range of injection rates. In the Darcian regime, the linear portion of Q vs dP must pass through the origin (0,0) when the ambient pressure represents static conditions (i.e. Q=0 and dP=0). After the onset of non-Darcian flow, proportionally less Q per unit dP occurs so that the interval transmissivity (T) is underestimated by by up to an order of magnitude using Darcy's Law based models. Onset of non-linearity depends on interval length and permeability, but for a 1.5 m interval it typically occurs at critical Reynolds number (Rec) of less than 1 to 5 corresponding to injection rates of 0.01-1 L/min. This Rec range for the onset of nonlinearity is consistent with literature values from laboratory experiments involving flow in single rough fractures. No literature Rec values are found for packer tests, but assessment of conventional packer testing indicates that the tests are most commonly done in the non-linear regime, causing underestimates of T and by extension, hydraulic aperture derived using the Cubic Law. Where contaminant transport analysis is the main purpose of the hydraulic tests, these errors can lead to substantial underestimates in the velocity in the fracture network and contaminant arrival times. The test method developed in this study is effective for use in investigations where flow regime identification is desired.
Types of Permeability Development in Carbonate Aquifers: Examples From Ontario
Paleozoic carbonates are common in Ontario and form highly-productive aquifers that are widely utilized for water supplies. However, carbonates may have rapid groundwater velocities with little attenuation of bacterial contamination. One such example is at Walkerton, where more than 2000 people became ill and seven died after the municipal wells became contaminated with pathogenic bacteria. Subsequent studies showed that almost all the flow into the wells was from a few solutionally-enlarged fractures (or channels) at depths up to 70 m below the water table. Measured velocities from tracer tests to one of the municipal wells showed that groundwater velocities over distances up to 350 m were 200 - 500 m/day. This implies that there is an interconnected network of channels with apertures of at least 3 mm. An important question is whether such channel networks are common in carbonate aquifers. Advances over the last forty years have resulted in a clear understanding of how dissolution processes enhance aquifer permeability. First, laboratory experiments established that there is a precipitous drop in dissolution rates as chemical equilibrium is approached. Then numerical models showed how channel pathways become enlarged and integrated over time, forming self-organized networks that typically have apertures in the millimeter to centimeter range. The models have shown that two end-member channel network types may be distinguished, one with many channels of similar size (microkarstic aquifers) and one where a small number of large channels conduct most of the flow (macrokarstic aquifers). Macrokarstic development often occurs in Ontario where overburden is minimal and is manifested by sinking streams, springs, and occasionally caves; some dozens of examples have been documented. Microkarstic development is much more common; tracer tests suggest that natural-gradient groundwater velocities often exceed 30 m/day. Wide-ranging studies of bacterial contamination of wells have shown that carbonate wells are amongst the most frequently contaminated, and the measured high groundwater velocities provide a satisfactory explanation for this vulnerability.