In borehole dilution experiments the change in tracer concentration within a well is used to infer the velocity, , of groundwater flowing through the well screen. The interpretation of such experiments typically uses a linear mixing model that requires that the tracer be prevented from diffusing vertically in the wellbore. We conducted a borehole dilution experiment in which a sodium-chloride tracer was introduced to the screened part of a well, but no packer was installed at the top of the screen. To interpret these data, we have derived an equation that predicts tracer concentration in a well when the tracer diffuses vertically while it is diluted by water flowing through the screen. We fit the field data to our theoretical results by constructing type curves from which match points can be used to infer .
Home | Back | Down
About half of North Carolina¹s population utilizes groundwater resources. On the Coastal Plain most municipalities and most rural homes rely exclusively on groundwater for water supply. This is particularly true on the Outer Banks where surface water resources are severely limited. In contrast, most municipalities in the Piedmont and mountain regions exploit surface water resources, although most rural homes have private water wells. There are essentially two reasons for this contrast: (1) groundwater on the Coastal Plain is abundant, of high quality and relatively easy to extract; and (2) because of the topographic relief in the Piedmont and mountains it has been relatively easy to develop surface water reservoirs. Moreover, groundwater in the Piedmont and mountain regions typically is difficult to extract in volumes large enough to meet municipal demand. The Coastal Plain aquifer system consists of layered aquifers that are of high quality and productivity. Groundwater on the Outer Banks occurs in surficial aquifers, for which the water table is typically a few feet below land surface (as well as in deeper confined aquifers, some of which are continuous with the Coastal Plain aquifer system). The shallow water table makes groundwater in coastal areas relatively easy to exploit, but also make it vulnerable to contamination.
In contrast to coastal settings, groundwater in the Piedmont and the mountains occurs in widely spaced (i.e., tens of feet) discrete fractures within bedrock, as well as in thick soils and saprolite with abundant clay minerals that restrict the rate of groundwater flow. When bedrock aquifers become contaminated it is difficult to monitor and predict contaminant transport because flow can be rapid through some fractures while nearby fractures may be unconnected hydraulically, therefore unaffected by contamination.
In this paper I present two approaches to monitoring groundwater quality and testing aquifer properties in North Carolina. These two approaches correspond to two contrasting hydrogeologic settings. In the first section I present the basis of a water quality study on Bodie Island, North Carolina. On Bodie Island the water table is very shallow, leading to an extensive wetland system on the west side of the island. On the east side of the island groundwater discharges to the ocean. The National Park Service is concerned that extensive development in South Nags Head is threatening water quality in the Park Service wetlands and on the beach-side of the island that is used extensively for recreation. At this site we are beginning a water quality study to identify the major pathways of septic effluent from the developed part of the island.
In the second part of this paper I present the results of research aimed at addressing the fundamental question of how we can best measure hydraulic properties of, and monitor water quality in, fractured bedrock aquifers of the Piedmont and mountain regions.
Abstract
Understanding the distribution of water-bearing fractures in crystalline rock is an important component in evaluating the availability and vulnerability of water resources throughout the Northeastern U.S. The State of Maine requests well drillers to report estimates of fracture depths and fracture yields for all bedrock wells drilled in the state. Using these data we analyze fracture-depth and fracture-yield data from 227 bedrock wells in Coastal Maine in order to understand how fracture locations and yields are distributed with depth. Numerical simulations and statistical tests show that it is not possible to infer how fractures are distributed with depth: fracture depths are consistent with several distributions, including uniform fracture density with depth. In order to understand how fracture-yield varies with depth, we group yield data into 50 foot depth intervals and compare distributions in each interval using non-parametric statistical tests. These tests show that the distribution of fracture yield in different depth intervals are statistically equivalent. These results imply that there is no empirical justification for limiting well depth when drilling for water resources in fractured bedrock in Coastal Maine.
Fluid electrical conductivity logging experiments introduced by Tsang et al. [1990] represent a new technology for studying the hydraulic properties of discrete fractures in open boreholes. These experiments consist of replacing water in the wellbore with de-ionized water and then pumping the well to induce formation water to return to the wellbore. During and after the fluid exchange, the entire well is logged to measure fluid electrical conductivity (FEC). At locations where formation water enters the well there are abrupt increases in the borehole FEC, thereby revealing the location of hydraulically conductive fractures. As pumping continues, high FEC water moves up the wellbore at a rate proportional to the total inflow rate below the observation point. Inflow rates and FECs of formation water associated with each fracture can be inferred by modeling the FEC profiles with the one-dimensional advective dispersion equation. In this paper I present an inverse model which estimates the inflow rates and formation FEC values that optimally fit observed FEC logs acquired during fluid exchange experiments described by Tsang et al. [1990]. With this inverse model inflow rates are constrained by the rate at which the well is pumped during the experiment. The forward model is solved numerically using a control-volume finite difference scheme with power-law upstream weighting and source-term linearization. The inverse problem is solved using the Gauss-Newton iterative method. The rows of the Jacobian matrix, or the sensitivity coefficients, are calculated numerically with the same algorithm that solves the forward problem. Both constrained and unconstrained inverse models are used to interpret fluid logging experiments performed in research wells penetrating Piedmont rocks of North Carolina.
The hydraulic properties of the surficial aquifer on Hatteras Island reflect the
coarsening-upward stratigraphy typical of barrier island settings. Modeling the aquifer
using bulk hydraulic properties can lead to specious results because the water table
response to recharge events is most strongly affected by the shallow aquifer properties,
while the annual water table variation is more influenced by aquifer properties at depth.
In order to understand and model the aquifer, therefore, it is necessary to quantify the
vertical variation in aquifer properties.
The surficial aquifer on Hatteras Island is bounded below by a leaky confining unit about
24m deep. This unit extends to the mainland and is part of the coastal plain aquifer
system. It is overlain by at two coarsening-upward sequences that are apparent on nearly
all well logs that have been acquired on the island.
In order to characterize the vertical heterogeniety in the surficial aquifer, we have
conducted a number of aquifer tests, including single-well and multiple-well pump tests.
Single-well pump tests were conducted in wells screened at different depths, and indicate
that hydraulic conductivity increases several-fold over a coarsening-upward cycle.
Vertical variation in storativity is apparent in multiple-well tests using wells that are
screened from the water table to the confining unit at 24m. Data from these tests can be
explained using a model that combines the response of a two-layer semiconfined-unconfined
aquifer system. The layers inferred from the multiple well test correspond to the distinct
coarsening-upward cycles inferred from well logs.
Dept. of Marine, Earth, & Atmospheric Sciences, North Carolina State University, Raleigh, NC, 27695, dave_evans@ncsu.edu;
US Geological Survey, MS 403 Box 25046, Federal Center, Denver, CO;
COLOG, 17301 W Colfax Ave. Suite 265, Golden, CO, 80401
Throughout the Piedmont region of the southeastern United States hydrogeologists must
often characterize the bedrock hydrology at scales smaller than the representative
elemental volume. This entails identifying hydraulically conductive fractures, measuring
their hydraulic properties, and inferring their interconnections. With these objectives in
mind we have conducted fluid electrical conductivity (FEC) logging experiments and
transient flow-meter aquifer tests in a research well field in Raleigh, North Carolina.
The well field consists of three wells cased through 30m of soil and saprolite, and open
below the casing to a depth of 122m in the Raleigh gneiss.
We conducted transient flow meters tests by pumping one of the test wells while logging
vertical flow rates above fractures intersecting the other wells. The temporal variation
in flow rates above each fracture are characteristic of the type of hydraulic connection
between that fracture and fractures intersecting the pumping well. In our test wells there
is one horizontal fracture at about 94 m that is hydraulically connected to all three test
wells. This fracture is also hydraulically connected to shallower fractures intersecting
two of the wells. In the same wells we also conducted FEC logging experiments in which
borehole water was replace with de-ionized water and the wells were pumped to induce the
flow of formation water back into the wells. The test well was logged vertically for FEC.
The presence of hydraulically conductive fractures was apparent from abrupt increases in
FEC at the fracture locations. Inflow rates from each fracture were inferred from the rate
that high-FEC water moved up the wellbore. Both the flow meter and FEC logging experiments
revealed fractures at the same locations and, when combined with drawdown data, indicated
comparable values for fracture transmissivity.
On August 31, 1993 Hurricane Emily, a category three storm, grazed the North Carolina barrier islands. An ongoing aquifer monitoring program in the Buxton Woods maritime forest recorded the water table response to the storm. The monitoring network consists of three transects extending from dune swale sedge wetlands to beyond the central dune ridge on the island. Rainfall was recorded at each transect. The local recharge varied with ecotype, depending primarily on depth of the water table. Ground water rise was most rapid in sedge wetlands. Ground water levels in the sedge rose an average of about 0.9m over a twohour period. In tree-shrub wetlands water levels rose rapidly during the storm and then quickly declined, whereas other topographic types showed a steady water level rise after the initial inundation. During recharge, the water table was extremely dynamic with hydraulic gradients temporarily reversing directions as a result of local variations in recharge rate.
Hydrogeologists have recognized for years the value of tracer tests to determine flow
velocities and transport parameters in groundwater systems. Quite often, however,
logistical or regulatory obstacles discourage the use of tracer tests, even when such
tests are the most appropriate means of acquiring needed data. Single-well tracer tests
can often overcome these obstacles, particularly when a benign tracer, such as de-ionized
water or a dilute sodium chloride solution, is used. Moreover, single-well tracer tests in
complex aquifers allow hydrogeologists to discern variations in flow rates within an
aquifer. In this paper I discuss a variety of single-well tracers tests and how they can
be used to infer groundwater flow rates into wells. Under pumping conditions these rates
can be combined with drawdown data to estimate transmissivity of discrete inflow zones.
Under non-pumping conditions, such tests can be used to estimate the horizontal component
of flow from field experiments. They tend to be less expensive and provide better
resolution of inflow zones than do many other technologies.
In all of the field tests presented here, the tracer is either de-ionized water or a
dilute sodium chloride solution. The concentration of these tracers can be measured
inexpensively using fluid electrical conductivity (FEC) probes. The FEC of most dilute
solutions varies linearly with concentration, so that FEC can be used directly as a proxy
for concentration.
In Revision for to Geophysical Research Letters
In some hydrogeologic environments, such as near salt domes, groundwater is subject to
large lateral density gradients arising from variations in salinity and temperature.
Accurate models of groundwater flow in such environments require that flow be explicitly
coupled to heat and mass transport. Non-linearities in the coupled heat and mass transport
equations can cause the solution to these equations to bifurcate such that it oscillates
between distinct convergence points. We have investigated this phenomena through a series
of flow and transport simulations in which we systematically change one flow parameter
(the Rayleigh number, the Lewis number, or the buoyancy ratio) while holding other flow
parameters constant. Our numerical experiments indicate that bifur cations can arise
as artifacts of (1) poor spatial discretization, and (2) the numerical scheme used to
resolve the non-linearities. The former problem is related to inadequately resolving high
spatial frequencies, which has been found to introduce spurious bifurcations in other
solution techniques, and can be corrected by guarding against false diffusion.
Bifurcations that arise from the scheme used to resolve the non-linearities can be
eliminated by iteratively solving the flow and transport equations using under-relaxation.
Iterative solutions using under-relaxation also eliminate some problems with
strongly-coupled transient simulations. Numerical simulations that produce chaotic flow
regimes when Newton-Raphson iterations are used produce non-chaotic flow patterns when
under-relaxation is used.