Document Actions

ijege-12_01-feinstein.pdf

background image
59
Italian Journal of Engineering Geology and Environment, 1 (2012)
© Casa Editrice Università La Sapienza
www.ijege.uniroma1.it
DOI: 10.4408/IJEGE.2012-01.O-04
D
aniel
FEINSTEIN
(*)
(*)
University of Wiscounsin , U.S Geological Survey - Milaukee, USA
MODELLO DI FLUSSO REGIONALE DELLE ACQUE SOTTERRANEE
DEL BACINO DEL LAGO MICHIGAN: PRIMI ELEMENTI
A REGIONAL GROUNDWATER-FLOW MODEL OF THE
LAKE MICHIGAN BASIN: PRELIMINARY ELEMENTS
A regional groundwater-flow model of the Lake Michigan Ba-
sin and surrounding areas in the northern midwest region of the
United States has been developed in support of the U. S. Geo-
logical Survey National Assessment of Water Availability and Use
Program-Great Lakes Basin Pilot (F
einstein
et alii, 2010). The
transient 2-million-cell model incorporates multiple aquifers and
pumping centers with drawdown that extend into deep saline wa-
ters. The 20-layer model simulates the exchange between a dense
surface-water network and heterogeneous glacial deposits overly-
ing stratified bedrock of the Wisconsin and Kankakee Arches and
the Michigan Basin in the Lower and Upper Peninsulas of Michi-
gan; eastern Wisconsin; northern Indiana; and northeastern Illinois.
It provides a platform for quantifying the regional sources and sinks
of groundwater (including recharge, pumping, and groundwater
flow to inland surface water and to Lake Michigan-all elements of
the groundwater budget that change with time) and for mapping the
direction and magnitude of flows in a series of aquifers (including
the source areas for wells and the locations of major groundwater
divides at various depths on both sides of Lake Michigan and the
migration of the divides in response to pumping).
Five datasets, which were prepared as part of the Great Lakes
Basin Pilot to serve as the foundation for model development, are
described in separate reports:
• a three-dimensional hydrogeologic representation of aquifers and
confining units above the Precambrian basement, with a maxi-
mum thickness of 15,000 ft in the middle of the Michigan Basin
(l
ampe
, 2009);
• maps of the coarse fraction of unconsolidated material at depth
intervals of 0-100 ft, 100-300 ft, and greater than 300 ft, overlaid
on existing interpretations of glacial categories (a
rihooD
, 2009);
• location, depth, and pumping rates of high-capacity public-sup-
ply, industrial, and irrigation wells from the early 20th century
through 2005 (B
uchwalD
et alii, 2010), a compilation that docu-
ments generally upward trends in withdrawals and some shifts
between deep and shallow pumping;
• maps of recharge derived from a soil-water-balance model that
reveals trends in the spatial and temporal distribution of inflow to
the water table (w
estenBroek
et alii, 2009);
• maps of salinity in hydrogeologic units that show the three-di-
mensional boundary between fresh and saline water, as well as
the distribution of high concentrations of dissolved solids in the
Michigan Basin (l
ampe
, 2009).
These datasets, along with boundary conditions linked to out-
lying Great Lakes, hydrologic coverages delineating the surface-
water network, and hydrogeologic information relating primarily
to hydraulic conductivity provided the input required by the SEA-
WAT-2000 model to simulate groundwater flow before pumping
(steady-state simulation) and after development (transient simu-
lation with 12 stress periods extending from 1864 to 2005). The
simulation uses a form of the groundwater-flow equation that takes
account of variable density (l
angevin
et alii, 2003). Two versions
of the model were calibrated: one for confined conditions and one
for unconfined conditions. Multiple target sets developed from ob-
servations of head and base flow and inversion methods using the
suite of PEST computer programs (D
oherty
, 2008a,b; D
oherty
et
alii, 2010) guided the adjustment of initial inputs. Comparison of
background image
MODELLO DI FLUSSO REGIONALE DELLE ACQUE SOTTERRANEE DEL BACINO DEL LAGO MICHIGAN: PRIMI ELEMENTI
60
D. F
einstein
some areas by metropolitan water-supply projects that substituted
Lake Michigan water for groundwater supplies). On the west side of
Lake Michigan, well withdrawals have caused a complete reconfigu-
ration of the deep divides. Before the advent of pumping, the deep
Lake Michigan groundwater basin boundaries extended to the west
of the Lake Michigan surface-water basin boundary, in some places
by tens of miles. Over time, the pumping centers have replaced Lake
Michigan as the regional sink for the deep part of the flow system.
The regional model results provide a broad picture of the status
of the groundwater resource and how it has responded to pumping.
However, there are limitations imposed by the relatively coarse grid
spacing. Laterally, the finite-difference cells are 5,000 ft on a side in
the Lake Michigan Basin and in adjoining areas. At this resolution,
the simulation of the water-table response to pumping is severely
constrained by the necessity of including enough of the surface-water
system in model cells to provide outlets for recharge and, thereby, to
avoid spurious simulated water-level mounding. The mounding that
occurs when discharge points are neglected can be offset by increasing
hydraulic conductivity; but this fix distorts the K
h
and K
v
fields relative
to field conditions. In order to avoid distorting the hydraulic conduc-
tivity input, more than half the water-table cells in the Lake Michigan
Basin model contain surface-water features, each of which is repre-
sented by a boundary condition with a fixed stage. The stage tends
to “staple” the water-table solution because there is generally a small
gradient between the average groundwater head solved for the cell and
the surface-water level assigned to the cell. The regional model by
itself cannot overcome this limitation; however, in conjunction with
techniques for inset models, it can lay the foundation for any number
of applications designed to address local management problems re-
lated to optimizing water supply and maintaining ecologic flows.
Two promising new techniques-Local Grid Refinement in MOD-
FLOW-2005 and Hybrid Analytic Element/Finite Difference Mode-
ling-could allow enhanced versions of the regional model to simulate
groundwater/surface-water interactions in the presence of pumping at
the necessary level of refinement while still honoring the regional pat-
tern of flow needed to properly simulate water availability. Research
aimed at demonstrating these two methods in a setting characterized
by pumping near headwater streams is part of the Great Lakes Basin
Pilot project (see h
aitjema
et alii, 2010; and h
oarD
, 2010).
The construction of alternative versions of the regional model re-
veals important insensitivities with respect to model design. One is
related to variable density. Whereas the specification of salinity dra-
matically affects groundwater conditions in the deep Michigan Basin
and even though the simulated drawdown around pumping centers
extends into the highly saline waters, model results indicate that vari-
able density can be neglected with negligible effect on the range of
simulated results in freshwater areas under either predevelopment or
stressed postdevelopment conditions. Relaxing the assumption that
saline concentrations are fixed through time also has little effect on
model output. A second finding of insensitivity is related to the level
updated parameter values, calibration statistics, and parameter sen-
sitivities demonstrated that the confined and unconfined versions of
the model produced solutions similar in most respects.
The output of the calibrated confined model was selected for
detailed presentation largely for reasons of numerical stability dur-
ing inversion and no loss of pumping to dry cells as a result of
drawdown. The simulated results, organized laterally into seven
subregions and vertically into five aquifer systems, included maps,
cross sections and tables of:
• regional predevelopment water table and head conditions at depth
in bedrock units;
• changes in water levels (drawdown and recovery) over time, by
aquifer system;
• changes in the magnitude and direction of shallow and deep flow;
• water budgets that quantify regional sources (such as recharge
and storage release) and sinks (such as base flow to streams and
discharge to wells) through time.
Analysis of the results by means of particle tracking, revealed:
• sources of water to shallow and deep wells by subregion;
• the changing configuration of the divides that delineate the Lake
Michigan groundwater basin and the postdevelopment groundwa-
ter basins around pumping centers; and
• the distribution of direct and indirect discharge of groundwater
to Lake Michigan and the modifying effects of pumping on the
distribution.
The multiple perspectives provided by the simulated model
output portray a regional groundwater flow system that, over time,
has largely maintained its natural predevelopment configuration but
locally has been strongly affected by well withdrawals. The quan-
tity of rainfall in the Lake Michigan Basin and adjacent areas sup-
ports a dense surface-water network and recharge rates consistent
with generally shallow water tables and a flow system generally
dominated by shallow circulation. At the regional scale, pumping
has not caused appreciable disruption of the shallow flow system;
but pumping has resulted in decreases in base flow to streams and
in direct discharge to Lake Michigan. Comparison of inset models
constructed along the Lake Michigan coastline suggests that the re-
gional model because of its coarse grid spacing and coarse repre-
sentation of surface water underestimates the direct discharge by
about 48 percent. When the bias is corrected, the results indicate
that about 2 percent of total groundwater flow is directly discharge
to the lake at a rate of about 0.5 ft
3
/s per mile of shoreline.
Well withdrawals have caused reversals in regional flow patterns
around pumping centers in deep, confined aquifers (most noticeably
in the Cambrian-Ordovician aquifer system on the west side of Lake
Michigan near the cities of Green Bay and Milwaukee in eastern Wis-
consin, and around Chicago in northeastern Illinois, as well as in some
shallow bedrock aquifers (for example, in the Marshall aquifer near
Lansing, Mich.). The shifts in flow have been accompanied by large
drawdowns with consequent local decrease in storage (moderated in
background image
A REGIONAL GROUNDWATER-FLOW MODEL OF THE LAKE MICHIGAN BASIN: PRELIMINARY ELEMENTS
61
Italian Journal of Engineering Geology and Environment, 1 (2012)
© Casa Editrice Università La Sapienza
www.ijege.uniroma1.it
of detail appropriate for the values assigned to the hydraulic conduc-
tivity of unconsolidated sediments. The availability of geologic de-
scriptions from hundreds of thousands of driller logs for household
wells permitted cell-by-cell mapping of hydraulic conductivity in the
top three model layers. When this distribution is zoned more broadly
on the basis of glacial categories alone (that is, on the basis of mate-
rial types such as clayey till and coarse outwash), the model results
give rise to a somewhat modified water-table solution, but the agree-
ment to calibration targets is only weakly compromised, and the find-
ings are very similar to the more detailed model with respect to the
regional drawdown response and the regional water budget.
In summary, the results of this modeling effort have yielded
• improved estimates of the various components of the water bud-
get for the region,
• improved estimates of the various hydraulic properties of the ge-
ologic units in the region, and
• a better understanding of the groundwater flow throughout the
region.
The regional model is also intended to support the framework pilot
study of water-availability and use at the scale of the entire Great Lakes
Basin. To that end, an ongoing effort has been undertaken to distill the
model’s findings using a series of sustainability indicators intended to
reveal overall patterns in the status of the water resource in terms of
the human effect on natural groundwater flows and on groundwater/
surface-water interactions. Moreover, the regional model is being ap-
plied in forecasting mode to shed light on the effects of possible future
levels of pumping on the groundwater system and is being used to test
hypotheses regarding the effect of climate variability and change on
water availability. All these aspects, along with a demonstration of the
procedures for embedding models and sample results related to eco-
logic flows, are discussed in detail in a USGS Professional Paper on the
comprehensive findings of the Great Lakes Basin Pilot (r
eeves
, 2010).
OPERE CITATE / REFERENCES
a
rihooD
l.D. (2009) - Processing, analysis, and general evaluation of well-driller logs for estimating hydrogeologic parameters of the glacial sediments in a ground-
water flow model of the Lake Michigan Basin: U.S. Geological Survey Scientific Investigations Report 2008-5184, 26 pp.
B
uchwalD
c.a., l
uukkonen
c.l. & r
achol
c.m. (2010) - Estimation of groundwater use for a groundwater-flow model of the Lake Michigan Basin and adjacent
areas, 1864-2005. U.S. Geological Survey Scientific Investigations Report [in press].
D
oherty
j. (2008a) - PEST, Model Independent Parameter Estimation-User manual (5
th
ed.): Brisbane, Australia, Watermark Numerical Computing, accessed
October 1, 2009, at http://www.pesthomepage.org/Downloads.php.
D
oherty
j. (2008b) - PEST, Model Independent Parameter Estimation-Addendum to user manual (5
th
ed.): Brisbane, Australia, Watermark Numerical Computing,
accessed October 1, 2009, at http://www.pesthomepage.org/Downloads.php.
D
oherty
j. & h
unt
R.J. (2010) - Approaches to highly parameterized inversion - A guide to groundwater model calibration using PEST: U.S. Geological Survey
Scientific Investigations Report [in press].
F
einstein
D.t., h
unt
r.j. & r
eeves
h.w. (2010) - Regional groundwater-flow model of the Lake Michigan Basin in support of Great Lakes Basin water availability
and use studies: U.S. Geological Survey Scientific Investigations Report [in press].
h
aitjema
h., F
einstein
D., h
unt
r. & g
usyev
m. (2010) - A hybrid finite difference and analytic element model for detailed surface-ground water modeling on a
regional scale: Ground Water [in press].
h
oarD
c.J. (2010) - Implementation of local grid refinement in MODFLOW for the Lake Michigan Basin regional groundwater flow model: U.S. Geological Survey
Scientific Investigations Report [in press].
l
ampe
D.c. (2009) - Hydrogeologic framework of bedrock units and initial salinity distribution for a simulation of ground-water-flow for the Lake Michigan Basin:
U.S. Geological Survey Scientific Investigations Report 2009-5060, 49 pp.
l
angevin
c.D., s
hoemaker
w.B. & g
uo
w
eixing
(2003) - MODFLOW-2000, the U.S. Geological Survey modular ground-water model-Documentation of the
SEAWAT-2000 version with the Variable-Density Flow Process (VDF) and the Integrated MT3DMS Transport Process (IMT): U.S. Geological Survey Open-File
Report 03-426, 43 pp.
r
eeves
h.w. (2010) - Water availability and use pilot-Multiscale assessment in the U.S. Great Lakes Basin: U.S. Geological Survey Reeves Professional Paper [in press].
w
estenBroek
s.m., k
elson
v.a., D
ripps
w.r., h
unt
r.j. & B
raDBury
k.r. (2009) - SWB-A modified Thornthwaite-Mather soil-water balance code for estimating
ground-water recharge: U.S. Geological Survey Techniques and Methods 6-A31: 65 pp.
Received April 2010 - Accepted November 2011
Statistics