WISCONSIN GEOLOGICAL AND NATURAL HISTORY SURVEY

The 2016 Groundwater Flow Model for Dane County, Wisconsin

Bulletin 110 • 2016 Michael J. Parsen Kenneth R. Bradbury Randall J. Hunt Daniel T. Feinstein Wisconsin Geological and Natural History Survey Kenneth R. Bradbury, Director and State Geologist

WGNHS staff Research associates William G. Batten, geologist Jean M. Bahr, University of Evan R. Larson, University of Eric C. Carson, geologist Wisconsin–Madison Wisconsin–Platteville Peter M. Chase, geotechnician Mark A. Borchardt, USDA– John A. Luczaj, University of Agricultural Research Station Wisconsin–Green Bay Linda G. Deith, editor Philip E. Brown, University of J. Brian Mahoney, University of Anna C. Fehling, hydrogeologist Wisconsin–Madison Wisconsin–Eau Claire Madeline B. Gotkowitz, hydrogeologist Charles W. Byers, University of Shaun Marcott, University of Brad T. Gottschalk, archivist Wisconsin–Madison (emeritus) Wisconsin–Madison Grace E. Graham, hydrogeologist William F. Cannon, U.S. Geological Survey Joseph A. Mason, University David J. Hart, hydrogeologist Michael Cardiff, University of of Wisconsin–Madison Irene D. Lippelt, Wisconsin–Madison Daniel J. Masterpole, Chippewa water resources specialist William S. Cordua, University Co. Land Conservation Dept. Sushmita S. Lotlikar, of Wisconsin–River Falls David M. Mickelson, University of administrative manager Robert H. Dott, Jr., University of Wisconsin–Madison (emeritus) Stephen M. Mauel, GIS specialist Wisconsin–Madison (emeritus) Donald G. Mikulic, M. Carol McCartney, outreach manager Charles P. Dunning, Illinois State Geological Survey Michael J. Parsen, hydrogeologist U.S. Geological Survey William N. Mode, University of Wisconsin–Oshkosh Jill E. Pongetti, office manager Daniel T. Feinstein, U.S. Geological Survey Maureen A. Muldoon, University J. Elmo Rawling III, geologist Michael N. Fienen, of Wisconsin–Oshkosh Caroline M.R. Rose, GIS specialist U.S. Geological Survey Beth L. Parker, University of Guelph Kathy Campbell Roushar, GIS specialist Timothy J. Grundl, University of Robert E. Pearson, Peter R. Schoephoester, GIS specialist Wisconsin–Milwaukee Wisconsin Dept. of Transportation Aaron N. Smetana, system administrator Nelson R. Ham, St. Norbert College Kenneth W. Potter, University Val L. Stanley, geologist Paul R. Hanson, of Wisconsin–Madison Esther K. Stewart, geologist University of Nebraska–Lincoln Todd W. Rayne, Hamilton College Carolyn M. Streiff, geophysicist Karen G. Havholm, University Daniel D. Reid, of Wisconsin–Eau Claire Wisconsin Dept. of Transportation Jay Zambito, geologist Randy J. Hunt, U.S. Geological Survey Randall J. Schaetzl, and approximately 15 graduate and Michigan State University undergraduate student workers John L. Isbell, University of Wisconsin–Milwaukee Allan F. Schneider, University of Emeritus staff Mark D. Johnson, Wisconsin–Parkside (emeritus) University of Gothenburg Madeline E. Schreiber, Virginia Tech John W. Attig, geologist Joanne L. Kluessendorf, Susan K. Swanson, Beloit College Bruce A. Brown, geologist Weis Earth Science Museum Kent M. Syverson, University Thomas J. Evans, geologist George J. Kraft, Central Wisconsin of Wisconsin–Eau Claire Ronald G. Hennings, hydrogeologist Groundwater Center Lucas Zoet, University of Frederick W. Madison, soil scientist Wisconsin–Madison Stanley A. Nichols, biologist Deborah L. Patterson, GIS specialist Roger M. Peters, subsurface geologist James M. Robertson, geologist

The Wisconsin Geological and Natural History Survey also maintains collaborative relationships with a number of local, state, regional, and federal agencies and organizations regarding educational outreach and a broad range of natural resource issues. Bulletin 110 n 2016

The 2016 Groundwater Flow Model for Dane County, Wisconsin

Michael J. Parsen Kenneth R. Bradbury Randall J. Hunt Daniel T. Feinstein Suggested citation: Parsen, M.J., Bradbury, K.R., Hunt, R.J., and Feinstein, D.T., 2016, The 2016 groundwater flow model for Dane County, Wisconsin: Wisconsin Geological and Natural History Survey Bulletin 110, 56 p.

Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Published by and available from: Wisconsin Geological and Natural History Survey 3817 Mineral Point Road n Madison, Wisconsin 53705-5100 608.263.7389 n www.WisconsinGeologicalSurvey.org Kenneth R. Bradbury, Director and State Geologist

ISSN: 0375-8265 ISBN: 978-0-88169-992-0

Cover photos Front, title page: Aerial photo of Madison, © Jeff Miller / UW–Madison Back: Pheasant Branch Conservancy, © Dick Lathrop Inside back: Lake Wingra, © Linda Deith Contents

Abstract ...... 1 Conceptualization of the Steady-state model calibration 30 groundwater system . . . . . 16 Steady-state calibration Introduction ...... 2 targets ...... 30 Scope ...... 2 Three-dimensional model Steady-state calibration construction and simulation parameters ...... 32 Setting ...... 4 of the groundwater flow system 18 Steady-state calibration Objectives ...... 4 results ...... 34 Model grid ...... 18 Important features Transient model calibration . . 39 Model construction . . . . . 18 of the new model ...... 4 Transient calibration targets 39 Hydrostratigraphic units . . . 18 Model distribution and use 5 Transient stress periods . . 40 Layering approach and Acknowledgments . . . . . 5 treatment of pinchouts 18 Transient calibration parameters ...... 41 Unlithified materials . . . . 22 Study methods ...... 6 Transient calibration results . 41 Bedrock units ...... 22 Review of previous studies . . . 6 Bedding-plane fractures . . 23 Model results 44 Location and identification High-conductivity zone Model solution and numerical of public and private wells . . . 6 near Fish and Crystal Lakes . 23 mass balance 44 Geophysical logging . . . . . 6 Boundary conditions . . . . 23 Comparison of 2010 and Water-use survey and Streams ...... 23 predevelopment simulations 44 data compilation ...... 6 Lakes ...... 24 Water levels and drawdown 44 Stream and spring flow Springs 24 Mass balance and flow to measurements 7 Specified-head boundaries . 24 surface water ...... 48 Recharge estimation . . . . . 7 No-flow boundaries . . . . 24 Demonstration of transient Hydrostratigraphy 8 model response ...... 50 Sources and sinks of water . . 24 Numerical simulation methods . 8 Recharge 24 Model limitations 52 Wastewater treatment Overview ...... 52 Geology, hydrostratigraphy, discharges 26 and hydrology of Dane County . . 9 Groundwater withdrawals . . 26 Limitations related to Quaternary geology and discretization ...... 52 hydrostratigraphy ...... 9 Hydraulic properties of the Limitations related to Bedrock geology and groundwater flow system . . . 27 model stability 53 hydrostratigraphy ...... 9 Hydraulic conductivity and Limitations related to a lack of Bedding-plane fractures . . 12 storage ...... 27 hydrogeologic knowledge . . 53 Fish and Crystal Lakes . . . 53 Surface-water features . . . . 12 Parameterization of unlithified Lakes and wetlands 13 material properties . . . . . 27 Parameter uncertainty . . . 53 Rivers and streams . . . . 14 Parameterization of bedrock Streamflow and lake level along the Yahara River . . . 53 Springs 14 properties ...... 29 Wetland drainage 54 Water use and wastewater Parameterization of streambed discharge 15 and lakebed properties 29 Summary 54 Model calibration 30 Literature cited 55 Calibration strategy . . . . . 30 The 2016 Groundwater Flow Model for Dane County, Wisconsin

25. Comparison of changes in streamflow between predevelopment and 2010 conditions 51 26. Transient drawdown simulation for a hypothetical well near Story Creek in southern Dane County 51 27. Detailed results for the hypothetical well showing drawdown for the creek, spring, and pumped well ...... 52

Spring Harbor Spring Tables 1. Hydraulic properties of hydro­ Sue Swanson stratigraphic units used in the Figures model 28 2. Conversion between map units 1. Location and extent of the 14. Histograms of hydraulic defined by Clayton and Attig Dane County model ...... 3 conductivity for the Wonewoc (1997) and hydrogeologic (layer 9) and Mount Simon 2. Bedrock stratigraphy and materials in model layers 1 and 2 29 corresponding layers in the 1996 (layer 12) aquifers ...... 36 3. Steady-state calibration and 2016 groundwater models. 10 15. Steady-state flow calibration target list ...... 30 3. Thickness and extent of the results 37 4. Simulated lakes and calibrated Eau Claire aquitard 11 16. Calibrated transmissivity leakance values ...... 38 4. Major hydrologic features distribution for layer 9 5. Steady-state mass balance in Dane County ...... 13 (Wonewoc aquifer) 38 summary ...... 44 5. Conceptual model of the 17. Precipitation, water levels, 6. Simulated predevelopment and groundwater system 17 and pumping during the 2012 calibration period 39 2010 streamflows at selected 6. Extents and elevations for sites in Dane County 49 key model layer surfaces 19 18. Locations of all transient calibration targets ...... 40 7. Simulated predevelopment and 7. Model layer thicknesses 2010 spring flows at principal 19. Transient calibration results for key model layers ...... 20 springs in Dane County 50 for selected well targets . . . . 42 8. Hydrostratigraphic cross sections across Dane County 21 20. Transient calibration results Appendices for selected flow and lake-level 9. Distribution of unlithified targets 43 Appendices are provided in an Excel file. materials and near-surface rock in model layers 1 and 2 22 21. Simulated steady-state results 1. Example geophysical log under predevelopment 2. Wastewater treatment plant 10. Location of all steady-state conditions (no pumping) 45 calibration targets ...... 25 discharges included in the model 22. Simulated steady-state results 11. Location of layer 12 (Mount 3. Pumping wells included in the under 2010 conditions model (steady-state and transient) Simon aquifer) Kx pilot points (pumping) 46 used during steady-state model 4. Head calibration targets 23. Simulated steady-state calibration 33 5. Borehole-flow calibration targets drawdown, comparing 12. Calibrated steady-state predevelopment and 2010 6. Lake-level calibration targets recharge distribution 34 conditions 47 7. A. Spring flow calibration targets 13. Steady-state hydraulic head, 24. Steady-state water balance, B. SFR2 nodes representing vertical head difference, and comparing predevelopment springs in the model lake-level calibration results with and 2010 conditions 48 8. Surface water flow calibration calibration summary statistics . 35 targets 9. Transient calibration targets Wisconsin Geological and Natural History Survey Abstract new groundwater flow model as an upper aquifer (layer 9) with a Model calibration used the parameter for Dane County, Wisconsin, bedding-plane fracture feature (layer estimation code PEST, and calibration Areplaces an earlier model 10) at its base. The Eau Claire aquitard targets included heads, stream and developed in the 1990s by the (layer 11) includes shale beds within spring flows, lake levels, and borehole Wisconsin Geological and Natural the upper portion of the Eau Claire flows. Steady-state calibration focused History Survey (WGNHS) and the Formation. This layer, along with over- on the period 2006 through 2010; the U.S. Geological Survey (USGS). This lying bedrock units, is mostly absent transient calibration focused on the modeling study was conducted in the preglacially eroded valleys 7-week drought period from late May cooperatively by the WGNHS and the along the Yahara River valley and in through July 2012. USGS with funding from the Capital northeastern Dane County. Layer 12 This model represents a significant Area Regional Planning Commission represents the Mount Simon sand- step forward from previous work (CARPC). Although the overall con- stone as the lowermost model layer. because of its finer grid resolution, ceptual model of the groundwater It directly overlies the Precambrian improved hydrostratigraphic dis- system remains largely unchanged, crystalline basement rock, whose top cretization, transient capabilities, the incorporation of newly acquired surface forms the lower boundary of and more sophisticated representa- high-quality datasets, recent research the model. tion of surface-water features and findings, and improved modeling and The model uses the USGS multi-aquifer wells. calibration techniques have led to MODFLOW-NWT finite-difference the development of a more detailed code, a standalone version of and sophisticated model represen- MODFLOW-2005 that incorpo- tation of the groundwater system. rates the Newton (NWT) solver. The new model is three-dimensional MODFLOW-NWT improves the and transient, and conceptualizes the handling of unconfined conditions county’s hydrogeology as a 12-layer by smoothing the transition from system including all major unlithified wet to dry cells. The model explicitly and bedrock hydrostratigraphic units simulates groundwater–surface-water and two high-conductivity horizontal interaction with streamflow routing fracture zones. and lake-level fluctuation. Model Beginning from the surface down, input included published and the model represents the unlithified unpublished hydrogeologic data from deposits as two distinct model layers recent estimates of aquifer hydraulic (1 and 2). A single layer (3) simu- conductivities. A spatial ground- lates the Ordovician sandstone and water recharge distribution was dolomite of the Sinnipee, Ancell, and obtained from a recent GIS-based, Council Circle Spring Prairie du Chien Groups. Sandstone soil-water-balance model for Dane Mike Parsen of the Jordan Formation (layer 4) and County. Groundwater withdrawals silty dolostone of the St. Lawrence from pumping were simulated for Potential applications of the model Formation (layer 5) each comprise 572 wells across the entire model include evaluation of potential sites separate model layers. The underlying domain, which includes Dane County for and impacts of new high-capacity glauconitic sandstone of the Tunnel and portions of seven neighboring wells, development of wellhead pro- City Group makes up three distinct counties—Columbia, Dodge, Green, tection plans, evaluating the effects layers: an upper aquifer (layer 6), a Iowa, Jefferson, Lafayette, and Rock. of changing land use and climate on fracture feature (layer 7), and a lower These wells withdrew an average of groundwater, and quantifying the aquifer (layer 8). The fracture layer 60 million gallons per day (mgd) over relationships between groundwater represents a network of horizontal the 5-year period from 2006 through and surface water. bedding-plane fractures that serve 2010. Within Dane County, 385 wells as a preferential pathway for ground- were simulated with an average with- water flow. The model simulates the drawal rate of 52 mgd. sandstone of the Wonewoc Formation 1 The 2016 Groundwater Flow Model for Dane County, Wisconsin Introduction he 2016 Dane County regional of the 1996 model, including no tran- opportunities to leverage existing groundwater model is part of sient calibration, fixed lake and stream data sets and update the ground- Tan ongoing effort to improve levels, and relatively coarse discretiza- water flow model. understanding of the hydrologic sys- tion, provided additional motivation The regional-scale groundwater mod- tem within Dane County, Wisconsin. to build an improved model. eling of Dane County has been made This model replaces a 1996 regional Although many of the same ground- possible through ongoing funding groundwater flow model (Krohelski water modeling needs remained rele- by the CARPC, the Dane County Land and others, 2000). The new model vant to the development of the 2016 and Water Resources Department, builds upon improvements in our model, new societal concerns have the Madison Metropolitan Sewerage understanding of the hydrogeology emerged over the past decade as District, the Madison Water Utility, of Dane County and takes advantage our understanding of Dane County’s and other municipalities and public of additional data collection efforts as hydrologic system has improved. water utilities across Dane County. well as advances in computer capabil- The recent interest in groundwater– This continued support has led to ities, data management, groundwater surface-water interactions, which the development, use, and regular modeling, and calibration. require greater spatial resolution for improvement of the groundwater Following its development, the orig- accurate representation, and the need model, ensuring its applicability and inal 1996 model has been regularly for a transient model with improved utility as a decision-support tool for used to evaluate a host of questions model calibration methods were years to come. ranging from long-term impacts of the primary drivers motivating the The WGNHS partnered with the USGS groundwater pumping and well siting development of an updated regional Wisconsin Water Science Center for to numerous site-specific research groundwater flow model. The release this modeling project. and applied environmental investi- of countywide lidar (light + radar) gation studies. Although the 1996 data in 2010, long-term baseflow model has remained a valuable tool estimates by USGS researchers in Scope for assessing groundwater resources, 2011 (Gebert and others, 2011), the The model area, shown in figure 1, our understanding of hydrogeology availability of numerous high-quality comprises Dane County and parts of across the county has improved, as geophysical logs, and a countywide seven adjacent counties (Columbia, has the data management, mod- groundwater recharge model (Hart Dodge, Green, Iowa, Jefferson, eling, and calibration software. and others, 2012) developed by the Lafayette, and Rock). Including Furthermore, certain simplifications WGNHS in 2009 provided further neighboring counties was necessary because the hydraulic boundaries of the groundwater system extend out- side Dane County. Water-use data for high-capacity wells and subsurface hydrogeologic data for areas beyond Dane County’s borders were com- piled and used in the updated model. When available, we also used existing data for neighboring counties, such as recharge estimates for Columbia County. In the absence of additional information, data and published maps for Dane County were extrapolated into neighboring counties.

Spring Creek

Mike Parsen 2 Wisconsin Geological and Natural History Survey

Figure 1. Location and extent of the Dane County model showing edge boundary conditions.

Rocky Run r ive h R n s si w

no- ow no- ow a on Cr isc ch W an ke Br La N SAUK COLUMBIA DODGE

C ra 51 w G s 39 h r l R e i iv v R 90 a 151 e in r ns o 94

sc c

L i L

W a

a i

t t

e a

r W t C

DANE

s 14 i d s e

n c 12

u o d

o n M

s e D in u A l 94 B r g la r i c i e f a l Madison l a t i m l i 12 18 e t

s 39 s 18 151 no- ow boundary 14 90

t

A t JEFFERSON

i i

r

m

m e i

IOWA r i v

E i

l l R

51 B

l k

r e c

a a o i R n g c c n a o h a n l o

P hk e g s o c K a n L t a o i n o i c n a i R l GREEN l ROCK iv I er no- ow boundary

0 5 miles

0 100 miles

3 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Land use in Dane County is pre- Setting dominantly agricultural, with most Important features Dane County is located in activity directed toward dairy farming of the new model south-central Wisconsin and straddles and row crops. The centrally located Although the overall area simulated the boundary between the Driftless Madison metropolitan area, com- by the 2016 model is essentially Area of southwestern Wisconsin and posed of the city of Madison and the the same as was simulated in the the area covered by the Laurentide adjacent cities of Monona, Middleton, 1996 model, the new model incor- Ice Sheet during the Wisconsin Fitchburg, and Verona, is the largest porates the following improve- Glaciation (Clayton and Attig, 1997). population center in the county. As ments to the conceptualization The western part of the county is of the 2010 census, the total popu- and numerical representation located within the Driftless Area, a lation of Dane County was 488,000 of the groundwater system: region of older landscapes not mod- (U.S. Census Bureau, 2012). ified by glacial ice. The Driftless Area Additional data typically contains dissected uplands ❚❚ Re-evaluation of existing hydro- and well-developed surface-water Objectives This project had three major geologic and hydraulic data and drainage systems. Hills are generally objectives: incorporation of recently acquired flat-topped and are often used for high-quality datasets including pastureland and row crops. Slopes are 1. Collect data to support geophysical logs, water-level and steep and commonly forested. In con- model improvement, includ- vertical-head-difference measure- trast, the glaciated eastern two-thirds ing refining hydrostratigraphy, ments, borehole-flow measure- of the county is characterized by roll- characterizing aquifer proper- ments, pumping test results, and ing, moderately hilly topography. The ties, updating water-use data, stream baseflow measurements. eastern part of the county contains and refining recharge rates. numerous drumlins. The drainage sys- ❚❚ A more complete evaluation 2. Refine the conceptual and tem is not as well developed and the of countywide water-use data, numerical model, including region contains lakes and marshes. including all active and inactive improving simulation detail high-capacity wells through 2010 According to the National Climatic through finer grid spacing than (historical, industrial, agricultural, Data Center, the average annual used in previous models, add- commercial, and municipal). precipitation in Dane County, as ing simulations of springs plus measured at the Dane County groundwater interactions with the ❚❚ Use of more-detailed estimates of Regional Airport, was 34.48 inches Yahara Lakes, and calibrating to recharge distribution from recently per year between 1981 and 2010. For transient conditions. published soil-water-balance mod- the same period, the mean annual eling for Dane County (Hart and 3. Develop a simulation tool for air temperature was 46.5°F, with an others, 2012). informing management deci- average monthly maximum of 81.6°F sions regarding groundwater use ❚❚ Incorporation of recent laboratory in July and an average monthly and development in Dane County. and field testing to better estimate minimum of 11.1°F in January (http:// Typical uses include simulating effective porosity of different aqui- www.ncdc.noaa.gov/oa/climate/ contributing areas for municipal fer and aquitard units. normals/usnormals.html). Sixty wells, examining current ground- percent of annual precipitation falls ❚❚ Incorporation of recently acquired water levels and flow directions, between May and September. lidar data on county topography, and establishing a starting point which was particularly useful for for future site-specific studies developing the streamflow routing using refined models. network used in the model.

4 Wisconsin Geological and Natural History Survey

Improvements to the MODFLOW 2005 code (Harbaugh, numerical model 2005) that incorporates a Newton Model distribution solver package, which greatly ❚❚ Refined grid spacing from 1,320 to and use improves model stability by 360 feet (ft) horizontally and from The groundwater flow model allowing dry nodes in the model to 3 layers to 12 layers vertically to described here is in the public remain part of the model solution provide greater spatial resolution. domain. The model files are avail- (for an example, see Feinstein and able both in native MODFLOW and ❚❚ Simulation of all significant others, 2012). in proprietary Groundwater Vistas streams in the county using the ❚❚ Transient capabilities to allow sea- (Environmental Simulations, Inc.) MODFLOW Streamflow-Routing sonal predictions of water levels formats. A companion user’s guide (SFR2) package (Niswonger and and streamflow. (Bradbury and others, 2016) includes Prudic, 2005). The 1996 model detailed lists of files required and simulated streams using the ❚❚ Advanced calibration using a instructions for running the model. less-sophisticated River (RIV) highly parameterized approach package, where stream-aquifer (Doherty and Hunt, 2010) for both interactions are not constrained steady-state and transient models. by a water balance in the stream, and streamflow routing is not explicitly simulated. The SFR2 package simulates a water bal- ance in each stream cell, which Acknowledgments includes streamflow routing The authors thank the Capital Area Regional Planning Commission components, and can simulate (CARPC), Dane County Department of Land and Water Resources, inflows and outflows between Madison Metropolitan Sewerage District (MMSD), Madison Water Utility streams and lakes. Stream stage is as well as many municipalities and water utilities across Dane County for computed from the water bal- their support of this groundwater modeling study. ance, and can limit stream-aquifer interactions (for example, losing We would also like to thank the following people who played a signifi- streams can go dry, shutting cant role in helping us to update and improve the model: Mike Kakuska off leakage to the aquifer). and Kamran Mesbah of CARPC; Joe DeMorett of the Madison Water Utility; Dave Taylor of MMSD; Brynn Bemis of the City of Madison; Kevin ❚❚ Simulation of major lakes in the Connors, Jeremy Balousek, and Michelle Richardson of the Dane County county using the MODFLOW Land and Water Resources Department; Jessica Meyer of the University Lake (LAK3) package. The 1996 of Guelph; Eric Oelkers of SCS Engineers; Ken Quinn of TRC; Steve model simulated lakes as static Gaffield of Montgomery Associates; Mike Schmoller and Matt Diebel of constant-head features. The the Wisconsin DNR; Sue Swanson of Beloit College; and David Liebl of new model allows lake levels to the University of Wisconsin–Extension. fluctuate dynamically in response to a simulated lake-water balance, Staff members and students at the Wisconsin Geological and Natural which includes groundwater flow, History Survey who greatly contributed to this project include Bill precipitation, evaporation, as well Batten, Pete Chase, Linda Deith, Anna Fehling, Madeline Gotkowitz, Dave as stream inflow and outflow. The Hart, Jake Krause, Irene Lippelt, Steve Mauel, Sam Patrick, Roger Peters, LAK3 package also allows lake Caroline Rose, Pete Schoephoester, and Greg Tracy. area, volume, and stream outflow U.S. Geological Survey staff contributing to this project include Cheryl to vary as a function of stage. Buchwald, Mike Fienen, Jim Kennedy, Rich Niswonger, Jason Smith, and ❚❚ Improved simulation of John Walker. multi-aquifer or cross-connecting Colleague reviews by Dave Hart (WGNHS), Andy Leaf (USGS), Robert wells using the MODFLOW Nauta (independent hydrogeologist), and the CARPC staff significantly Multi-Node Well (MNW2) package. improved this report. ❚❚ Use of MODFLOW-NWT (Niswonger and others, 2011), a stand-alone version of the USGS  5 The 2016 Groundwater Flow Model for Dane County, Wisconsin Study methods Attig (1997) and Brown and others Review of (2013), and studies by Fritz (1996) and Geophysical logging previous studies Aswasereelert and others (2008) also Modern downhole geophysical log- ging is an important tool for under- The initial phase of the project provided insight into the regional standing subsurface hydrostratigra- involved review of geological and hydrogeology. The development of phy. Numerous geophysical logs have hydrogeological studies conducted improved techniques for estimating been collected in Dane County over within Dane County over the past recharge by Dripps and Bradbury the past several decades, typically several decades. Early groundwater (2007) and Westenbroek and others including vertical profiles of tempera- modeling (McLeod, 1975) served (2010) and their application to Dane ture, fluid conductivity, resistivity, as a historical reference point for County by Hart and others (2012) natural gamma radiation, and bore- development of more recent mod- also improved understanding of the hole diameter (caliper). Many more els in Dane County. Regional-scale spatial distribution of recharge within recent logs also include optical and groundwater characterization by the the county. acoustic borehole imaging as well as USGS and WGNHS in the 1990s and borehole-flow measurements. These 2000s (Bradbury and others, 1999; Location and additional data can provide more Krohelski and others, 2000) repre- identification of public detailed information about hydro- sented a major improvement in the stratigraphy, including the spatial conceptual model and numerical and private wells extent and thickness of aquitards and representation of the groundwater The location and pumping rates of evidence for preferential flow along system. A series of inset models based both public and private high-capacity apparent bedding-plane fractures. on the regional Dane County model wells were compiled by the USGS into Appendix 1 contains an example of a were subsequently developed to a water-use database. A high-capacity complete suite of geophysical logs. investigate flow in fractured bedrock well is, by definition, any well that and spring systems at local scales. The is constructed on a high-capacity The high quality of geophysical logs, inset models led to improved under- property, where the pumping rate compared to other sources of subsur- standing of preferential flow through from all combined wells is equal to face data, such as well construction fracture features and their importance or exceeds 70 gallons per minute, reports and geologic logs from drill for regionally fed spring systems or roughly 100,000 gallons per day cuttings, make them a primary source (Hunt and Steuer, 2000; Parent, 2001; (gpd) (Wisconsin Administrative of data for delineating distinct hydro- Swanson, 2001, 2007; Swanson and Code NR 812.07). stratigraphic units. others, 2001, 2006; Anderson, 2002), Water-use estimates were made for all and other areas of interest such as wells based on data available from the Water-use survey and Fish and Crystal Lakes (Krohelski, Wisconsin Public Service Commission 2002). Work by McLeod (1974), Meyer data compilation (PSC) and from individual water (2005, 2013), Meyer and others (2008), Following the collection and com- utilities. Pumping rates were aver- Macholl (2007), Bahr and others pilation of pumping data for all aged over the 5-year period of 2006 (2010), Gellasch and others (2013), high-capacity wells, a water-use through 2010 for the steady-state and Sellwood and others (2014) survey was sent to all 27 public water model. Historical pumping rates further demonstrated the importance utilities within Dane County during were also obtained, from as early as of preferential flow through fracture the spring of 2012. Results from the late 1800s, for the purposes of networks and better characterized this survey confirmed the location evaluating changes in pumping since groundwater flow within the county. and pumping rate for each munic- predevelopment. ipal well and provided information Foundational research on the regional The steady-state model includes 572 about future pumping. Wells that geology and hydrology by Cline wells. The locations and well construc- had recently become inactive or (1965), Ostrom (1965), Olcott (1972), tion characteristics of each well were selected for future abandonment and Day and others (1985) provided a checked against records in the DNR well were identified and modified accord- basis for reinterpreting the hydrostra- construction reports (WCRs) as well as ingly in the water-use database. tigraphy of Dane County. More recent information maintained by the WGNHS. geologic mapping by Clayton and 6 Wisconsin Geological and Natural History Survey

A survey was also conducted for all at the outlets of Lakes Mendota, Dane County municipal wastewater Stream and spring Monona, Waubesa, and Kegonsa were treatment facilities in the spring of flow measurements calculated based on an evaluation of 2012 to estimate discharge rates Baseflow measurements for perennial historical time-series streamflow data to surface water during 2010. Of streams in Dane County were used from USGS stream gaging stations the 15 treatment plants surveyed to calibrate the model. A statewide along the Yahara River. in Dane County, the following 12 study to estimate average annual reported effluent data: Belleville, Blue recharge from 1970 to 1999 by the Recharge estimation Mounds, Cambridge, Cross Plains, USGS (Gebert and others, 2011) Groundwater recharge in Dane Deerfield, Madison Metropolitan provided estimates of baseflow County comes from precipitation Sewerage District (MMSD), Marshall, for 66 partial-record stations and (rain and snowmelt) at the land Mazomanie-Arena, Mount Horeb, eight long-term stream gaging sta- surface. A recent investigation Oregon, Stoughton, and Sun Prairie. tions across the study area. During the estimated recharge across Dane For the remaining three, we estimated course of the modeling project, the County (Hart and others, 2012) using discharge rates by applying a 6 per- WGNHS made 23 one-time stream- the soil-water-balance code (SWB) cent increase (that is, the average per- flow measurements and 18 one-time (Westenbroek and others, 2010). SWB cent increase for the reporting treat- spring flow measurements within uses readily available soil type, land ment plants between 2000 and 2010) the study area. In most cases these cover, topographic, and climatic data to the 2000 rates published by CARPC measurements were made under to estimate groundwater recharge. (Dane County Regional Planning low-flow conditions and at least The SWB recharge model accounts Commission, 2004). The MMSD several days after a rainfall event. For for precipitation, evapotranspira- treatment plant discharges to two a DNR study to evaluate the aquatic tion, interception, surface runoff, locations, one on Badfish Creek and health of stream habitats across Dane soil-moisture storage, and snow- another at the headwater of Badger County (Diebel and others, 2014), melt at daily time steps. The spatial Mill Creek, raising the total number over 100 flow measurements were distribution of groundwater recharge of Dane County discharge locations made within the study area. Eighty was estimated for both present and accounted for by the model to 16. of these measurements were located past climate and land-use conditions. A final water-use survey was con- along perennial streams and included The results indicated significant ducted during the fall of 2012 to in model calibration. Four additional temporal variability in average annual obtain transient pumping data for estimates of streamflow representing infiltration to the unsaturated zone the drought period from late May to a 75 percent flow exceedance (Q75) in response to climatic variability. July 2012 from all municipal water utilities. We obtained withdrawal data from the following 11 water utilities in Dane County: DeForest, Fitchburg, Madison, McFarland, Middleton, Mount Horeb, Oregon, Stoughton, Sun Prairie, Verona, and Waunakee. These data were then extrapolated to other wells for which no withdrawal data were available for this period.

Starkweather Creek—stream gaging

Mike Parsen 7 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Estimated recharge varied spatially set of subsurface data. An initial cells. The model is transient and across the county from less than interpretation of hydrostratigraphic three-dimensional. It explicitly sim- 5 inches to more than 15 inches in a contact elevations for bedrock was ulates groundwater–surface-water typical year. made using geophysical logs. The interaction with streamflow rout- lateral extent and thickness of each ing and lake-level fluctuation. The Hydrostratigraphy hydrostratigraphic unit was then con- Groundwater Vistas graphical user Hydrostratigraphic interpretation was strained using the remaining subsur- interface was used to facilitate model important for discretizing the model face point data and existing geologic, input and visualize model output. The layering. Geologic and hydrogeologic depth-to-bedrock, and bedrock model was calibrated using param- data were compiled, including well elevation maps. eter estimation code (PEST) and construction reports with driller’s guidelines outlined by Doherty and lithological descriptions, geologi- Numerical Hunt (2010). Calibration targets used cal logs of drill cuttings and core as simulation methods for history matching included heads, described by WGNHS geologists, stream and spring flows, lake levels, The model uses the USGS geophysical logs collected by WGNHS and borehole flows. Steady-state MODFLOW-NWT finite-difference staff and other researchers, as well calibration focused on the period code (Harbaugh, 2005; Hunt and as Pleistocene, bedrock geology, and between 2006 and 2010; the tran- Feinstein, 2012), with a Newton bedrock elevation maps for Dane sient calibration simulated a 7-week solver to improve the handling of County. Geophysical logs represented drought period between late May and unconfined conditions by smooth- the highest quality and most reliable July 2012. ing the fluctuation of wet and dry

Culver Springs

Ken Bradbury

8 Wisconsin Geological and Natural History Survey Geology, hydrostratigraphy, and hydrology of Dane County down a surface topography compara- of Precambrian age. The top of Quaternary geology ble to that of the Driftless Area; low- this Precambrian surface varies in and hydrostratigraphy ering hills and filling low-lying areas elevation between roughly 400 ft Dane County straddles the boundary with debris and sediment (Clayton below and 400 ft above mean sea between the glaciated area to the and Attig, 1997). Modern-day fluvial level within Dane County. This surface east and the unglaciated, or Driftless, and lacustrine sediments were sub- elevation was developed by using area to the west (fig. 1). Weathering sequently deposited in lakebeds and information from 58 wells in Dane and glacial processes over the past along river and stream channels in County and an additional 300 wells tens of thousands of years have both the glaciated and Driftless Area. from across southern Wisconsin and northern Illinois, which were drilled to created a diverse range of landforms, The till dominating the eastern por- Precambrian bedrock. Although little surface-water features, and depos- tion of the county consists primarily is known about the hydrogeologic its of glacial, fluvial, lacustrine, and of gravelly, clayey, silty sand, which properties of the crystalline bedrock eolian materials across the landscape, exhibits a relatively uniform range in Dane County, it is known to have over the bedrock. Several geologists, of hydraulic conductivity of about little porosity and limited hydraulic including Alden (1918), Mickelson 0.5–0.7 ft/day in field tests (Rayne conductivity (Bradbury and others, (1983), and Clayton and Attig (1997) and others, 1996). Offshore sedi- 1999). The Precambrian surface is have described these deposits and ment of glacial lakes and postglacial, directly beneath several hundred feet provide a framework for interpreting pre-modern lakes also form import- of permeable sandstone, and, for the their hydrogeologic significance. ant landscape features in central purposes of this model, represents and eastern Dane County. In several Across the county, the overall depth the base of the aquifer system. of Pleistocene sediment varies from areas, these deposits of fine-grained absent to a thin cover of windblown silt and clay overlie coarser-grained Paleozoic bedrock of the Elk Mound and hillslope sediment on the uplands till, forming a confined or par- Group overlies the Precambrian and valley walls of the Driftless Area, tially confined aquifer within the crystalline rocks throughout all of to several hundred feet of unlithified unlithified sediments (Clayton and Dane County. The Elk Mound Group sediments within the Yahara River and Attig, 1997). The presence of this consists of units of sandstone and Wisconsin River valleys (Clayton and confined unlithified aquifer sys- shale including, from oldest to Attig, 1997). Most of the Pleistocene tem was mapped by Fritz (1996). youngest, the Mount Simon, Eau sediment in Dane County is com- Claire, and Wonewoc Formations. posed of till of the Horicon Member Bedrock geology and The Mount Simon Formation is made of the Holy Hill Formation, which was hydrostratigraphy up of coarse- to medium-grained deposited by the Green Bay Lobe sandstone, which ranges in thickness The bedrock geology of Dane County of the Laurentide Ice Sheet during across Dane County from roughly consists of Precambrian age crystal- the Late Wisconsin glaciation. Earlier 100 ft in the northeast to 800 ft in line rock overlain by successive units deposits from the early and middle the south. Across most of the county, of younger Paleozoic age sandstone Pleistocene are believed to be present the Mount Simon ranges from 300 and dolomite (fig. 2). Cline (1965) and in some of the deeper preglacial to 600 ft thick. The Mount Simon is Ostrom (1965) describe these units valleys beneath the modern-day a major aquifer, which serves as the in detail, while more recent mapping Yahara and Wisconsin Rivers. These water supply for many of the largest by Brown and others (2013) inter- earlier sediments would have been high-capacity wells in Dane County. prets the spatial extent of these units. deposited on the preglacial, erosional The coarse- to medium-grained sand- Investigations by Mickelson (1983) topography that was carved into the stone of the Mount Simon Formation and Clayton and Attig (1997) also Paleozoic bedrock surface millions gradually transitions upwards to the discuss these bedrock units. of years before glaciers first reached fine-grained, silty sandstone of the Dane County. As glaciers moved into The base of the system consists of Eau Claire Formation. Interbedded eastern Dane County, they ground igneous and metamorphic rocks shales occur in the lower quarter of the Mount Simon, and the formation 9 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Figure 2. Bedrock stratigraphy and corresponding layers in the 1996 and 2016 groundwater models. (General bedrock stratigraphy adapted from Brown and others, 2013.)

GENERAL BEDROCK STRATIGRAPHY GROUNDWATER FLOW MODEL Age Stratigraphic name Model layers, names Era Period Group Formation 1996 model 2016 model Type Unlithified I (fine-grained lake deposits within 1 glacial Lake Yahara area; elsewhere, 1 Sand and gravel till and meltwater stream deposits) Unlithified II 2 (till and meltwater stream deposits) Maquoketa Galena Sinnipee Decorah Ordovician Platteville 3 Upper bedrock aquifers Glenwood Ancell St. Peter Prairie du Chien

Jordan 2 Upper bedrock 4 Jordan Trempealeau St. Lawrence 5 St. Lawrence

Paleozoic 6 Tunnel City—upper Lone Rock, Tunnel City Tunnel City 7 Mazomanie (fracture layer) Cambrian 8 Tunnel City—lower 9 Wonewoc Wonewoc Wonewoc 10 Elk Mound (fracture layer) Eau Claire —a Confining unit 11 Eau Claire aquitard Mount Simon 3 Lower bedrock 12 Mount Simon aquifer Precambrian Various unnamed units No-flow boundary a In the 1996 model, the shaley part of the Eau Claire Formation was represented by a leakance term to account for the unit's vertical hydraulic conductivity and thickness.

10 Wisconsin Geological and Natural History Survey coarsens and contains numerous peb- interval of the Eau Claire Formation is surface fell below this layer. Outcrops bles near its basal contact with the roughly 70 ft thick in that vicinity. This of the Wonewoc are exposed along underlying Precambrian rocks. shale interval is particularly distinct the banks of the Wisconsin River from the natural gamma signature valley. Recent studies at multiple field The upper portion of the Eau Claire in numerous geophysical logs. The sites within Dane County provide Formation contains significant absence of this gamma signature evidence for preferential flow of amounts of shale and siltstone, form- in northeastern Dane County sug- groundwater along bedding-plane ing an important aquitard, known as gests the absence of the Eau Claire fracture features within the sand- the Eau Claire aquitard, across much aquitard in these areas. The top-of- stone of the Wonewoc Formation of the county. The shaley interval near bedrock elevation surface developed (Anderson, 2002; Macholl, 2007; Bahr the top of the Eau Claire Formation by Olcott (1972) suggests that the and others, 2010; Gellasch and others, is absent in well logs in northeastern Eau Claire aquitard is also absent 2013; Sellwood and others, 2014). Dane County. This shaley interval beneath parts of the Yahara Lakes. is also absent beneath portions of The Tunnel City Group overlies the central Dane County where erosion The overlying Wonewoc Formation Elk Mound Group and consists of along the pre-glacial Yahara River consists of medium- to fine-grained medium to very fine-grained glau- valley scoured down through this sandstone, forming an important conitic sandstone. In Dane County, layer into the Mount Simon aquifer. regional aquifer system above the the Tunnel City Group ranges from The Eau Claire aquitard ranges in Eau Claire aquitard. Sandstone of absent to roughly 140 ft thick. thickness from absent to over 70 ft the Wonewoc Formation varies in Although these rocks yield less than in western Dane County (fig. 3). A thickness from absent to roughly the underlying sandstones of the Elk geophysical log of WGNHS test hole 200 ft. The Wonewoc is absent within Mound Group, the sandstone of the along County Highway A in eastern portions of the Yahara River, Black Tunnel City Group is an important Iowa County (WGNHS ID: 25000512), Earth Creek, and Wisconsin River source of groundwater to wells in indicates that the shaley facies valleys where the preglacial erosional many parts of Dane County. In central

Figure 3. Thickness and extent of the Eau Claire aquitard. Includes locations of geophysical borehole logs used in the interpretation.

Thickness, Eau Claire (layer 11)

70

50 ! Geophysical 70 10 70 borehole log ! 50 ! ! ! 30 ! Thickness of Eau ! Claire aquitard (ft) ! 74 ft 30 ! ! ! contour ! ! interval 10 ft ! ! ! ! ! ! ! ! ! ! ! < 1 ft ! ! ! ! ! ! Layer absent !!!! ! ! ! ! ! ! Interstate ! ! ! highways US highways

Major streams 10 and lake outlines

0 5 miles

11 The 2016 Groundwater Flow Model for Dane County, Wisconsin

0.5 ft when imaged in borehole walls and can transmit significant quan- tities of water. Gellasch and others (2013) showed that such fractures can dominate the transmissivity of an individual borehole in Madison and reported hydraulic conductivities ranging from 26 to 1,560 ft/day mea- sured using short-interval (2.3 feet) straddle packers. Swanson and others (2006) observed similar fractures in an investigation of the Nine Springs basin in Dane County. They reported fracture hydraulic conductivities as high as 1,130 ft/day and were able to simulate springs in a numerical flow model by including a fracture layer having a hydraulic conductivity of White Clay Spring 400 ft/day. WGNHS scientists have Mike Parsen commonly observed such fractures in Dane County, the Tunnel City Group above this unit and acts as a minor bedrock wells installed in the county. outcrops at or near the level of the aquifer where saturated. The com- These fractures, which appear to cor- principal lakes within the Yahara bined thickness of these two bedrock relate from well to well, are generally watershed. In many areas, the Tunnel units varies from absent to roughly found about midway between the top City Group sandstone is the first 60 ft thick across Dane County. and bottom of the Tunnel City Group bedrock unit encountered at depth and within a few feet of the bottom of The uppermost Paleozoic units of along the Madison Isthmus and other the Wonewoc Formation. Ordovician age are above the water low-lying areas near the Yahara lakes. table in most of the county and were Recent investigations by Swanson Surface-water features lumped into a single model layer. (2001, 2007), Swanson and others Dane County contains a diversity This layer includes dolostone of (2006), and Meyer and others (2008) of geomorphologic landforms the Prairie du Chien Group, dolos- demonstrated the existence of dis- that influence the distribution of tone and sandstone of the Ancell crete bedding-plane fractures within surface-water features across the Group, shale and dolostone of the the Tunnel City Group that contribute county. The county consists of five Sinnipee Group, and shale of the to the preferential flow of ground- distinct physiographic areas including Maquoketa Formation, which forms water. Many of the larger spring the Wisconsin River valley, valleys and a thin cap at the top of the highest features in central Dane County, for ridges of the Driftless Area, hills and ridge at Blue Mounds State Park. example Nine Springs, Culver Springs, hummocks between the Johnstown Frederick Springs, and several springs Bedding-plane fractures and Milton moraines, the Yahara River Valley, and drumlins and marshes around Lake Wingra are likely associ- Over the past decade, downhole of eastern Dane County (Day and ated with these fracture features. investigations using video logs, others, 1985). Within these areas, The upper Cambrian bedrock units optical-borehole imaging (OBI), four principal surface watersheds are of the Trempealeau Group is made borehole flow meters, and other commonly demarcated: the Wisconsin up of the St. Lawrence and Jordan technology have regularly detected River basin in the northwest; the Formations. The St. Lawrence is clas- permeable horizontal discontinuities Sugar and Pecatonica River basins to sified as a silty dolostone or dolomitic in bedrock wells in Dane County and the west and south; the Yahara River siltstone, contains trace amounts of elsewhere. These discontinuities usu- basin forming a central corridor from glauconite (Brown and others, 2013), ally appear coincident with bedding north to south; and the Crawfish River, and exhibits relatively low hydraulic planes and are typically nearly hori- Koshkonong Creek, and Maunesha vertical conductivity. The sandstone zontal in orientation. These features River basin to the east (fig. 4) (Day of the Jordan Formation lies just range in thickness from about 0.05 to 12 Wisconsin Geological and Natural History Survey and others, 1985). Each basin contains lock and dam structure at Tenney Park In addition to the five major lakes a variety of surface-water features and makes its way across the Madison within the Yahara River watershed, including lakes, wetlands, streams, Isthmus to Lake Monona. Lake Wingra, there are dozens of smaller named and springs. There are also a num- which is fed by several local springs lakes. Most named lakes are located ber of municipal treatment plants near the UW-Madison Arboretum, is within kettles, low-lying marsh whose treated effluent contributes also connected to Lake Monona via areas, or behind impoundments and to the flow of various streams across Wingra Creek. From Lake Monona, dams. Lakes at higher elevations in the county. the gradient flattens and the Yahara the watershed are typically iso- River flows through several marshes lated seepage lakes with no natural Lakes and wetlands and shallow lakes before entering stream outlets or headwater lakes The largest lakes in Dane County are Lake Waubesa near McFarland. Due to that feed perennial streams. Lakes located within the central Yahara the low hydraulic gradient between lower in the landscape are usually River corridor and form a chain these lakes, backwater effects (where flow-through lakes connected to connected by the Yahara River and a water backs up due to an obstruc- streams or low-lying seepage lakes. series of smaller streams, lakes, and tion) are common along this stretch Fish and Crystal Lakes, seepage lakes wetlands. From north to south the of the Yahara River. At the outlet of in northwest Dane County, are the Yahara River enters Lake Mendota Lake Waubesa, the Yahara River again two largest lakes outside of the Yahara through Cherokee Marsh, a broad passes through several wetlands and River basin that are within the model wetland area that receives additional shallow lakes before entering Lake domain. Water levels in Fish and streamflow from Token Creek and Kegonsa north of Stoughton. Once Crystal Lakes have varied significantly the watershed upstream of Cherokee the Yahara River leaves Lake Kegonsa, over the past decades with more Marsh. At the outlet of Lake Mendota, it continues south until joining the recent flooding (Krohelski, 2002). the Yahara River abruptly drops 5 ft in Rock River southwest of Edgerton. elevation as water passes through the

Figure 4. Major hydrologic features in Dane County.

! ! ! ! ! Major hydrologic Crystal ! Fish Lake ! Lake ! ! ! !! . ! ! features R ! n ! in ! ! ! ! ! s #* i !! ! ! ! n ! ! o s ! ! ! ! ! c ! ! ! ! is a ! ! ! ! ! ! W ! !! ! ! ! M B !! !! ! a ! ! ! u ! ! ! ! ! ! n r ! ! e . ! s ! !! e ! ! h R ! *# ! ! ! ! ! a Wastewater v ! ! a ! ! R ! i ! ! r ! ! ! ! ! . ! R ! ! a discharge locations ! ! h ! ! ! ! #* a ! ! ! n ! ! #* i ! Y ! !! ! ! ! ! ! s !! ! #* nB !! ! ! ! Springs l o! ac ! Craw sh- c k E Yahara !!Basin ! ! ! s ! ar ! ! ! ! ! i th ! ! ! High-capacity Cr ! ! ! W ! . ! ! ! ! ! !Maunesha- ! #* ! ! ! wells ! ! ! ! ! L. Mendota ! ! ! ! ! Koshkonong ! ! ! ! ! ! ! ! ! ! Watershed !! ! ! ! ! ! ! !! ! ! ! Basin boundaries ! L. Monona K !! ! o !! ! L. Wingra ! s #*! !! hk ! ! ! !!! ! ! ono Cr. Major streams ! ! ! ! !! ! ng ! !!! ! ! ! ! ! ! ! ! Su ! ! ! !! g ! ! ! ! ! and lakes a ! ! ! r ! ! ! ! R !! ! ! ! ! i ! ! ! ! ! v ! !! ! ! #* #* ! e ! ! L. Waubesa ! ! ! ! ! ! r ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! #* #* ! ! !! ! ! ! ! !! ! #* ! #* ! ! ! !! L. Kegonsa ! B ! ad Sugar-Pecatonica Basin! ! ! fi sh ! ! ! #* ! ! ! W ! ! ! C ! ! e ! r ! s ! ! ! . #*! t B ! !!! ! !! ! ! ra ! nc ! h ! Y ! Sug a ar ha R ! r ! . ! !! a ! ! !! R ! ! . 0 5 miles !! !#* ! ! ! ! #* 13 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Rivers and streams The Crawfish River, Koshkonong Common methods for lowering Rivers and streams within Dane Creek, and Maunesha River water- water levels and increasing stream- County lie within four principal shed within eastern Dane County flow include channelizing creeks, surface watersheds (fig. 4). Within the is a relatively low-hydraulic-gradi- installing ditches and tile drains, as Wisconsin River watershed, several ent system characterized by the well as harvesting aquatic plants and streams drain the northern flanks presence of many drumlins and removing debris from stream and of Military Ridge and the western low-lying wetlands and marshes. The river channels. In contrast, techniques edge of the terminal moraine west principal streams are tributaries of for maintaining higher water levels of Middleton, Springfield, and Dane. the Koshkonong River, south of Sun in lakes and streams commonly The major perennial stream net- Prairie and east of Cottage Grove, and include the installation of dams and works include Dunlap Creek, Black the Maunesha and Crawfish Rivers, in impoundments. As an example, along Earth Creek, Blue Mounds Creek, the northeast. Streams meander con- the Yahara River, dam structures and several smaller tributaries to siderably more than in the western at the outlets of Lakes Mendota, Black Earth and Blue Mounds Creeks. and central portions of Dane County Waubesa, and Kegonsa are used to Wastewater discharges at Cross and are often connected to drainage manage maximum and minimum lake Plains and Mazomanie contribute to ditches and fed by field tiles that were levels while extensive weed cutting streamflow along Black Earth Creek; installed to manage water levels and maintains more consistent water flow discharge at Roxbury is directed convert wetland areas to cultivated during the summer months. acreage. Wastewater discharge from to the Wisconsin River through a Springs dedicated canal and does not enter Sun Prairie, Deerfield, Cambridge, Springs are important hydrologic any streams within the county. and Rockdale contribute flow to the Koshkonong River, while discharge features in Dane County and support The Sugar River and Pecatonica River from Marshall contributes to the diverse ecosystems by providing a watershed, within southwest Dane Maunesha River. steady supply of cool, clear water County, drains the southern slopes to many streams and lakes. In 2007, Within each of the four principal of Military Ridge and the western an inventory of Wisconsin’s springs watersheds, the natural hydrological edge of the terminal moraine. The identified 230 springs in Dane County system has been altered by humans primary streams include the Sugar based on historical records going to manage water levels and stream- River, Sugar River West Branch, Mount back decades (Macholl, 2007). Of flows, preferentially draining water Vernon Creek, and a few tributaries these, over 100 had recorded flow from certain areas while maintain- of the Pecatonica River in southwest rates of less than 0.01 cubic feet Dane County. Wastewater discharges ing elevated water levels in others. at Blue Mounds join the Pecatonica River system to the west; discharges at Mount Horeb, Verona, and Belleville Culver Springs contribute to flow within the Sugar River system. Within the Yahara River watershed, many of the tributaries feed directly into the Yahara lakes. Token Creek west of Sun Prairie, Pheasant Branch Creek in Middleton, Wingra Creek at the outlet of Lake Wingra, Nine Springs Creek near Fitchburg, and Starkweather Creek in Madison are important spring-fed creeks that contribute flow to the Yahara River. Wastewater discharges at Badfish Creek, Oregon, Brooklyn, and Stoughton all contribute to surface-water flow along the lower Yahara River. Grace Graham 14 Wisconsin Geological and Natural History Survey per second (cfs) or 4.5 gpm. Of the discharge it to 16 outlet locations remaining springs, 20 had recorded Water use and (the Madison Metropolitan Sewerage flows of 0.1 cfs to 0.25 cfs (45 gpm to wastewater discharge District treatment plant discharges to 112 gpm), and 20 had recorded flows Within Dane County, the average two locations) (fig. 4). The amount of of greater than 0.25 cfs (112 gpm). groundwater withdrawal rate during water discharged at these locations This set of principal springs is con- the period from 2006 to 2010 was is rarely the same as the amount of centrated within the Yahara River, estimated to be 52 million gallons groundwater originally withdrawn Wisconsin River, and Sugar-Pecatonica per day (mgd) from a total of 385 due to evaporation and gains and River watersheds (fig. 4), with over high-capacity wells (fig. 4). Although losses from the sanitary sewer system half occurring within the Yahara River some residents obtain water from before it reaches the wastewater basin. Springs are notably absent from private low-capacity wells (wells with treatment plant. A portion of pumped the Crawfish-Maunesha-Koshkonong a pumping capacity less than 70 groundwater is commonly applied River watersheds. Work by Swanson gallons per minute), the vast majority directly to the land for irrigation or and others (2006) suggests that of Dane County’s residents, busi- watering lawns and gardens, and preferential groundwater flow along nesses, industries, and farmers rely never reaches the sanitary sewer sys- bedding-plane fractures could on high-capacity wells for their water tem. By contrast, heavy rains or rapid contribute to the location of these supply. Groundwater withdrawal snowmelt can contribute runoff to spring features. Springs tend to be rates were compiled for all public and wastewater. Leaky sanitary sewers can located in areas where bedrock units private high-capacity wells within the also lead to groundwater entering the containing bedding-plane fractures, model domain based on responses to sewer system or wastewater leaking such as the Tunnel City Group, out- water-use surveys. For those utilities from the sewer, leading to gains or crop at the surface or are truncated that did not respond to the survey, losses in the total amount of waste- by younger unlithified sediments at withdrawal rates and well locations water arriving at the treatment plant. depth. Overlying unlithified materials were obtained from PSC, WGNHS, or The largest discharge of treated could then serve as vertical conduits USGS records. These records reflect wastewater in Dane County is from to flow, transporting groundwater the highest quality data available at the Madison Metropolitan Sewerage to its surface outlet at the spring. the time of model construction. District (MMSD), which releases water Of the 52 mgd pumped from to the headwaters of Badfish Creek high-capacity wells in Dane County, near Oregon (41.7 mgd) and Badger 46 mgd are attributable to pump- Mill Creek near Verona (3.3 mgd). ing from 104 municipal supply The discharge to Badger Mill Creek wells within the county. For the was designed to return ground- entire active model domain, which water pumped from within the includes portions of seven neigh- Sugar River watershed, near Verona boring counties, the total number and Fitchburg, back to that same of high-capacity wells is 572 with watershed as surface water. The an average pumping rate between next-largest discharges of treated 2006 and 2010 of 60 mgd. wastewater are from Sun Prairie to Koshkonong Creek (3.2 mgd), Much of the water produced from followed by Stoughton to the wells eventually reaches a waste- Yahara River (1.2 mgd), and Oregon water plant for treatment. Once to Badfish Creek (1.1 mgd). The treated, wastewater is discharged to remaining 11 wastewater treatment nearby streams where it reenters the plants discharge a combined total of hydrologic system as surface water. 2.5 mgd to surface water. See appen- In Dane County there are 15 waste- dix 2 for a full list of treatment plants water treatment plants that process and discharge volumes. approximately 53 mgd of water and

15 The 2016 Groundwater Flow Model for Dane County, Wisconsin Conceptualization of the groundwater system conceptual model of the Tunnel City Group and Wonewoc City Group outcrops directly at the groundwater system is a Formation. The Eau Claire shale under- surface or subcrops (that is, where it Asynthesis and interpretation of lies the Wonewoc sandstone aquifer is truncated at depth by unlithified what is known about the study area, and serves as an important regional deposits). In areas where the Tunnel and/or a collection of hypotheses aquitard between the upper and City Group subcrops, the unlithified about how the groundwater system lower aquifer systems. The Eau Claire deposits are believed to serve as functions, which are subsequently shale is largely absent within the preferential conduits for water to flow tested and refined in the modeling northeastern portion of Dane County vertically to surface springs (Swanson process (Anderson and others, 2015). as well as along the preglacially and others, 2006). The conceptualization of regional eroded valleys of the Yahara lakes Water enters the groundwater flow aquifers, aquitards, and boundary area. The underlying Mount Simon system as recharge to the water table. conditions outlined in the 1996 model aquifer, which consists of sandstones Recharge rates vary across the county (Krohelski and others, 2000) served of both the Mount Simon Formation depending on the topography, soil as the starting point for updating and the lower part of the Eau Claire type, and land use. When water the conceptual model of the hydro- Formation, beneath the shaley inter- falls on the land surface as rain or geologic system. Figure 5 illustrates val, forms the lower bedrock aquifer. snow melt, a portion infiltrates and the updated conceptual model and The Mount Simon aquifer overlies the reaches the water table as recharge. figure 2 compares the hydrostratigra- Precambrian crystalline basement Once part of the groundwater sys- phy of the two models. rock, which is assumed to be imper- tem, water moves horizontally and meable and forms the lower bound- In the 2016 model, the unlithified vertically toward discharge areas ary of the groundwater flow system. materials were divided into two such as streams, lakes, wetlands, units to account for locally confined Vertical hydraulic gradients, where and pumping wells. Water within conditions created by the presence hydraulic heads vary significantly shallow groundwater aquifers, such of shallow, fine-grained glacial and with depth, occur in Dane County. as unlithified deposits or shallow post-glacial lacustrine deposits Upward gradients, and upward flow, bedrock, typically circulates through within the greater Yahara River valley. often occur beneath natural discharge the groundwater system on the scale Outside of the Yahara River valley, areas such as the Wisconsin River, the of years to hundreds of years depend- the unlithified materials are present Yahara River, and major springs in ing on the proximity to discharge across much of the glaciated portion the county. However, near the main features. The closer recharging water of the county and within narrow pumping centers there is often a is to a discharge feature, the shorter alluvial valleys of the Driftless Area. In strong potential for downward ambi- its flow path, and the more quickly, these areas, the distinction between ent borehole flow from the shallow on average, it travels through the the unlithified units is not applica- aquifer system to the Mount Simon. groundwater system and discharges ble and both layers were assigned Numerous flow meter logs in these to the surface. In contrast, water in the the same hydraulic properties. The cross-connecting wells have docu- deeper Mount Simon aquifer typically distinction between these two mented these flows, which can be as takes hundreds to thousands of years unlithified aquifer units was based on high as 100 gpm. to circulate through the groundwater previous hydrogeologic mapping of system unless captured by a well. Water intersecting fracture net- the shallow sand and gravel aquifer works within the sandstones of the system (Fritz, 1996). Tunnel City Group and the Wonewoc The upper bedrock aquifer system Formation moves preferentially along (rock units above the Eau Claire these horizontal features until reach- aquitard) was lumped as a single unit ing a well, flowing into the unlithified in the 1996 model. The current model aquifer, or discharging directly to the subdivides the upper Paleozoic into surface. Many of the major spring eight layers (fig. 2) and accounts for networks in Dane County are located preferential groundwater flow along in areas where the fracture network bedding-plane fractures within the within the sandstone of the Tunnel 16 Wisconsin Geological and Natural History Survey EAST Creek Koshkonong Koshkonong recharge Door Creek high-capacity municipal well Lake Lake Monona recharge potentiometric surfacepotentiometric water table water recharge

groundwater groundwater ow direction ow

k e e r C t n o m r e V

b a

k e e r C s r e v l Conceptual model of the groundwater system shown with and without vertical exaggeration. without and with shown system the groundwater model of Conceptual E Yahara area; elsewhere, till and meltwater stream deposits stream till and meltwater elsewhere, area; Yahara deposits stream Unlithi ed II = till and meltwater Unlithi ed I = ne-grained deposits within glacial Lake lake a b Tunnel City—upper Tunnel layer City— fracture Tunnel City—lower Tunnel Unlithi ed I Unlithi ed II Upper bedrock Jordan Lawrence St. Wonewoc layer fracture Wonewoc— Eau Claire Mount Simon Precambrian 1 2 3 4 5 6 7 8 9 Model layer name Model layer 10 11 12 pro le with no verticalpro le exaggeration: vertical 50x exaggeration: WEST Figure 5. Figure 17 The 2016 Groundwater Flow Model for Dane County, Wisconsin Three-dimensional model construction and simulation of the groundwater flow system 2000) used uniform node dimen- Model grid sions of 1312.4 x 1312.4 ft (roughly Hydrostratigraphic The three-dimensional finite- 40 acres) throughout the entire model units difference groundwater flow model grid. The number of active nodes covers a 50- by 60-mile land area Layering approach and in each model layer varies slightly treatment of pinchouts subdivided into 3,056,020 nodes because the active extent of each (479 rows, 638 columns, and 12 lay- model layer is slightly different. The Dane County numerical model ers). To allow for additional node uses the concept that each model refinement within Dane County, while Model construction layer represents a single hydrostrati- maintaining acceptable run times, the graphic unit. This layering philosophy Each component of the hydrogeo- model grid is divided into near-field is convenient for keeping track of logic system, including hydrostrati- and far-field areas. The near-field area the properties of multiple units, but graphic units (aquifers, fracture layers, encompasses all of Dane County and poses problems where the units pinch and aquitards), boundary conditions has uniform row and column dimen- out due to erosion or nondeposition (streams, lakes, and springs), ground- sions of 360 x 360 ft. The area of a because model layers must be contin- water withdrawals, wastewater single model cell is approximately uous across the model domain. Where treatment plant discharges, and 3 acres. By contrast, the far-field node hydrostratigraphic units pinch out or hydrologic properties, is associated dimensions are non-uniform and are absent, the layer was assigned a with discretized cells that constitute increase in length using a multiplier of nominal thickness of 0.2 ft and given a three-dimensional numerical grid. 1.4, effectively stretching each node vertical hydraulic conductivity (Kv) The following sections describe dimension with distance from the and horizontal hydraulic conductivity how each of these components was near-field boundary. The far-field area (Kh) comparable to adjacent layers. incorporated into the model. Figure 2 extends outward from the near-field When overlying layers are pinched shows the relation of geologic units boundary into the seven neighboring beginning with layer 1, Kv and Kh to model layers. Maps of the extent counties (Columbia, Dodge, Jefferson, of the first unpinched underlying and thickness of each model layer are Rock, Green, Lafayette, and Iowa). layer were assigned. For example, shown in figures 6 and 7, respectively. For comparison, the previous Dane if layer 1 was pinched, Kv and Kh of Figure 8 shows the model layering in County model (Krohelski and others, layer 2 were assigned to layer 1; or if cross section. layers 1 and 2 were pinched, Kv and Kh of layer 3 were assigned to layers 1 and 2. In contrast, if underlying layers are pinched, Kv and Kh of the first unpinched overlying layer were assigned to all underlying pinched cells. For example, if only layer 3 was pinched, Kv and Kh of layer 2 were assigned to layer 3; or if layers 3 and 4 were pinched, Kv and Kh of layer 2 were assigned to layers 3 and 4. These pinchout layers were used, for exam- ple, across the Driftless Area of west- ern Dane County where Quaternary sediments are absent. They were also used below the Madison lakes where the pre-glacial lake basin is eroded down through the younger Fractured dolomite near Fitchburg Paleozoic rocks. Mike Parsen 18 Wisconsin Geological and Natural History Survey

Figure 6. Extents and elevations for key model layer surfaces. Figure 6:

Model layer surface elevation

Layers 1-2 (Unlithi ed materials) Layer 3 (Upper bedrock) Layer 4 (Jordan)

Layer 5 (St. Lawrence) Layers 6-8 (Tunnel City) Layers 9-10 (Wonewoc)

Layer 11 (Eau Claire) Layer 12 top (Mount Simon) Layer 12 bottom (Precambrian surface)

Range in surface elevation by layer* 1479 1479 1500 *Elevation ranges are Layer absent based on the extracted 1200 raster for Dane County Major lakes 968 939 909 only, not the entire 900 808 model domain. 660 652 600 736 614 600 417 558 536 509 300 352 305 Elevation (ft)Elevation 0 Mean sea level -300 0 10 miles -414 -600 1-2 3 4 5 6-8 9-10 11 12 12 top bottom Layer

19 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Figure 7. Model layer thicknesses for key model layers. Figure 7:

Model layer thickness

Layers 1-2 (Unlithi ed materials) Layer 3 (Upper bedrock) Layer 4 (Jordan)

Layer 5 (St. Lawrence) Layers 6-8 (Tunnel City) Layers 9-10 (Wonewoc)

Layer 11 (Eau Claire) Layer 12 (Mount Simon)

Thickness range by layer* 806 800 Layer absent *Thickness ranges are based on the extracted Major lakes raster for Dane County 600 546 only, not the entire model domain. 400 373

203 Thickness (ft) 200 155 0 10 miles 62 62 74 85 0 1-2 3 4 5 6-8 9-10 11 12 Layer

20 Wisconsin Geological and Natural History Survey 1 A 1

B 1,009 ft 1,009 1 C 1 F 1 1 1 B C A 1 F F vertical 15x exaggeration: 1 E E 1 D D 1 E miles 5 0 C A B F 1 D E b a D A Yahara area; elsewhere, till and meltwater stream deposits stream till and meltwater elsewhere, area; Yahara deposits stream Unlithi ed II = till and meltwater Unlithi ed I = ne-grained deposits within glacial Lake lake Tunnel City—upper Tunnel (not shown) layer City— fracture Tunnel City—lower Tunnel Unlithi ed I Unlithi ed II Upper bedrock Jordan Lawrence St. Wonewoc (not shown) layer fracture Wonewoc— Eau Claire Mount Simon Precambrian a b 1 2 3 4 5 6 7 8 9 Model layer name Model layer 10 11 12 B Hydrostratigraphic cross sections across Dane County. Dane sections across cross Hydrostratigraphic C Figure 8. Figure 21 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Unlithified materials corridor, layer 1 is assigned a thick- aquifers were each interpreted as Model layers 1 and 2 represent the ness of 10 ft and assigned distinct distinct hydrostratigraphic units and unlithified materials between land hydraulic properties (described in were identified as model layers 4 and surface and the top of bedrock. the Hydraulic Properties section). 5, respectively. The model treats layers These unlithified sediments were Outside this corridor, the thickness 3 through 5 as convertible because divided into two hydrostratigraphic of layers 1 and 2 are equally divided each of these layers either outcrops units based on geologic mapping and the same hydraulic properties or can be dewatered somewhere in by Clayton and Attig (1997) and Fritz are assigned to both layers. Figure 9 the model domain. Layers 6 through (1996), as well as WCR data. The pres- shows the distribution of materials 12 are simulated as always confined ence of extensive lacustrine deposits, in model layers 1 and 2. The model because they remain saturated associated with the emplacement of simulates both layers as convertible everywhere in the model through Glacial Lake Yahara and Glacial Lake between confined and unconfined all model runs. The sandstone of the Middleton, provided justification for depending on the head in the cell. Tunnel City Group was subdivided into three model layers, layer 6 for subdividing the hydrostratigraphy Bedrock units within the Yahara lakes corridor. The the upper massive section, layer 7 for The upper bedrock units, of thickness of these silt and clay-rich the bedding-plane fracture near the Ordovician age, were combined into lacustrine deposits was estimated middle of the formation, and layer 8 a single hydrostratigraphic unit in from WCR data within central Dane for the lower massive section. Two model layer 3. All units are not found County. For the five principal Yahara layers, 9 and 10, represent sandstone across the county, and insufficient lakes and Fish and Crystal Lakes, of the Wonewoc Formation. Layer information exists to subdivide these lake bathymetry is used as the top 9 represents the upper Wonewoc units hydrostratigraphically. The of layer 1. Where lacustrine deposits Formation, with layer 10 simulating underlying Jordan and St. Lawrence were mapped within the Yahara lakes the bedding-plane fracture at the

Figure 9. Distribution of unlithified materials and near-surface rock in model layers 1 and 2.

Unlithi ed materials and near-surface rock

Fish Lake— high-hydraulic conductivity zone Windblown sand Modern stream sediment Glacial meltwater sediment Lake sediment O shore sediment Subglacial till Hummocky till Near-surface rock

Interstate highways US highways

0 5 miles

22 Wisconsin Geological and Natural History Survey base of the formation. Layer 11 rep- with groundwater is incompletely the vertical hydraulic gradient, the resents the Eau Claire aquitard. The understood. During model calibration vertical hydraulic conductivity (Kv), Mount Simon aquifer is simulated as (described later), the current extrapo- and thickness (b) of the streambed, a single hydrostratigraphic unit (layer lation of Dane County hydrostratigra- as well as the area (A) that the stream 12) below the Eau Claire aquitard phy was not able to simulate the lake occupies within the model cell. There (where present). The Mount Simon levels and steep hydraulic gradients are a total of 11,440 SFR2 nodes in is the lowest hydrostratigrapic unit observed near the lakes. To address the model. The SFR2 stream network in the model, with the Precambrian the error in the conceptual model, includes all the streams shown within crystalline bedrock forming a a high-conductivity zone was created the county in figures 4 and 10. lower no-flow boundary beneath it in model layers 1 through 10 (see A streambed conductance term gov- (Krohelski and others, 2000). fig. 9) and adjusted between the lake erns the exchange of water between area and the Wisconsin River to the surface-water features and the Bedding-plane fractures west. This modification allowed both groundwater system. SFR2 calculates Numerous hydrogeologic investiga- the steep gradient and lake levels to both stream stage and fluxes for each tions (for example, Swanson, 2007; be acceptably simulated. However, individual reach (SFR2 model cell) Gellasch and others, 2013; Sellwood the evidence that there is enhanced allowing for a detailed mass balance and others, 2014) have identified hydraulic conductivity in this area at each model node along the stream. high-conductivity near-horizontal is based on indirect hydrologic fractures associated with bedding data rather than direct geological The surface water network used in planes in the Tunnel City Group and characterization. the groundwater model was based the Wonewoc Formation in Dane on the perennial stream network as County. The model simulates these Boundary conditions mapped by the Dane County Land bedding-plane fracture features and Water Resources Department. Streams as two continuous model layers, High-resolution surface-elevation each 0.1-ft thick. Layer 7 simulates a The SFR2 package (Niswonger and data from a 2009 lidar survey of Dane fracture within the Tunnel City Group, Prudic, 2005) simulates streamflow County was subsequently used to and layer 10 simulates a fracture along perennial streams and through route the perennial stream network. 1 within the sandstone of the Wonewoc minor named lakes which are con- Lidar data was obtained from the Formation. This conceptualization nected to the perennial river system. Dane County Land Information Office. represents a simplification of reality The package simulates each stream A regression expression was used to because there is likely more than one as a number of connected segments calculate stream width for each SFR2 fracture in each of these stratigraphic and tracks the gains and losses to cell based on the total stream length positions, and a single continuous groundwater along each segment. In upstream of that particular cell. The fracture is unlikely to extend across SFR2, a reach is defined as a section regression equation used is Y = 0.0091 the entire county. However, the of a stream associated with a specific * X 0.7103, where X represents stream abundance of field evidence suggests finite-difference model cell, and a length in feet and Y represents stream that higher-conductivity features segment is a group of reaches having width in feet, with R2 = 0.68. This do commonly occur at these strati- uniform or linearly changing prop- regression expression was developed graphic positions. This simulation erties (such as streambed elevation, using stream length data from GIS approach has been successfully used streambed thickness, streambed evaluation of the perennial stream in the past in regional models (Rayne hydraulic conductivity, or stream network and stream width data from and others, 2001). width). Stream segments are linked to form drainage networks, and also a historical study on surface-water High-conductivity zone near link directly with simulated lakes. The features of Dane County (Poff and Fish and Crystal Lakes calculation of groundwater exchange Threinen, 1961). Fish and Crystal Lakes are set in the with each stream node depends on Stream slope was estimated for northwest corner of Dane County 1 SFR2 nodes depending on whether and are characterized by atypical Minor lakes are those lakes which are not modeled using the LAK3 package. the stream was located within the changes in lake levels (Krohelski, The LAK3 package was only used to Driftless Area or the glaciated area 2002). This area of Dane County has model the main Yahara lakes (Mendota, of the model. An average stream few wells for subsurface information, Wingra, Monona, Waubesa, and slope was calculated for both areas and the interaction of these lakes Kegonsa) as well as Fish and Crystal Lakes in northwestern Dane County. and then applied to all perennial 23 The 2016 Groundwater Flow Model for Dane County, Wisconsin

streams within each area. Stream historical flow greater than 0.2 cfs direction of the hydraulic gradient), slope data was obtained from the (approximately 90 gpm or 17,280 cfd) the lakes represented as constant same 1961 report as stream width. were considered for inclusion in the heads are located in the far-field of Streams within the Driftless Area were model and evaluated in the field. the model domain and have little assigned an average slope of 0.0048 Spring Harbor spring, which today has impact on the numerical solution ft/ft (25.3 ft/mile), while streams in a low flow, was included because of within the near-field of the model. glaciated areas were assigned an aver- its higher historical flow. Spring loca- The area of the model most sensitive age slope of 0.0018 ft/ft (9.4 ft/mile). tions were recorded by GPS and new to the specified heads is the near-field flow measurements were collected for boundary in northwest Dane County Lakes springs where possible. Historical flow along the Wisconsin River and Lake The LAK3 package (Merritt and measurements were used for several Wisconsin. During the calibration Konikow, 2000) was used to simulate springs that have been studied in phase it was observed that changes in the major Dane County lakes within greater detail by other researchers. specified heads or hydraulic conduc- the Yahara River watershed (Mendota, Examples include Frederick Springs tivity impacted the water levels in Wingra, Monona, Waubesa, and at Pheasant Branch (Hunt and Steuer, nearby Fish and Crystal Lakes. Kegonsa), and Fish and Crystal Lakes 2000) and Culver Springs at Token in northwestern Dane County (fig. 10). Creek (Parent, 2001). No-flow boundaries The LAK3 package simulates lake No-flow boundaries represent places The model simulates springs using level in terms of the lake bathymetry where groundwater cannot cross the the MODFLOW SFR2 package. Each and the balance of water inflows and model boundary. No-flow boundaries spring included in the model is con- outflows. The water budget accounts were placed along the perimeter of nected to a surface-water feature and for inflows such as precipitation the model near groundwater divides routed as part of the SFR2 stream net- on the lake surface, direct ground- outside of the county along the work. Many of the prominent spring water baseflow, and contributions northern model perimeter between networks in Dane County consist of from tributary streams. The budget Rocky Run and North Branch Crawfish multiple spring boils and ground- also accounts for outflows from the River, along the southern perimeter water seeps, and many spring features lake including evaporation from the between the Rock River and East are distributed among several SFR2 lake surface, losses to groundwater, Branch Pecatonica River, and along nodes. Most major (high-flow) springs streamflow to an outlet, and any the western perimeter between the were connected to underlying aqui- direct diversion of lake water. As an East Branch Pecatonica River and Blue fers by initially setting the vertical example of streamflow routing, the Mounds Creek (fig. 1). The base of the hydraulic conductivity (Kv) of model SFR2 package routes streamflow from model (top of Precambrian crystalline nodes directly below the spring to a the Yahara River into Lake Mendota, rock) is also a no-flow boundary. high value. Allowing flexibility in verti- between Lakes Mendota and Monona cal conductivity under the springs and so forth down the Yahara River was required in order to match mea- Sources and basin. There are a total of 6,697 LAK3 surements of spring discharge. nodes in the model covering over sinks of water 19,000 acres. Smaller named lakes, Specified-head boundaries Recharge such as Cherokee Lake, Indian Lake, Surface-water features along the The largest source of water to the and Stewart Lake, were included as perimeter of the active model domain model is recharge at the land sur- SFR2 rather than LAK3 nodes. are simulated as specified-head face. Recharge was applied to the Springs boundaries. Specified-head nodes uppermost active layer in the model. in the model form a buffer around Initial estimates of recharge for Dane The model simulates 18 prominent Dane County coinciding with the County were based on results from spring systems across Dane County Wisconsin River, Lake Wisconsin, the 2009 Dane County recharge (fig.10). Simulated springs were Rocky Run, Crawfish River, Rock River, model (Hart and others, 2012), which initially selected based on information Lake Koshkonong, Pecatonica River used a soil-water-balance (SWB) from the Wisconsin Springs Inventory East Branch, and Blue Mounds Creek model approach. Results of SWB (Macholl, 2007) and through conver- (fig. 1). Although specified-head models also provided estimates of sations with local water-resources features can serve as sources or sinks recharge for Iowa and Columbia professionals and Dane County land- for groundwater (depending on the Counties. In neighboring Dodge, owners. Springs with an estimated Green, Jefferson, Lafayette, and Rock 24 Wisconsin Geological and Natural History Survey

Figure 10. Location of all steady-state calibration targets. Streams and lakes shown as modeled by the SFR2 and LAK3 packages.

Steady-state calibration: groundwater targets

Vertical head dierence (well) targets " Head (well) targets

Borehole ow targets

Streams and lakes as modeled by the SFR2 and LAK3 packages Interstate highways US highways

Steady-state calibration: surface water targets

Lake and stream level targets Spring ow targets Stream ow targets

Streams and lakes as modeled by the SFR2 and LAK3 packages Interstate highways US highways

0 5 miles

25 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Counties, for which SWB models have over the 5-year period from 2006 to aquitard into the Mount Simon aqui- not been developed, uniform distribu- 2010 to obtain a representative rate. fer. Well withdrawals are represented tions of recharge were used based on Average withdrawal rates were used in the model using the MODFLOW values obtained from areas mod- to avoid outliers that were particularly multi-node well (MNW2) package eled using a SWB model. The initial high during dry years or particularly (Konikow and Harbaugh, 2009). This estimates of the recharge were varied low during wet years. A common package accounts for the exchange piecewise by multipliers across the past well construction practice (still of groundwater between a well and model domain during history match- sometimes used) in Dane County each model layer open to the well. ing. Therefore, calibrated recharge was to drill wells into the Mount This is particularly important in Dane rates maintain the relative recharge Simon Formation but to case the well County given the large number of distribution calculated by Hart and only partway through the overlying high-capacity wells that span multiple others (2012) but have different actual Paleozoic units. This construction aquifers. Model output includes a recharge rates. technique maximized well yields but detailed accounting of node-by-node left an open conduit for groundwater mass balance and hydraulic head for Wastewater treatment discharges to flow vertically across the Eau Claire each multi-node well simulated. Wastewater discharges from 14 waste- water treatment plants corresponding to 15 outlet locations were directly added to the SFR2 package as stream inflow (MMSD accounts for one plant and two outlets). Discharges from wastewater treatment plants were estimated based on actual reported discharge rates from 2010. The Roxbury wastewater treatment plant discharge was not included in the SFR2 package because it discharges to a tributary of the Wisconsin River. The Wisconsin River serves as a con- stant head boundary which is negligi- bly impacted by this discharge. A list of all wastewater treatment plants in Dane County and those included in the model is provided in appendix 2. Groundwater withdrawals The location of all high-capacity withdrawal wells in Dane County are included in figure 4 and a detailed table with corresponding well data is included in appendix 3. Withdrawal rates for each well were averaged

Badger Mill Creek outfall

Madison Metropolitan Sewerage District

26 Wisconsin Geological and Natural History Survey Hydraulic properties of the groundwater flow system Hydraulic conductivity Parameterization of and storage unlithified material Hydraulic conductivity and storage properties properties (confined storativity and Parameterization refers to the specific yield) of the aquifer and assignment of hydraulic properties to aquitard units are fundamental model spatially variable geologic materials, parameters. The model inputs are in and is a key part of model develop- the form of hydraulic conductivity ment. The distribution of unlithified (ft/day), specific storage (1/ft), and materials (sand and gravel, clay, and specific yield (dimensionless). till) in the model is based on the most Initial hydraulic properties for each recent Quaternary mapping in Dane model layer were determined through County (Clayton and Attig, 1997), a review of the numerous reports and and these detailed map units were theses cited elsewhere in this docu- translated into piecewise-constant ment and earlier model information zones of uniform hydraulic conductiv- (Bradbury and others, 1999; Krohelski ity and specific yield in layers 1 and 2. and others, 2000). The areal variation Geologic materials having similar of hydraulic conductivity in the two hydraulic properties were combined. major aquifer units—the Wonewoc Because the thickness of saturated and Mount Simon Formations—was unlithified materials in the county also initially estimated by using is generally small compared to the specific-capacity test data from total thickness of the bedrock aquifer several hundred water-supply system, the hydraulic properties of wells. The TGUESS code (Bradbury layers 1 and 2 were specified identi- and Rothschild, 1985), which treats cally. However, in the vicinity of the specific capacity data as short-term Madison lakes, Fritz (1996) delineated single-well pumping tests, was used an area where well- to moderately to convert these tests to hydraulic well-sorted sand and gravel lies conductivity estimates. beneath clayey lacustrine sediment. In this area, these coarser-grained sed- During model calibration (described iments were incorporated in model in the Model Calibration section) the layer 2 as a piecewise-constant zone. initial hydraulic property estimates Everywhere else, the hydraulic prop- were varied within reasonable ranges erties of layers 1 and 2 were assigned to achieve an acceptable fit between to be the same as the properties measured and simulated water levels of the uppermost bedrock. Table 2 and stream and spring flows. Table 1 and figure 9 show how the various summarizes the layering and hydrau- Quaternary map units were combined lic properties used in the model. and assigned to hydraulic units.

27 The 2016 Groundwater Flow Model for Dane County, Wisconsin 0.15 0.10 0.10 0.15 0.10 0.05 0.05 0.30 0.05 0.10 0.05 0.05 0.05 0.10 0.001 0.001 0.001 0.0001 Same as Estimated n Estimated model layer Sy — — — — — — — 0.3 0.01 0.26 0.11 0.07 0.07 0.015 0.066 0.068 0.073 0.0035 Calibrated -6 -5 -4 -3 -3 -5 -4 -4 -4 -4 -4 -5 -5 -4 -4 -5 -5 -4 Ss 8.3e 2.4e 1.6e 2.6e 1.2e 2.6e 2.1e 9.1e 1.9e 5.0e 7.1e 1.4e 1.1e 2.5e 2.1e 1.0e 5.0e 7.2e (1/day) Same as model layer — — — — — — — Sy (range) Sy 0.1 (0.1–0.4) 0.2 (0.1–0.4) STORAGE PARAMETERS STORAGE 0.1 (0.01–0.3) 0.1 (0.01–0.2) 0.1 (0.01–0.2) 0.1 (0.01–0.4) 0.01 (0.01–0.3) 0.05 (0.01–0.2) 0.01 (0.01–0.3) 0.01 (0.01–0.4) 0.01 (0.01–0.3) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) Initial -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -4 -3 -4 –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e –1e -7 -7 -7 -7 -7 -7 -5 -5 -5 -5 -5 -5 -5 -5 -5 -7 -7 -7 Same as model layer Same as model layer (1/day) (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e (1e Ss (range) -4 -4 -4 -4 -4 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -4 -4 1e 2e 1e 2e 1e 1e 1e 1e 1e 1e 1e 1e 1e 1e 1e 1e 1e 1e )

-2

–3 -3 -3 –5e –5,060) –2,380) 0.2 4.3 0.5 1.0 3.8 3.8 5.0 1.4 0.1 0.5 167 5e 1e -4 25.4 0.01 0.01 150.0 -3 -4 2.8e (ft/day) (0.5–9.3) Kv (range) Kv (4e (2e (5e Calibrated

1 60 8.1 1.0 5.7 5.0 0.5 0.1 0.3 332 20.0 25.4 10.0 13.5 29.3 10.4 uppermost bedrock 0.015 300.0 (ft/day) Layer takes properties of Layer Kh (range) (0.031–237) (0.49–1,000) (0.02–2,680) (0.015–2,330) ) -3

-4 –5e 3 1 — — — 30 20 30 0.1 0.1 5e -5 10,000 (ft/day) HYDRAULIC CONDUCTIVITYHYDRAULIC 5 (3–28) 1 (0.1–2) 1 (0.1–2) (0.01–10) Kv (range) Kv 0.5 (0.1–1) 0.1 (0.1–5) 0.5 (0.1–2) 0.5 (0.1–1) (5e ) Initial -2

-3 -10 –5e 3 1 — — — 12 30 30 5e -4 1e (1–30) 1 (1–3) (ft/day) 5 (3–28) 1.3 (1–5) 1 (0.3–1) 10 (1–20) 50 (1–60) Kh (range) 3.6 (1–10) 1.1 (1–10) (5e 20 (1.6–26) (fracture layer) (fracture (fracture layer) (fracture Kh = horizontal hydraulic conductivity, Kv = vertical hydraulic conductivity, Ss = specific storage, Sy = specific yield, n = effective porosity, — = not assigned. porosity, effective n = Sy = specific yield, Ss = specific storage, conductivity, = vertical hydraulic Kv conductivity, Kh hydraulic = horizontal Hydrostratigraphic unit Hydrostratigraphic Mount Simon and Crystal Lakes, Fish zone cond. high-hydraulic Tunnel City Tunnel City—lower Tunnel Wonewoc Wonewoc Eau Claire Near-surface rock sediment Modern stream Subglacial till Glacial meltwater sediment sediment Offshore Hummocky till Lake sediment sand Windblown “Upper bedrock” (Ordovician) Jordan Lawrence St. City—upper Tunnel Hydraulic properties of hydrostratigraphic units used in the units model. hydrostratigraphic of properties Hydraulic 3 4 5 6 7 8 9 12 10 11 1, 2 1–10 Model layer(s) Abbreviations: Abbreviations: Table 1. Table 28 Wisconsin Geological and Natural History Survey

Clayton’s and Attig’s Quaternary map- In contrast to the layers above, the ping extends only to the boundaries model varies the hydraulic conduc- Parameterization of Dane County. Materials outside tivity in layers 7, 9, 10, 11, and 12 con- of streambed and the county were interpreted from tinuously by location. Such spatially recent studies in Iowa County (Batten variable K fields are more geologi- lakebed properties and Attig, 2010), unpublished maps cally realistic than the single-value Vertical hydraulic conductivity (Kv) of of Columbia County, and regional approach used in the overlying layers streambeds (assuming bed thick- Quaternary maps. and is justified by the availability ness equal to 1 foot) was previously of field data (head measurements, measured in the field at 12 sites in Parameterization of specific capacity tests, packer tests) the upper Yahara and Sugar Rivers as over parts of these units. In addition, well as the Koshkonong, Sixmile, Black bedrock properties these model layers represent the main Earth, Garfoot, and Pheasant Branch Parameterization of hydraulic con- aquifers present in Dane County and Creeks (Bradbury and others, 1999). ductivity (K) and specific storage model output is more sensitive to Streambed Kv from these measure- (Ss) varies by layer in the bedrock their properties than to properties of ments varied from 1.6 ft/day to 37 ft/ units (model layers 3–12); table 1 the upper, less continuous layers. The day, with a mean of 8.1 ft/day. The summarizes these values. The model distribution of hydraulic conductivity MODFLOW stream conceptualization simulates layers 3, 4, 5, 6, and 8 using (both Kh and Kv) in these spatially divides streambed Kv by thickness single values of parameters for the variable layers uses the pilot point to calculate streambed conductance. entire layer (as shown in table 1). This approach. Pilot points represent Both Kv and bed thickness are gen- single-value approach recognizes locations where the parameter of erally poorly known and vary along that there is little or no spatial data interest (K in this case) is either mea- the length of the stream. Assuming (layer-specific head or K measure- sured or estimated. Each pilot point is a 1-foot thickness means that all ments) to guide more detailed param- treated as a variable parameter in the variability in streambed conductance eterization for these stratigraphic calibration process (described below), is ascribed to Kv, and while this is not intervals. Where the formations rep- and an interpolated hydraulic con- strictly true in the field it is an appro- resented by these layers are absent or ductivity field based on these point priate assumption for the purposes of eroded away, the model cells in the values represents the layer hydraulic simplifying the stream network at a eroded layer take on the properties of conductivity. See Doherty and others regional scale. adjacent layers, as described earlier in (2010) for details regarding the pilot The combination of the vertical the treatment of pinchouts (p. 18). point method. hydraulic conductivity of the lakebed sediment and the thickness of these materials in the lakebed regulates the rate of lake-water exchange with Table 2. Conversion between map units defined by Clayton and Attig groundwater. This lumped parameter (1997) and hydrogeologic materials in model layers 1 and 2. See is called leakance, and is defined as table 1 for hydraulic property assignments for these materials. Kv/b, where Kv is the vertical hydraulic conductivity of the lakebed and b is Map units Description the thickness of these materials above gd, ge, h Little or no Quaternary material—use upper bedrock properties an aquifer. Leakance values were sm, sp Modern stream sediment assumed uniform over nearshore and gs, gt, gb, gp Subglacial till offshore zones in each lake, where the su, se, so, sc Glacial meltwater sediment nearshore consisted of a single node width (360 feet) adjacent to the lake og, op, or Offshore sediment shore and the offshore consisted of gh, gk Hummocky till the remainder of the lake area, but wtr Lake sediment varied among lakes. w Windblown sand

29 The 2016 Groundwater Flow Model for Dane County, Wisconsin Model calibration Calibration strategy Steady-state Steady-state calibration targets Measurements from wells and A calibrated model is one that has: model calibration surface-water features were used (1) an acceptable history matching The steady-state calibration was for the steady-state calibration. where the model approximates a generally based on average hydro- Calibration targets from wells con- series of field-measured calibra- geologic conditions between 2006 sisted of water levels (i.e., hydraulic tion targets, and (2) reasonable and 2010. This time period repre- heads, or heads), vertical head dif- parameters used to obtain the sented the most updated data record, ferences, and borehole flows. Target history-matching fit (Anderson included a variety of high-quality data from surface-water features and others, 2015). There were two calibration targets, and reduced included lake levels, spring flows, history-matching steps in the devel- the influence of particularly wet and streamflows (fig. 10). Each target opment of the Dane County model. or dry years. This evaluation of the was classified into one of 15 groups The first, and most extensive, step steady-state period is critical because (summarized in table 3) and assigned calibrated the model to steady-state it defines average equilibrium condi- an observation weight based on the conditions by varying recharge, tions, and served as the starting point general reliability of the measure- hydraulic conductivity, and stream for transient calibration. The objective ment, number of measurements in and lakebed properties. The second of the steady-state history match- that group, and importance for model step was a transient calibration to the ing was to obtain the best match of objectives. Weights are important drought conditions of late May to July time-averaged field observations and for emphasizing (higher weight) or 2012. During the transient calibration to obtain reasonable parameter esti- de-emphasizing (lower weight) a cali- only storage parameters (specific mates from the observation data. bration target during history match- storage and specific yield) were varied ing. When more than one measure- to match observed data. ment was available for a single well, water levels were averaged to deter- mine a single calibration target value.

Table 3. Steady-state calibration target list. Number of Target group Name Description Sources targets Water level head_1 Research wells WGNHS, other researchers 65 head_4 Wisconsin groundwater-level monitoring USGS 16 network wells head_11 Wisconsin surface water network USGS 10 head_19 Prime research wells Univ. of Guelph, WGNHS, 79 other researchers head_23 Remediation and site investigation wells DNR, other sources 181 Vertical-head head_2 Miscellaneous wells Various sources 5 difference head_21 Research wells WGNHS, other researchers 38 head_22 Remediation and site investigation wells DNR, other sources 45 Borehole flow bh_flow Research wells WGNHS 16 Lake level lake_1 Lakes Mendota, Wingra, Monona, Waubesa, USGS 5 and Kegonsa lake_2 Fish and Crystal Lakes USGS 2 Spring flow springs Research measurements WGNHS, other researchers 18 Streamflow flux Research measurements WGNHS, DNR, other researchers 128 flux1 Surface water gaging stations USGS 8 flux2 Surface water partial record stations USGS 66 30 Wisconsin Geological and Natural History Survey Water-level and vertical-head-dif- ference targets include field measurements collected from a variety of sources within the model domain. Many measurements were obtained from ongoing ground- water research projects by WGNHS and USGS staff, UW-Madison faculty and graduate students, research- ers from the University of Guelph (Ontario, Canada), as well as ongoing remediation efforts by local environ- mental consultants, and the City of Madison Engineering Department. A total of 351 water-level targets, 88 vertical-head-difference targets, and 16 borehole-flow targets were included in model calibration. Field measurements obtained by the USGS, Nine Springs WGNHS, or other researchers were generally given the highest weights Mike Parsen due to high level of documentation of measuring point elevation, data of wells across Dane County. The borehole to best represent back- collection, and data archiving. In measurement consists of lowering ground groundwater flow conditions. contrast, measurements collected and raising a probe, such as a spinner In each instance, the measured as part of ongoing groundwater flow meter or heat-pulse flow meter, vertical flow rates across a particular remediation/investigation projects or in the well and correcting for bore- hydrostratigraphic unit were com- other research projects were assigned hole diameter to determine the pared to fluxes simulated by the lower calibration weights because vertical flow of groundwater within MNW2 package for the corresponding these are considered more uncertain. the uncased section of the borehole. model layer. A list of all borehole-flow Water-level targets recorded by the When taken at multiple depths in targets, their corresponding model Wisconsin DNR’s well construction the well, flow measurements provide layers, and associated weight used for report program were not weighted information about the vertical direc- parameter estimation is included in due to their higher level of uncer- tion of flow and the rate of ground- appendix 5. tainty relative to other calibration tar- water flow from the aquifer into the Lake-level targets were included gets. Zero weight means such targets borehole. Sudden jumps in flow rate for each of the principal Yahara were evaluated visually but did not are typical signatures of preferential lakes (Mendota, Wingra, Monona, formally factor into the calculation flow along bedding-plane fractures, Waubesa, and Kegonsa), as well as of a best fit during history matching. while steady changes in flow rate are Fish and Crystal Lakes. These levels A list of all head targets and asso- more characteristic of matrix flow were obtained from historical mea- ciated weights used for parameter through porous media. An example surements made by the USGS and estimation is included in appendix 4. of a borehole-flow log is included in represent lake levels under average the geophysical log in appendix 1. Borehole-flow targets represent a conditions. Zero weight was given Sixteen borehole-flow targets from unique category of calibration target to the Fish and Crystal Lakes targets three different wells were included not included in the calibration of the due to steep observed horizontal in the steady-state calibration. Flow previous Dane County model. Since hydraulic gradients that were difficult measurements for each well were 2000, borehole-flow measurements to simulate with numerical stability; targeted for periods under ambient, have increasingly been collected however, the history matching to non-pumping, conditions within the during geophysical evaluations these zero-weighted targets was still

31 The 2016 Groundwater Flow Model for Dane County, Wisconsin

acceptable (see Results section). A list one-time streamflow measurements considered a highly parameterized of all lake-level targets and associated performed by WGNHS and DNR staff model (Hunt and others, 2007) that weight used for parameter estimation under baseflow conditions during provides more flexibility to model is included in appendix 6. 2010 and 2011, as well as streamflow calibration. Such flexibility allows the averages for USGS gaging stations history-matching process to extract Spring-flow targets were included that were not included in the 2011 more information from the observed for 18 of the principal spring systems Gebert study. While WGNHS mea- data, but requires additional compu- within Dane County. Due to the surements were obtained especially tational power. challenge of accurately simulating for the Dane County groundwater these features in a finite-difference History matching was performed on modeling project, DNR measurements grid, observed flow rates at certain a large parallel computer array at the were conducted as part of a sepa- springs were history matched by com- USGS Wisconsin Water Science Center. rate surface-water modeling study paring simulated flow over several During this phase of the history for Dane County (Diebel and others, SFR2 nodes to combined observed matching, the following parameters 2014). Four estimates of streamflow flow from multiple spring boils. The were adjusted: were included for the outlets of group of SFR2 nodes represented Lakes Mendota, Monona, Waubesa, Recharge: Recharge varies spatially cumulative flow from the entire and Kegonsa. Based on time series across the landscape. The initial spa- spring complex, not just one partic- evaluations of streamflow, the Q75 tial distribution of recharge was taken ular outlet feature such as a spring duration period was selected to rep- from Hart and others (2012), and boil. Spring flow measurements were resent long-term baseflow through then adjusted regionally using one compiled from a variety of sources these lakes.2 While the total discharge multiplier over the glaciated area and including the WGNHS, USGS, DNR, from the Yahara Lakes system is well another multiplier over the ungla- UW-Madison researchers, and histori- represented, the local distribution of ciated areas of the model domain. cal measurements from the Wisconsin flow between lakes is not as well con- This approach maintained the initial Springs Inventory. The WGNHS and strained. These estimates were based distribution of relative recharge rates DNR measurements were made on historical time-series streamflow of Hart and others (2012) within the during 2010 and 2011 while the other data from USGS stream gaging two regions while providing the flexi- measurements were made between stations along the Yahara River, bility needed to obtain a good history 1997 and 2003. A list of all spring flow downstream from Lakes Mendota, match to observed measurements. targets and associated weights used Waubesa, and Kegonsa. Since no for parameter estimation is included Horizontal hydraulic conductivity: stream gaging station is located at the in appendix 7a, and a list of SFR2 During history matching, hydraulic outlet of Lake Monona, streamflow nodes representing flow contribution conductivity was estimated at pilot data near the outlet of Lake Waubesa for each spring complex is included in point locations and then hydraulic was used as a surrogate. A list of all appendix 7b. conductivity values were assigned to streamflow targets and associated individual model cells by interpolation Streamflow targets included USGS weights used for parameter estima- between the pilot points using krig- statistical estimates of baseflow at tion is included in appendix 8. ing. The calibration process adjusted long-term stream and partial-record both horizontal and vertical hydraulic gaging stations, as well as synoptic Steady-state calibration parameters conductivity in each active model one-time streamflow measurements Model calibration was undertaken at layer, using the initial values shown made by the WGNHS and DNR. The the U.S. Geological Survey Wisconsin in table 1. In the unlithified sediment most highly weighted streamflow Water Science Center by using PEST layers 1 and 2, horizontal hydraulic targets were the USGS baseflow (Doherty and others, 2010; Doherty conductivity varied piecewise by estimates at eight long-term gaging and Hunt, 2010). The Dane County material units as shown in figure 9. stations (Gebert and others, 2011). model calibration effort adjusted These two layers were parameter- The second most highly weighted 2,580 parameters during steady-state ized into zones that aligned with the targets were USGS baseflow estimates history matching. Therefore, it is material units. Upper bedrock layers at 66 USGS partial-record stations 3, 4, 5, 6, 8, and 11 were parameter- (Gebert and others, 2011). The lowest 2 The Q75 value for a stream is the ized using single values of hydraulic target weights were assigned to flow rate (Q) which occurs or is conductivity corresponding to each exceeded at least 75% of the time.

32 Wisconsin Geological and Natural History Survey of the Paleozoic bedrock hydrostrati- izontal hydraulic conductivity. Only Simon being a primary aquifer that is graphic units. In the remaining layers field-based estimates from wells with costly to drill; thus, directly measured (7, 9, 10, and 12), a highly parameter- open (uncased) well intervals in either values are likely more important for ized approach was used and hydraulic the Mount Simon or upper Wonewoc citing future wells than indirect values conductivity was represented using aquifer were included. For the inferred during history matching to pilot points (Doherty and others, Wonewoc, TGUESS values of hydraulic sparse head observations. 2010) assigned to the Paleozoic bed- conductivity were used for initial val- Vertical hydraulic conductivity: In a rock in each layer. Pilot points were ues in 337 co-located pilot points but similar manner to horizontal hydraulic used to simulate the Wonewoc and then they, along with 260 other pilot conductivity, initial vertical hydraulic Mount Simon aquifers (layers 9 and points, were allowed to vary during conductivity values (table 1) were 12, respectively), and the Tunnel City history matching. For the Mount adjusted during history matching. In and Wonewoc fracture layers (layers Simon aquifer, TGUESS values from 37 the unconsolidated sediment layers 1 7 and 10, respectively). Horizontal locations were used to set collocated and 2, vertical hydraulic conductivity hydraulic conductivity was estimated pilot point hydraulic conductivity to was varied by the piecewise-constant at over 1,800 pilot points in these a constant value, and history match- zones representing the geologic four layers, which allowed hydrau- ing then varied the remaining 305 units as shown in figure 9. Vertical lic conductivity to vary spatially pilot point values (point distribution hydraulic conductivity in bedrock within the hydrostratigraphic unit. shown in fig. 11). Thus, Mount Simon layers 3 through 10 was parameter- TGUESS measurements were more In the Dane County model, field ized using single values of vertical strictly enforced during the history measurements of hydraulic conduc- hydraulic conductivity applied to the matching to ensure final calibrated tivity were formally considered during entire unit and corresponding to the parameters agreed with measured calibration in wells where specific Paleozoic bedrock hydrostratigraphy. values. The more strict application of capacity data were used (Bradbury Vertical hydraulic conductivity in the TGUESS is consistent with the Mount and Rothschild, 1985) to estimate hor-

Figure 11. Location of layer 12 (Mount Simon aquifer) Kx pilot points used during steady-state model calibration.

Transmissivity, Mount Simon (layer 12)

Pilot points ! Actual measurements ( History matched values

Transmissivity (ft2/day)

500 1000 5000 10000

Interstate highways US highways

0 5 miles

33 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Eau Claire aquitard and Mount Simon To ensure that calibration parameters Steady-state calibration results sandstone are represented using 308 stayed within geologically reasonable The final steady-state calibration and 443 pilot points, respectively. ranges, we used “soft knowledge”, or consisted of a good model fit for most Tikhonov Regularization, to constrain Streambed and lakebed conduc- observations and reasonable values parameter variation (Doherty and tance: Streambed properties were for the 2,580 calibration parame- Hunt, 2010). Soft knowledge was varied using single multipliers for ters. Figure 12 shows the calibrated interjected into the calibration by lim- two regions of the model domain recharge distribution. The average cal- iting the hydraulic conductivity values reflecting the glaciated and unglaci- ibrated recharge rate over the model to be within pre-calibration estimated ated areas. Lakebed leakance values domain is 7.7 in/yr with a range of ranges, limiting the Kh/Kv ratio to be were divided into two zones for 0.0 to 35.4 in/yr, with most less than greater than 1 for every unit, setting each lake using lake bathymetry—a 12 in/yr. For comparison, recharge measured Kh as initial values for shallow water nearshore zone which estimates using stream baseflow for pilot points associated with TGUESS is relatively more transmissive, and selected basins in the county range measurements, and maintaining the remaining lake nodes represent- from 1.1 to 15 in/yr (Gebert and reported well production for pump- ing a less transmissive offshore zone. others, 2011), with estimates from the ing wells. Assisted singular value During history matching, the entire largest basins in the range of 5.8 to decomposition (SVDA) (Doherty and lakebed leakance was adjusted using 9.5 in/yr. Hunt, 2010) was used to ensure that a single multiplier parameter for each the calibration problem was mathe- Cross plots of observed versus lake. This maintained the relative leak- matically solvable (that is, to create simulated hydraulic heads, ver- ance difference within the lake while a well-posed problem from a highly tical head differences, and lake allowing history matching flexibility parameterized ill-posed problem). levels (fig. 13) show good agree- to account for differences in lake ment, with some outliers. hydrogeologic setting.

Figure 12. Calibrated steady-state recharge distribution.

Groundwater recharge

Recharge (inches per year) 0 5 6 7 8 9 10 11 12

Unde ned

Major streams and lake outlines

0 5 miles

34 Wisconsin Geological and Natural History Survey

Figure 13. Steady-state hydraulic head, vertical head difference and lake-level calibration results with calibration summary statistics.

Hydraulic head Vertical head di erence 1,000 70

60 950

t-msl) 50 f ( s

900 c e n e

r 40 e t-msl) i f ( d

s d d

850 a

a 30 e e h

h l

d c a e i t t r a 20 l e u v

800 m d i e S t a l 10 u m i

750 S 0

700 -10 700 750 800 850 900 950 1,000 -10 0 10 20 30 40 50 60 70 Observed heads (ft-msl) Observed vertical head di erences (ft-msl) Research wells (prime) Research wells (other) Remediation wells USGS groundwater network wells Remediation wells Research wells USGS surface water network Other wells

Lake level Calibration summary statistics 880 Head 875 Measurement Head difference Residual mean 3.92 2.58 870 Absolute residual mean 9.26 6.1 Crystal Lake RMS error 15.48 11.64 t-msl)

f 865 (

s Range of observations 287.06 877.89 l e v e

l Fish Lake 860 Number of observations 355 86 e k a l

d e

t 855 a l u m i S 850 Lake Mendota Lake Wingra 845 Lake Monona Lake Waubesa Lake Kegonsa 840 840 845 850 855 860 865 870 875 880 Observed lake levels (ft-msl)

Weighted Unweighted

35 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Histograms (fig. 14) comparing field In figure 14, the calibrated mean well pumping tests (Bradbury and measurements of hydraulic conduc- hydraulic conductivity for the Mount Muldoon, 1990). It was also noted tivity with the spread of node-by- Simon aquifer (layer 12) is about 65 that enhanced hydraulic conductivity node calibration results for layers percent higher than the mean from in the fractures in layers 7 and 10 did 9 and 12 show that the calibration the TGUESS estimates. This apparent not persist beyond the glaciated area. appropriately reproduces the range discrepancy is likely due to the rela- Rather, in the unglaciated areas these and distribution of the field data. tively small number of reliable specific layers retained the properties of the capacity tests in the Mount Simon vertically adjacent unfractured bed- and the clustering of these tests in the rock. The model also generally repro- Madison area. Commonly, hydraulic duces field-measured streamflows conductivity values appropriate for and spring discharges well across the regional models are significantly county (fig. 15). larger than values derived from single

Figure 14. Histograms of hydraulic conductivity for the Wonewoc (layer 9) and Mount Simon (layer 12) aquifers.

Wonewoc (layer 9) Mount Simon (layer 12) 0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2 fraction of values fraction of values

0.1 0.1

0 0 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 Log K, ft/day Log K, ft/day

FIELD DATA Wonewoc Mount Simon Geometric mean (ft/day) 5.6 4.9 Log geometric mean 0.75 0.69 Number of tests 375 37 CALIBRATION DATA Geometric mean (ft/day) 5.7 8.1 Log geometric mean 0.76 0.91

36 Wisconsin Geological and Natural History Survey

Figure 15. Steady-state flow calibration results.

Stream ow 20,000

16,000 ) d 1,000 c f 12,000 ow (

m a e r s t

d 8,000 e t a l u m i S 4,000

0 0 4,000 8,000 12,000 16,000 20,000 Observed stream ow (1,000 cfd) USGS—partial record stations USGS—gaging stations Miscellaneous sources

Spring ow Observed Simulated 700

600

500

400

300 Spring ow (1,000 cfd)

200

100

0 Flynn Garfoot Lower Spring Big Council & Duck Big Small Big Englehart Nevin Nursery Syene Frederick Zeier Rd. Story Culver Creek Springs Garfoot Harbor Springs Dancing Pond Donald Donald Springs Drive Fish Springs Road Springs Springs Creek Springs Springs Springs Spring Sands Springs Park Park Springs Hatchery Springs Spring Springs Springs Springs Springs Flynn Garfoot Creek Lake Lake Wingra Mount Vernon Nine Springs Creek Pheasant Stark- Story Token Creek Mendota Creek Branch weather Creek Creek

37 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Figure 16 shows calibrated trans- Table 4. Simulated lakes and missivity distributions for layer 9 calibrated leakance values. (Wonewoc aquifer); and figure 11 shows distributions for layer 12 Calibrated leakance (Mount Simon aquifer). Model number, Nearshore Offshore After model calibration, streambed lake name (1/day) (1/day) Kv in the calibrated model (again, 1 Mendota 1.3 0.1 assuming a bed thickness equal 2 Wingra 0.8 0.08 to 1 ft) ranged from 2 to 40 ft/day, 3 Monona 0.9 0.09 with an overall average of 7.7 ft/day. 4 Waubesa 2.1 0.2 Streambed Kv was set at 15 ft/day for streams in the glaciated area of Dane 5 Kegonsa 1.0 0.1 County and 2 ft/day in the unglaci- 6 Fish 1.1 1.0 ated area. Table 4 shows the lake sed- 7 Crystal 2.1 2.0 iment leakance values that regulate the rate of lake water exchanged with groundwater for the lakes simulated with the LAK2 package.

Figure 16. Calibrated transmissivity distribution for layer 9 (Wonewoc aquifer).

Transmissivity, Wonewoc (layer 9)

Transmissivity (ft2/day)

500 1000 5000 10000

Interstate highways US highways

0 5 miles

38 Wisconsin Geological and Natural History Survey

water pumping. The figure also shows included in the model domain. The Transient model an expanded view of the transient model simulates changes in hydro- calibration calibration period. During this period, logic stress such as recharge and lake Transient models are needed to regional groundwater levels declined evaporation. by several feet, while total county simulate field data where head or Transient calibration targets flow changes with time. Simulating groundwater pumping increased transient groundwater flow was an from about 63 mgd in May to nearly Transient head and flow targets con- important objective of the modeling 91 mgd in July. History matching sisted of time-series measurements of project as many societal questions focused on transient changes in water levels in wells, lake levels, and cannot be adequately addressed hydraulic head and associated flow in streams (appendix 9). Time using a steady-state model. A groundwater–surface-water exchange series are relatively sparse compared transient model was derived from in response to changes in pumping to steady-state targets in Dane the steady-state model to simulate rates and the absence of recharge. County. Groundwater recharge during drought conditions of May 31 to Because any changes to calibrated the 2012 drought (May 31–July 18) July 18, 2012. As shown in figure 17, hydraulic conductivity values would was assumed to be zero, consistent during this 7-week drought, only only degrade the steady-state calibra- with the period occurring during the two small rainfall events occurred tion results, the transient calibration height of the growing season when and regional groundwater levels focused on only the values of the available soil moisture was likely steadily declined due to an absence of storage terms, specific storage and/or removed by evapotranspiration. recharge and an increase in ground- specific yield, for hydrogeologic units Calibration targets for the transient

Figure 17. Precipitation, water levels, and pumping during the 2012 calibration period.

well #13000064 head Precipitation, water levels, and well #13001355 head pumping during calibration period precipitation 2 875 928 calibration period 874 1.5 927 873 1 926 872

daily precip, in 0.5 925 871 well #13000064 head, feet 0 870 924 well #13001355 head, feet Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

1.5 874 928 calibration period SP4 SP3 91 mgd 88 mgd 1 873 SP2 927 81 mgd SP1 0.5 872 71 mgd 926 daily precip, in 63 mgd PUMPING RATE well #13000064 head, feet 0 871 925 well #13001355 head, feet 17 21 25 29 05 09 13 17 21 25 29 05 09 13 17 21 25 29 Jun Jul 39 The 2016 Groundwater Flow Model for Dane County, Wisconsin

calibration consisted of water-level Transient stress periods account for over 70 percent of total measurements at 25 wells and four The transient calibration run used pumping within the model domain lakes, and flow records at 14 streams the calibrated steady-state model as and are assumed to represent general (fig. 18). At each site, continuous the starting condition. Recharge was water demand trends. time-series records were converted to set to zero, as was precipitation on Although operations varied from period mean and ranges (lake levels) lake surfaces. Lake evaporation rates well to well, analysis of the pumping and daily mean and daily difference were initially set to typical summer records showed that the pumping (that is, the difference between values of 0.15 to 0.23 inches per day. generally fell into four stress periods, adjacent daily values) for comparison To estimate groundwater withdrawal specifically May 31–June 13 (14 days), with model results. Temporal differ- rates during the transient calibration June 14–27 (14 days), June 28–July encing of observed data is needed to runs, we sent a water-use survey to 8 (11 days), and July 9–18 (10 days), ensure the history matching focuses major public water utilities in the for a total of 49 days. During each on dynamics of the groundwater fall of 2012. These data were then of these stress periods, simulated system and does not simply reflect extrapolated to other wells for which pumping for wells with records was the parameter’s need to overcome no withdrawal data was obtained. set to actual reported rates when shortcomings in the steady-state Withdrawal data were obtained from possible; pumping rates for wells model (Doherty and Hunt, 2010). 11 water utilities in Dane County: lacking records was estimated to Observation weight varied by target DeForest, Fitchburg, Madison, increase proportionally to the rates type and length of observed time McFarland, Middleton, Mount Horeb, in the wells having records. For the series; transient observed data and Oregon, Stoughton, Sun Prairie, calibration runs shown in figure 17, simulated results were pre- and Verona, and Waunakee. These records stress periods (SP1–4) were divided post-processed using a time-series represent approximately 12 percent into time steps equal to the number processor (TSPROC) (Westenbroek of the high-capacity wells within the of days in each period (14, 14, 11, and and others, 2012)). model domain. Although this is a rel- 10 days, respectively); the time step atively small percentage of wells, they length was increased from 0.5 days at

Figure 18. Locations of all transient calibration targets.

Transient calibration targets

" " !

Head (well) targets "" ! Stream ow targets ! ! Lake level targets " " " " ! " ! " ! ! " " " "

Interstate highways ! US highways "

Major streams ! and lakes ! "

!

! 0 5 miles

40 Wisconsin Geological and Natural History Survey the initial step in each stress period to example) suggests that the transient 1.5 days in the final step to facilitate model stress regime did not accu- numerical convergence. rately reflect actual stresses such as local pumping changes. Likewise, Transient calibration parameters predicted baseflows and lake levels Specific storage, specific yield (for (fig. 20) are in close agreement for convertible layers that can change some targets (such as Lake Monona) from confined to unconfined), while the match is rather poor for and lake surface evaporation were others (such as Black Earth Creek). adjusted during transient history Lake levels are particularly difficult matching while holding all other to match because the measured lake model parameters constant. These level responds to multiple factors storage parameters were zoned other than groundwater interaction, according to hydrostratigraphic unit, such as wind, dam operation, and and varied uniformly within an indi- weed growth in the Yahara River— vidual unit. Initial and final calibration processes that are not included in values are shown in table 1. A total of the transient groundwater model. 33 parameters, including lake evapo- Therefore, while the model can ration for the four stress periods and reliably represent groundwater–lake 29 storage parameters, were adjusted interactions, it is not intended for use during transient calibration. as a transient lake-level model. Transient calibration results The calibrated values for specific The overall transient history match- storage (table 1) for the sandstone ing was deemed acceptable, though aquifer layers (layers 9 and 12) are -5 was not as well matched as the in the range of 10 1/ft, which is steady-state calibration. A lesser within the range of the expected fit is expected because the actual specific storage for sandstone (Li magnitude and time of important and Bradbury, unpub. data), and the changes to stress, such as pumping observation that some (but not all) of rate, were only approximately known. the simulated transient head declines In addition, fewer parameters were shown in figure 19 are less steep than allowed to vary during the history observed declines suggests that, at matching. The transient calibration least locally, the actual specific stor- must trade-off fit between multiple age might be less than the calibrated observation types (for example, values. The final transient calibration between fitting predicted stream- is a compromise fit among all model flows and predicted water levels) targets and viewed as acceptable for as well as possible across the entire the purposes of this model and for county. Figures 19 and 20 (heads the relatively short calibration period and flows, respectively) show typical chosen. Application of the model to comparisons between measured and problems where transient results are simulated targets. For water levels in critical will likely require additional wells (fig. 19), it is important to note evaluation of storage parameters. that matching the slopes of the head decline was the objective of transient calibration rather than matching the head values themselves. For some wells, such as Savannah Valley and Madison city well #21, the match is considered close, while a poor match in other wells (Pheasant Branch, for 41 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Figure 19. Transient calibration results for selected well targets. WELL HEADS Madison at W. Washington Ave. (13001297) Madison city well #21 (13000064) 850 880 observed simulated 878 849

876 848 head, ft head, ft 874

847 872

846 870 0 10 20 30 40 50 0 10 20 30 40 50 observation day observation day

Madison sentry well #29, port 2 (13006008-2) Madison sentry well #29, port 4 (13006008-4) 860 830

825

856 820

815 head, ft head, ft 852 810

805

848 800 0 10 20 30 40 50 0 10 20 30 40 50 observation day observation day

Pheasant Branch (13001440) Savannah Valley (13006104) 864 896

863 895

862 894 head, ft head, ft

861 893

860 892 0 10 20 30 40 50 0 10 20 30 40 50 observation day observation day 42 FIGURE 19 Wisconsin Geological and Natural History Survey

Figure 20. Transient calibration results for selected flow and lake-level targets.

STREAMS Pheasant Branch (at Hwy M) Black Earth Creek (near Cross Plains) 8 11 observed simulated 10 6

9 4 8 discharge, CFS discharge, CFS 2 7

0 6 0 10 20 30 40 50 0 10 20 30 40 50 observation day observation day

Yahara River (at Stoughton) Yahara River (at McFarland) 160 160

120 120

80 80 discharge, CFS 40 discharge, CFS 40

0 0 0 10 20 30 40 50 0 10 20 30 40 50 observation day observation day LAKES Lake Mendota Lake Monona 850 845

849.8 844.8

849.6 844.6

849.4 844.4 lake stage, ft lake stage, ft 849.2 844.2

849 844 0 10 20 30 40 50 0 10 20 30 40 50 observation day observation day 43 FIGURE 20 The 2016 Groundwater Flow Model for Dane County, Wisconsin Model results Figure 22 shows simulated Model solution Comparison of 2010 steady-state conditions during the and numerical and predevelopment 2010 calibration run. This simulation includes all high-capacity wells pump- mass balance simulations ing at 2010 average rates, with the Using the Newton solver, the cali- Water levels and drawdown rates shown by the relative diameters brated steady-state model converges Comparison of simulated steady-state of the circles on the figure (see appen- in about 3 minutes on a 64-bit com- predevelopment conditions to the dix 3 for a list of wells and pumping puter using an Intel Core i7 2.8GHz 2010 calibration shows how ground- rates). The water table is very similar processor. The overall mass-balance water conditions in Dane County to the predevelopment water table error for the solution is less than have changed over the past century. (fig. 21 - top) except for drawdown 0.02 percent, using a head-change Predevelopment here refers to condi- near the Madison area (discussed criterion of 0.01 (steady-state) and tions prior to the use of high-capacity later). The Mount Simon potentio- 0.3 (transient) ft and a global flux wells in the county. This simulation metric surface shows inflection or criterion of 20,000 (steady-state) and is identical to the 2010 steady-state bull’s-eyes around the pumping wells 9,000 (transient) cfd. calibration simulation except that all in the Madison area, showing the An overall mass-balance summary wells were removed and the wastewa- impact of deep pumping in lowering (table 5) shows inputs and outputs ter treatment discharges to selected heads in the deep aquifer. of water to the model. Lakes and streams were eliminated. It assumes Present-day pumping has caused sig- streams can act as both sources recharge equal to present-day condi- nificant drawdown near the Madison and sinks of water depending on tions. More detailed hindcast simu- metropolitan area, in the center of the their location relative to pump- lation of changes in the hydrology of county. Figure 23 shows drawdown ing wells and their position in the county (such as wetland drainage, for the water table and Mount Simon the landscape. Cross-connecting dam construction, and land-use potentiometric surface (calculated as wells can also be a water source, change) is possible but was beyond the difference between predevelop- transferring a net amount of over the scope of this project. ment and 2010 conditions). Simulated 4 cfs between model layers. Figure 21 shows the simulated water water-table declines (fig. 23 - top) table and Mount Simon potentio- range from less than 5 ft to more metric surface for predevelopment than 40 ft in the Madison area, but Table 5. Steady-state mass conditions. The water table has the Madison lakes and Yahara River balance summary. significant complexity, particularly in maintain the water table where they Inflow Outflow the western Driftless Area, because occur, so there is negligible decline Boundary (cfs) (cfs) the numerous surface-water features at the lakeshore or below the lakes. More significant declines occur in the Recharge 1,090.2 0.0 are well-connected to groundwater. Lows in the Mount Simon potenti- Mount Simon potentiometric surface Specified head 10.3 150.4 ometric surface reflect topographic (fig. 23 - bottom) which simulations Lake 11.2 19.9 lows, and this potentiometric surface show has declined more than 70 ft in Well 6.4 97.5 is much smoother than the water parts of Madison, including a nearly Stream 20.4 871.6 table because the Mount Simon is not 50-ft decline below the Madison lakes. Other smaller cones of depres- Storage 0.0 0.0 directly connected to surface-water features. The simulation clearly shows sion occur in outlying cities such as TOTAL 1,138.5 1,139.4 the divides between the Yahara River Verona, Stoughton, and Sun Prairie; basin and the Wisconsin River and little or no drawdown occurs near the Sugar-Pecatonica River basins to the county boundaries. These overall pat- west, represented as highs on the pre- terns are consistent with simulations development water table and potenti- of the earlier Dane County model of ometric surface contour maps. Krohelski and others (2000).

44 Wisconsin Geological and Natural History Survey

Figure 21. Simulated steady-state results under predevelopment conditions (no pumping).

Water table, predevelopment 900

950 Elevation (feet above sea level) 850 900 1020 ft 750 max: 1020 ft

contour 800 interval 25 ft

min: 720 ft 900 720 ft 1000 850

Major streams and lake outlines 950 850

950 900 850

900

Potentiometric surface, Mount Simon (layer 12), predevelopment 950

Elevation (feet above sea level) 900 1020 ft max: 986 ft 800 contour interval 25 ft

min: 744 ft 900 720 ft 850

Major streams and lake outlines 850 850 950

850

900

0 5 miles

45 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Figure 22. Simulated steady-state results under 2010 conditions (pumping).

Water table, 2010 950

Well pumping rate (mgd) 900 <0.1 750 0.1 - 0.5 800 0.5 - 1.0 850 850 >1.0

900 Elevation (feet above sea level) 950 1020 ft max: 1000 ft

contour 850 interval 25 ft 950 900 950 min: 720 ft 720 ft 850 Major streams and lake outlines 900

Potentiometric surface, Mount Simon (layer 12), 2010 950 Well pumping rate (mgd) 750 <0.1 800 0.1 - 0.5 900 850 0.5 - 1.0 800 >1.0

Elevation 775 (feet above sea level) 850 900

1020 ft 825 max: 985 ft 775 contour 950 interval 25 ft

min: 730 ft 720 ft

850

Major streams 900 and lake outlines

0 5 miles

46 Wisconsin Geological and Natural History Survey

Figure 23. Simulated steady-state drawdown, comparing predevelopment and 2010 conditions.

Drawdown in water table

5 ft of drawdown 5 Drawdown, in 10 ft intervals

10 10 Well pumping rate 10 (mgd) 30 <0.1 40 20 0.1 - 0.5 5 0.5 - 1.0 20 30 >1.0

5

Major streams 10 and lakes 5 5

Drawdown in potentiometric surface, Mount Simon (layer 12)

5 ft of drawdown Drawdown, in 10 ft intervals 5 10 80 Well pumping rate (mgd) 30 90 <0.1 120 40 0.1 - 0.5 40 0.5 - 1.0 30 130 >1.0 80 20

Major streams 10 and lakes 5 60

5

0 5 miles

47 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Mass balance and flow to streams. With the addition of aquifers into wells) also occurs, and to surface water pumping from wells totaling 52 mgd this flow is lumped together with the The model quantifies the overall countywide, recharge remains the pumping outflow (52 mgd) to yield mass balance of groundwater and largest source and streams the largest the total well outflow (54 mgd) shown how that mass balance has changed sink, but groundwater discharge to in figure 24. streams decreases by over 40 mgd, from predevelopment to the present. The impact of pumping on individual and discharge to lakes declines from Figure 24 compares predevelopment streams varies according to proximity 19 mgd to 12 mgd. Figure 24 shows to 2010 conditions for Dane County, to the drawdown and stream size. 2 mgd of inflow from wells in 2010. using pie charts to illustrate the CARPC requested a compilation of This 2010 inflow arises from ambient changes to different components of streamflow impacts from pumping borehole inflow (flow from wells into the water balance. For both condi- at specific locations relevant for aquifers) in wells open to more than tions the largest source of ground- regional planning decision-making one aquifer. An equal amount of water is recharge, and the largest (M. Kakuska, CARPC, personal com- sink of groundwater is discharge ambient borehole outflow (flow from

Figure 24. Steady-state water balance, comparing predevelopment and 2010 conditions.

PREDEVELOPMENT 2010

Lakes Lakes Streams (6 mgd) (6 mgd) Boundary ow (5 mgd) Boundary ow (18 mgd) (19 mgd) Wells Streams (2 mgd) (4 mgd) INFLOW

Recharge Recharge (455 mgd) (455 mgd)

Boundary ow Wells (41 mgd) Constant head (54 mgd) Constant head (18 mgd) (19 mgd) Boundary ow (43 mgd) Lakes (19 mgd)

Lakes (12 mgd) OUTFLOW Streams Streams (402 mgd) (361 mgd)

48 Wisconsin Geological and Natural History Survey munication). Table 6 summarizes the Badfish Creek receives about 42 mgd predevelopment to 2010. Declines simulated changes in streamflow at (64 cfs) of treated effluent from the in spring flow from predevelopment 23 locations. Declines in streamflow Nine Springs Wastewater Treatment to 2010, range from as little as 0.01 from predevelopment to 2010 range plant (see appendix 2). As a result, its mgd to 1.54 mgd (0.02 cfs to 2.38 cfs). from 0.31 mgd to 33.24 mgd (0.51 cfs flow in 2010 was significantly greater Overall, the percentages of total simu- to 51.43 cfs). In terms of percentages than predevelopment flow. Note, lated spring flow, range from declines of total simulated streamflow, the though, that changes to streamflow of less than 2 percent to 100 percent percent declines range from less than between the two periods refer only to (completely dry). 3 percent to over 75 percent. For baseflow, not total streamflow. Figure 25 depicts the change in flow several of the streams, the simulated Similarly, the impacts of pumping on for the entire streamflow routing flows increased from predevelopment individual spring flows varies across (SFR2) network between predevelop- to 2010 due to contributions from Dane County. Table 7 summarizes the ment and 2010 conditions. The largest wastewater treatment plants. Badfish changes in spring flow at 18 spring percent decreases in flow occur in Creek is a prime example (table 6). locations simulated in the model from streams located within central Dane

Table 6. Simulated predevelopment and 2010 streamflows at selected sites in Dane County. Simulated Simulated Measurement location predev. 2010 flow Difference Difference Difference Stream/river (at or near area listed) flow (cfs) (cfs) (cfs) (mgd) (%) Badfish Creek CTH-A 11.58 75.47* 63.89 41.30 551.7 Badger Mill Creek STH-69 3.63 4.21* 0.58 0.38 16.1 Black Earth Creek Stagecoach Rd. near Cross Plains 4.94 3.52 –1.42 –0.92 –28.8 Black Earth 33.24 31.28* –1.96 –1.27 –5.9 Dorn Creek CTH-M 6.27 5.65 –0.62 –0.40 –9.9 Koshkonong Creek Deerfield 27.35 29.79* 2.43 1.57 8.9 Hoopen Rd. near Rockdale 62.52 64.66* 2.14 1.38 3.4 Bailey Rd. near Sun Prairie 0.77 5.02* 4.25 2.75 549.8 Maunesha River Greenway Rd. south of US-151 17.25 16.43 –0.82 –0.53 –4.8 Mount Vernon Creek STH-92 19.15 18.49 –0.67 –0.43 –3.5 Nine Springs Creek US-14 12.17 6.87 –5.30 –3.42 –43.6 Pheasant Branch Parmenter St, Middleton 2.85 1.19 –1.66 –1.07 –58.3 Sixmile Creek North Madison St, Waunakee 5.96 5.11 –0.85 –0.55 –14.3 Spring Creek Lodi 22.44 21.92 –0.51 –0.33 –2.3 Starkweather Creek East Branch—at Milwaukee St. 3.01 0.73 –2.28 –1.47 –75.7 West Branch—at Milwaukee St. 8.85 4.16 –4.69 –3.03 –53.0 Sugar River above Badger Mill Creek near 21.16 18.60 –2.55 –1.65 –12.1 Riverside Rd. West Branch—at STH-92 near 18.96 19.19* 0.24 0.15 1.2 Mt. Vernon Token Creek US-51 20.34 17.98 –2.36 –1.53 –11.6 Wingra Creek Beld St., Madison 3.77 1.82 –1.95 –1.26 –51.6 Yahara River Lake Windsor Country Club 6.77 6.28 –0.49 –0.31 –7.2 Stoughton 207.94 156.51* –51.43 –33.24 –24.7 Lake Waubesa outlet at Stoughton Rd. 157.60 109.04 –48.57 –31.39 –30.8 Abbreviations: cfs = cubic feet per second, mgd = million gallons per day * Includes discharge from wastewater treatment plants; see appendix 2 for outfall locations and volumes. 49 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Table 7. Simulated predevelopment and 2010 spring flows at principal springs in Dane County. Simulated Simulated predev. 2010 flow Difference Difference Difference Stream/lake Spring name flow (cfs) (cfs) (cfs) (mgd) (%) Flynn Creek Flynn Creek Springs 2.41 2.33 –0.08 –0.05 –3.3 Garfoot Creek Garfoot Springs 0.95 0.88 –0.06 –0.04 –6.8 Lower Garfoot Springs 0.85 0.62 –0.22 –0.14 –26.3 Lake Mendota Spring Harbor Spring 0.11 0.00 –0.11 –0.07 –100.0 Lake Wingra Big Springs 0.57 0.34 –0.23 –0.15 –40.5 Council Circle & 0.18 0.06 –0.12 –0.08 –66.8 Dancing Sands Springs Duck Pond Springs 0.81 0.48 –0.34 –0.22 –41.6 Mount Vernon Creek Big Donald Park Springs 7.69 7.54 –0.15 –0.10 –1.9 Small Donald Park Springs 1.50 1.48 –0.02 –0.01 –1.5 Nine Springs Creek Big Springs 1.01 0.68 –0.34 –0.22 –33.5 Englehart Drive Springs 0.35 0.13 –0.23 –0.15 –64.2 Nevin Fish Hatchery Springs 5.63 3.25 –2.38 –1.54 –42.2 Nursery Springs 1.68 1.19 –0.49 –0.32 –29.2 Syene Road Springs 0.60 0.25 –0.35 –0.22 –58.1 Pheasant Branch Frederick Springs 4.14 3.48 –0.66 –0.43 –16.0 Starkweather Creek Zeier Road Springs 1.09 0.47 –0.62 –0.40 –57.0 Story Creek Story Spring 5.48 4.88 –0.61 –0.39 –11.0 Token Creek Culver Springs 8.91 7.97 –0.94 –0.61 –10.6 Abbreviations: cfs = cubic feet per second, mgd = million gallons per day

County, which are coincident with the across Dane County. The most notice- which discharges approximately regional cone of depression. In con- able gains in flow are associated with 4 cfs. The hypothetical well is located trast, the smallest percent decreases effluent discharges to Badger Mill 3,000 ft east of the creek and spring in flow are concentrated within west- Creek at Verona, Badfish Creek north complex (fig. 26), simulated as cased ern and eastern Dane County, outside of Oregon, the West Branch of the through the Eau Claire aquitard, of the main groundwater withdrawal Sugar River south of Mount Horeb, and open to the entire thickness of centers. Stream reaches simulated as and the Koshkonong Creek south of the Mount Simon sandstone. The dry (zero flow) during both prede- Sun Prairie. model simulates the pumping well velopment and 2010 conditions are at 1,200 gpm, a typical rate for large areas which were included in the Demonstration of high-capacity wells in southern model as streams but did not receive Wisconsin. The model simulates groundwater discharge during either transient model drawdown and spring flow prior to simulation period. These streams response pumping, after 30 days of pumping, are concentrated in headwater areas One potential use of the tran- and at steady-state. The model also and represent ephemeral streams sient model is for evaluation of simulates the transient response of that flow during wet periods of the the time-variant impacts of new the well and the spring during and year but often become dry during high-capacity wells. To demonstrate after a 30-day pumping period. baseflow conditions. Streams that this capability, we added a hypothet- gained flow between predevelop- ical new well in the vicinity of Story ment and 2010 conditions are those Creek in the Town of Oregon in the that currently receive treated effluent southern part of Dane County. Story from wastewater treatment plants Creek begins at a spring complex, 50 Wisconsin Geological and Natural History Survey

Figure 25. Comparison of changes in streamflow between predevelopment and 2010 conditions.

Change in stream ow, predevelopment to 2010 # M au n e Change in stream ow (%) s h a

R - 95 to -100% . . R

a r a # h -75 to -95% a # Y B # la -50 to -75% ck Ea rth C#r.

decrease -25 to -50%

-5 to -25% # 0 to -5% Koshkonong Creek S ug + 0 to 100% a r # # R iv e r

increase > + 100% # # # # #* Wastewater B discharge locations ad fi sh # C W r Major lakes . # es t Br an Ya ch ha Sug r ar R. a R 0 5 miles # . #

Figure 26. Transient drawdown simulation for a hypothetical well near Story Creek in southern Dane County.

Hypothetical well near Story Creek 6 8

Spring Storytown Road 6 Transient drawdown in aquifer after 30 days, in 2 ft intervals

S

t

o 12

r

y

C

r

e

e

k 10

20 Well

18 22

16 County Hwy A 14 Imagery from the National Agriculture Imagery Program (NAIP) 2010

0 500 1000 feet

51 The 2016 Groundwater Flow Model for Dane County, Wisconsin

The simulated hydrograph (fig. 27) Figure 27. Detailed results for the hypothetical well showing shows that following the initiation drawdown for the creek, spring, and pumped well. of pumping, the water level in the Simulated ow pumped well falls nearly 25 ft and 5 continues to decline slowly through- out the 30-day pumping period. Story Creek (at Hwy A) 4.5 Following the cessation of pumping steady state after 30 days, the water level in the pumped well rebounds rapidly and 4 Story Spring continues to rise slowly throughout

the 40-day recovery period (fig. 27). simulated ow, cfs 3.5 steady state Even though the spring complex is over 3,000 ft from the well and in the 3 surficial aquifer, its flow declines by 0 10 20 30 40 50 60 70 about 0.25 cfs during the pumping Simulated drawdown period and recovers once the well is 0 turned off. It is interesting to note that Pumped well the decline and recovery of spring flow happens more slowly than the 10 decline and recovery in the pumping well, a result which is consistent with hydrogeologic theory. 20 simulated drawdown, feet steady state 30 0 10 20 30 40 50 60 70 time, days Model limitations simulates flow between ground- in the model is 360 ft, and cannot Overview water and surface water, the model capture geologic complexity, such as The current model is a state-of-the-art is not a surface-water model, and facies changes, erosional channels, representation of the ground- can only simulate the groundwater sand and gravel lenses, silt layers, frac- water system and groundwater– component of issues such as flood- tures, and other features that occur surface-water interactions in Dane ing, stormwater runoff, and stream at smaller areal dimensions. Similarly, County at the county or regional management. This model provides a the vertical discretization of the scale. The model is intended to inform regional foundation for smaller-scale groundwater system into 12 discrete questions at the regional scale and is studies, however, and can be used as layers required significant lumping of very well suited to problems involving a starting point for refined study and hydrogeologic properties and does regional pumping, water balance, and simulation of site-specific problems. not represent features that exist at groundwater–surface-water interac- The following sections summarize finer scales. tions. The model is not intended for model limitations. site-scale questions and problems This model contains two layers (layers that involve small stresses to the Limitations related 6 and 10) intended to represent regional system (such as the effects bedding-plane fractures that are of additional pumping from a new to discretization present across the entire county. The private well), where the prediction The hydrogeologic system in Dane existence of such features is now of interest depends on fine-scale County has been significantly simpli- well documented (Swanson, 2001, details that are not well represented fied during model discretization. The 2007; Swanson and others, 2006; in the regional model. Although it smallest horizontal grid dimension Gellasch and others, 2013), but the field appearance of these fractures is 52 Wisconsin Geological and Natural History Survey of numerous horizontal features over properly specified, the model can the county because the density of a limited stratigraphic zone rather become unstable, require very long calibration targets is spatially uneven. than a single continuous feature. run times, or fail to converge. This Parameters that are not sensitive to Although such detail is not important problem is especially troublesome the calibration data, and those that for groundwater flow at the county for transient simulations. Therefore, take on surrogate values to compen- scale, the model’s omission of detail modifications to the base steady-state sate for other model deficiencies add in these fracture features at the site and transient model stresses and/or further model uncertainty. scale might lead to inaccurate model parameters may require the user to The hydrogeologic data, especially predictions, particularly with respect explore a range of solver settings to pumping tests and specific capacity to transport of contaminants. obtain acceptable results. estimates of hydraulic conductivity, The discretization of time rep- are most abundant in populated resents another model limitation. Limitations related to areas such as the Madison metro- Steady-state simulations assume a lack of hydrogeologic politan area, and much less abun- that hydrologic conditions, including dant in outlying rural areas where pumping and recharge, are constant knowledge water-supply wells are scarce. In over time. Pumping and recharge are Fish and Crystal Lakes addition, the hydraulic data are both obviously transient phenomena, The model was unable to satisfactorily heavily biased toward the most-used and so the results of steady-state simulate the stages and fluctuations aquifers of the Wonewoc and Mount simulations represent long-term of Fish and Crystal Lakes, in north- Simon Formations. Parameters in averages. The transient model is suit- west Dane County, using stratigraphy areas of sparse calibration targets able for time-dependent problems; extrapolated from farther south in are relatively more uncertain. If, as however, the transient calibration the county, and sparse stratigraphic intended, this model is used as a basis documented here reflects a relatively information near the lakes. Successful for creation of more detailed inset simple drought scenario and a limited calibration of these lakes required models, additional data collection observed data set for history match- insertion of a high-hydraulic conduc- and re-calibration will usually be ing. Therefore, the model’s ability tivity zone extending from the lakes necessary. to simulate more complex transient to the Wisconsin River in model layers stresses and responses in areas with- Streamflow and lake level 1–10. Therefore, the model’s simula- along the Yahara River out observed transient data will vary, tion capabilities in the area of these both in space and with complexity of two lakes are uncertain. Additional Due to the low hydraulic gradient the specified transient stress. hydrostratigraphic characterization and management of lakes along the and subsequent modeling is needed Yahara River, from Lake Mendota to Limitations related to more accurately resolve the local Lake Kegonsa, it was not feasible to to model stability groundwater flow system and its develop rating curves that could accu- interaction with these lakes. rately attribute flow rates to specific The Dane County model features a lake-level elevations. The combination number of mathematically nonlinear Parameter uncertainty of backwater effects between the processes such as groundwater-lake All models are simplifications of an lakes and engineering management interaction, surface-water routing and unknowably complex natural system. decisions to maintain lake levels groundwater exchange, and wetting As such, there can be no expectation above winter minimums and sum- and drying of model cells. Although of a model forecast without some mer maximums (for example, raising the Newton solver offers improved uncertainty (Hunt and Zheng, 2012). and lowering dam outlet elevations, handling of these nonlinearities Uncertainty in model parameters is aquatic plant cutting practices), con- (Hunt and Feinstein, 2012), certain one well-recognized source of fore- tributed several variables that could reasonable combinations of model cast uncertainty. Even after calibra- not be adequately resolved at the parameters can cause the model tion, hydrogeologic parameters can temporal and spatial resolution used solver settings used for the calibration only be approximately known, and in this modeling effort. documented here to be non-optimal. the level of uncertainty varies across When the solver settings are not

53 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Wetland drainage From predevelopment to the present Summary day, Dane County has experienced he 2016 groundwater flow flows, lake levels, and borehole flows. significant alteration of its landscape. model for Dane County, Steady-state calibration focused on One very significant hydrologic TWisconsin, exemplifies the 5-year period of 2006 through alteration has been the drainage state-of-the-art model construc- 2010; the transient calibration focused of wetlands and the construction tion and calibration. The model was on the 7-week drought period from of networks of agricultural drain- developed by a team of WGNHS and late May through July 2012. age ditches and field tiles. These USGS scientists with support from This model represents superior constructed features route water the Capital Area Regional Planning simulation capabilities of the ground- directly to streams and away from the Commission, and replaces an earlier water system over previous work groundwater system, with the effect model developed in the 1990s. The because of its finer grid resolution, of decreasing groundwater recharge new model is three-dimensional and improved representation of hydro- where they occur. An offsetting effect transient, and conceptualizes the stratigraphy, transient capabilities, is that draining wetlands has reduced county’s hydrogeology as a 12-layer and focus on sophisticated handling evapotranspiration losses. Accounting system including major hydro- of surface-water features. for drainage ditches and field tiles stratigraphic units and two known was considered beyond the scope of high-conductivity fracture zones. As a tool for sustainably manag- the current model and not explicitly ing Dane County's groundwater The model uses the USGS simulated in the model. They were, resources, the model has many MODFLOW-NWT finite-difference however, indirectly accounted for potential applications. These can code, with a Newton solver to by the recharge multiplier, which include evaluation of potential sites improve the handling of unconfined constrains the magnitude of recharge for and impacts of new high-capacity conditions by more robustly handling to achieve a “best fit” to the baseflow wells, development of wellhead pro- the transition between wet and dry stream targets, as well as SWB, which tection plans, evaluating the effects cells. The model explicitly simulates uses present-day land-cover condi- of changing land use and climate on groundwater–surface-water interac- tions. Omitting the dynamics of these groundwater, and quantifying the tions with streamflow routing and hydraulic features probably has little relationships between groundwater lake-level fluctuation. Model calibra- impact on the overall model solution and surface water. tion used the parameter estimation but increases model uncertainty near code PEST, and calibration targets ditches and wetlands. included heads, stream and spring

Pond in Story Creek headwaters

Ken Bradbury 54 Wisconsin Geological and Natural History Survey Literature cited Alden, W.C., 1918, The Quaternary Bradbury, K.R., Swanson, S.K., Krohelski, Feinstein, D.T., Fienen, M.N., Kennedy, J.L., geology of southeastern Wisconsin: J.T., and Fritz, A.K., 1999, Hydrogeology Buchwald, C.A., and Greenwood, M.M., U.S. Geological Survey Professional of Dane County, Wisconsin: Wisconsin 2012, Development and application Paper 106, 356 p. Geological and Natural History Survey of a groundwater/surface-water flow Anderson, K.M., 2002, Hydrogeologic Open-File Report 1999-04, 66 p. model using MODFLOW-NWT for the controls on flow to Frederick Springs Brown, B.A., Massie-Ferch, K., and Peters, Upper Fox River Basin, southeastern in the Pheasant Branch Watershed, R.M., 2013, Preliminary bedrock Wisconsin: U.S. Geological Survey Middleton, Wisconsin: Madison, WI, geology of Dane County, Wisconsin: Scientific Investigations Report University of Wisconsin-Madison, Wisconsin Geological and Natural 2012-5108, 124 p. MS thesis, 172 p. History Survey Open-File Report Fritz, A.M.K., 1996, Aquifer contam- Anderson, M.P., Woessner, W.W., and 2013‑01, map, scale 1:100,000. ination susceptibility of Dane Hunt, R.J., 2015, Applied groundwater Clayton, L., and Attig, J.W., 1997, County, Wisconsin: University of modeling, 2nd edition: [London?], Pleistocene geology of Dane Wisconsin-Madison, MS thesis, 149 p. Academic Press, 630 p. County, Wisconsin: Wisconsin Gebert, W.A., Walker, J.F., and Kennedy, Aswasereelert, W., Simo, J.A., and Geological and Natural History J.L., 2011, Estimating 1970–99 average LePain, D.L., 2008, Deposition of Survey Bulletin 95, 64 p. annual groundwater recharge in the Cambrian Eau Claire Formation, Cline, D.R., 1965, Geology and Wisconsin using streamflow data: U.S. Wisconsin: Hydrostratigraphic ground-water resources of Dane Geological Survey Open-File Report implications of fine-grained cra- County, Wisconsin: U.S. Geological 2009-1210, 14 p. plus appendices, tonic sandstones: Geoscience Survey 1779-U, 64 p. available at http://pubs.usgs.gov/ ofr/2009/1210. Wisconsin, v. 19, no. 1, p. 1–21. Dane County Regional Planning Bahr, J.M., Hart, D.J., and Leaf, A.T., Commission, 2004, Dane County Gellasch, C.A., Bradbury, K.R., Hart, D.J., 2010, DTS as a hydrostratigraphic water quality plan: Summary plan, and Bahr, J.M., 2013, Characterization characterization tool: Madison, 2004: Dane County Regional Planning of fracture connectivity in a siliciclastic University of Wisconsin, Water Commission, 104 p. bedrock aquifer near a public supply well (Wisconsin, USA): Hydrogeology Resources Institute, 18 p. Day, E.A., Grezebieniak, G.P., Osterby, Journal, v. 21, no. 2, p. 383–399. Batten, W.G., and Attig, J.W., 2010, K.M., and Brynildson, C.L., 1985, Preliminary geology of Iowa County, Surface waters of Dane County: Harbaugh, A.W., 2005, MODFLOW-2005: Wisconsin: Wisconsin Geological and Wisconsin Department of Natural The U.S. Geological Survey modular Natural History Survey Open-File Resources, 99 p. ground-water model—the ground-water flow process: U.S. Report 2010-01, map and cross sec- Diebel, M., Ruesch, A., Menuz, D., 2014, Geological Survey Techniques and tions, 1:100,000 scale. Ecological limits of hydrologic alter- Methods 6-A16, variously p. Bradbury, K.R., and Muldoon, M.A., 1990, ation in Dane County streams: Project Hydraulic conductivity determinations report to the Capital Area Regional Hart, D.J., Schoephoester, P.R., and in unlithified glacial and fluvial materi- Planning Commission, 26 p. Bradbury, K.R., 2012, Groundwater recharge in Dane County, Wisconsin: als, in Nielsen, D.M., and Johnson, A.I., Doherty, J.E., Fienen, M.N., and Hunt, R.J., Estimating recharge using a GIS-based eds., Ground water and vadose zone 2010, Approaches to highly parame- water-balance model: Wisconsin monitoring: Philadelphia, American terized inversion: Pilot-point theory, Geological and Natural History Survey Society for Testing Materials, ASTM STP guidelines, and research directions: Bulletin 107, 11 p. 1053, p. 138–151. U.S. Geological Survey Scientific Bradbury, K.R., Parsen, M.J., and Fehling, Investigations Report 2010-5168, 36 p. Hunt, R.J., Doherty, J., and Tonkin, M.J., 2007, Are models too simple? A.C., 2016, The 2016 groundwater flow Doherty, J.E., and Hunt, R.J., 2010, Arguments for increased parame- model for Dane County, Wisconsin— Approaches to highly parameterized terization: Ground Water, v. 45, no. 3, User’s Manual: Wisconsin Geological inversion: A guide to using PEST for p. 254–262. and Natural History Survey Bulletin groundwater-model calibration: 110 – Supplement, 31 p. U.S. Geological Survey Scientific Hunt, R.J., and Feinstein, D.T., 2012, Bradbury, K.R., and Rothschild, E.R., 1985, A Investigations Report 2010-5169, 59 p. MODFLOW-NWT: Robust handling of dry cells using a Newton Formulation computerized technique for esti- Dripps, W.R., and Bradbury, K.R., 2007, of MODFLOW-2005: Ground Water, mating the hydraulic conductivity of A simple daily soil-water balance v. 50, no. 5, p. 659–663. aquifers from specific capacity data: model for estimating the spatial and Ground Water, v. 23, no. 2, p. 240–246. temporal distribution of groundwater recharge in temperate humid areas: Hydrogeology Journal, v. 15, no. 3, p. 433–444. 55 The 2016 Groundwater Flow Model for Dane County, Wisconsin

Hunt, R.J., and Steuer, J.J., 2000, Simulation ———, 2013, A high resolution ver- Rayne, T.W., Bradbury, K.R., and Muldoon, of the recharge area for Frederick tical gradient approach to hydro- M.A., 2001, Delineation of capture Springs, Dane County, Wisconsin: U.S. geologic unit delineation in frac- zones for municipal wells in fractured Geological Survey Water-Resources tured sedimentary rocks: Guelph, dolomite, Sturgeon Bay, Wisconsin, Investigations Report 2000-4172, 33 p. Ontario, Canada, University of USA: Hydrogeology Journal, v. 9, no. 5, Hunt, R.J., and Zheng, C.M., 2012, The cur- Guelph, Ph.D. thesis, 225 p. p. 432–450. rent state of modeling: Ground Water, Meyer, J.R., Parker, B.L., and Cherry, J.A., Sellwood, S.M., Bahr, J.M., and Hart, D.J., v. 50, no. 3, p. 329–333. 2008, Detailed hydraulic head profiles 2014, Evaluating aquifer flow condi- Konikow, L.F., and Harbaugh, A.W., 2009, as essential data for defining hydro- tions using heat as an in-well tracer: Revised multi-node well (MNW2) geologic units in layered fractured sed- American Water Resources Association package for MODFLOW ground-water imentary rock: Environmental Geology, Wisconsin Section annual meeting, flow model: U.S. Geological Survey v. 56, no. 1, p. 27–44. March 2014. Techniques and Methods 6-A30, 67 p. Mickelson, D.M., 1983, A guide to the Swanson, S.K., 2001, Hydrogeological Krohelski, J.T., 2002, Simulation of Fish, glacial landscapes of Dane County, controls on spring flow near Madison, Mud, and Crystal Lakes and the Wisconsin: Wisconsin Geological and Wisconsin: University of Wisconsin– shallow ground-water system, Dane Natural History Survey Field Trip Guide Madison, Ph.D. dissertation, 436 p. County, Wisconsin: U.S. Geological Book 6, 53 p. ———, 2007, Lithostratigraphic Survey Water-Resources Investigations Niswonger, R.G., Panday, S., and Ibaraki, controls on bedding-plane frac- Report 2002-4014, 17 p. M., 2011, MODFLOW-NWT, a Newton tures and the potential for dis- Krohelski, J.T., Bradbury, K.R., Hunt, R.J., Formulation for MODFLOW-2005: crete groundwater flow through and Swanson, S.K., 2000, Numerical U.S. Geological Survey Techniques and a siliciclastic sandstone aquifer, simulation of groundwater flow in Methods 6-A37, 44 p. southern Wisconsin: Sedimentary Dane County, Wisconsin: Wisconsin Niswonger, R.G., and Prudic, D.E., Geology, v. 197, no. 1–2, p. 65–78. Geological and Natural History Survey 2005, Documentation of the Swanson, S.K., Bahr, J.M., Bradbury, K.R., Bulletin 98, 31 p. streamflow-routing (SFR2) package and Anderson, K.M., 2006, Evidence Macholl, J.A., 2007, Inventory of to include unsaturated flow beneath for preferential flow through Wisconsin’s springs: Wisconsin streams—a modification to SFR1: U.S. sandstone aquifers in southern Geological and Natural History Survey Geological Survey Techniques and Wisconsin: Sedimentary Geology, Open-File Report 2007-03-DI, 21 p. Methods 6-A13, 50 p. v. 184, no. 3–4, p. 331–342. plus digital data. Olcott, P.G., 1972, Bedrock topography of Swanson, S.K., Bahr, J.M., Schwar, M.T., and McLeod, R.S., 1974, A digital-computer Dane County, Wisconsin: Wisconsin Potter, K.W., 2001, Two-way cluster model for estimating drawdowns in Geological and Natural History Survey analysis of geochemical data to con- the sandstone aquifer in Dane County, Open-File Report 1972-03, map, strain spring source waters: Chemical Wisconsin: Wisconsin Geological and 1:100,000 scale. Geology, v. 179, no. 1–4, p. 73–91. Natural History Survey Information Ostrom, M.E., 1965, Cambro-Ordovician U.S. Census Bureau, 2012, Wisconsin: 2010 Circular 28, 91 p. stratigraphy of southwest census of population and housing: U.S. ———, 1975, A digital-computer model Wisconsin: Wisconsin Geological Government Printing Office, CPH-2-51, for estimating hydrologic changes in and Natural History Survey 180 p., https://www.census.gov/prod/ the aquifer system in Dane County, Information Circular 6, 57 p. cen2010/cph-2-51.pdf. Wisconsin: Wisconsin Geological and Parent, L., 2001, An improved hydro- Westenbroek, S.M., Doherty, J., Walker, Natural History Survey Information geologic model for the Token Creek J.F., Kelson, V.A., Hunt, R., and Cera, Circular 30, 40 p. Watershed: Final report to the T.B., 2012, Approaches in highly Merritt, M.L., and Konikow, L.F., 2000, Wisconsin Department of Natural parameterized inversion: TSPROC, Documentation of a computer Resources: University of Wisconsin– a general time-series processor to program to simulate lake-aquifer Madison, Department of Geology and assist in model calibration and result interaction using the MODFLOW Geophysics, 14 p. summarization: U.S. Geological Survey ground water flow model and Poff, R.J., and Threinen, C.W., 1961, Surface Techniques and Methods, book 7, the MOC3D solute-transport water resources of Dane County: chap. C7, 79 p., available at http:// model: U.S. Geological Survey Madison, Wis., Wisconsin Conservation pubs.usgs.gov/tm/tm7c7. Water-Resources Investigations Department, 61 p. Westenbroek, S.M., Kelson, V.A., Dripps, Report 2000-4167, v. 1, 146 p. Rayne, T.W., Bradbury, K.R., and Mickelson, W.R., Hunt, R.J., and Bradbury, Meyer, J.R., 2005, Migration of a mixed D.M., 1996, Variability of hydraulic K.R., 2010, SWB—A modified organic contaminant plume in a conductivity in uniform sandy till, Thornthwaite-Mather Soil-Water- multilayer sedimenteray rock aquifer Dane County, Wisconsin: Wisconsin Balance code for estimating ground- system: Ontario, Canada, University of Geological and Natural History Survey water recharge: U.S. Geological Survey Waterloo, M.S. thesis, 626 p. Information Circular 74, 19 p. Techniques and Methods 6-A31, 60 p. 56

Bulletin 110 Bulletin n The 2016 Groundwater The Flow Model for Dane County,

Published by and available from: Wisconsin Geological and Natural History Survey 3817 Mineral Point Road n Madison, Wisconsin 53705-5100 608.263.7389 n www.WisconsinGeologicalSurvey.org Wisconsin Kenneth R. Bradbury, Director and State Geologist

This report is an interpretation of Issued in furtherance of Cooperative the data available at the time of Extension work, Acts of May 8 and Our Mission The Survey conducts earth-science preparation. Every reasonable efort June 30, 1914, in cooperation with surveys, field studies, and research. We has been made to ensure that this the U.S. Department of Agriculture, n provide objective scientifc information 2016 interpretation conforms to sound University of Wisconsin–Extension, about the geology, mineral resources, scientific principles; however, the Cooperative Extension. University of water resources, soil, and biology report should not be used to guide Wisconsin–Extension provides equal of Wisconsin. We collect, interpret, site-specifc decisions without opportunities in employment and disseminate, and archive natural verification. Proper use of the report programming, including Title IX and resource information. We communicate is the sole responsibility of the user. ADA requirements. If you need this the results of our activities through information in an alternative format, The use of company names in this publications, technical talks, and contact the Office of Equal Oppor­tunity document does not imply endorsement responses to inquiries from the and Diversity Programs or the Wisconsin by the Wisconsin Geological and Natural public. These activities support Geological and Natural History History Survey. informed decision making by Survey (608.262.1705). government, industry, business, and ISSN: 0375-8265 individual citizens of Wisconsin. ISBN: 978-0-88169-992-0