Question & Answer

Locals are concerned about stepping back 60 years of flood protection to “…basically having almost no flood protection.”  People will be shocked, then frightened, and eventually, angry. Rational or not, that will be the reaction.


Strong reactions are understandable as people face the uncertain consequences of more extreme storms, and then grapple with the risks, benefits, and costs of flood control dams. We would not expect people in the Coon Creek and West Fork Kickapoo valleys to be passive about a reversal in 60 years of flood protection strategy.




Would you say, given that dams control only 19% of the watershed in Coon Creek, and 35% of the watershed in the West Fork, that flood control (such as a 500 or 1,000-year event) is coming from historic achievements in land treatment?


The 1996 study by William Krug, USGS, found that changes in agricultural practices reduced the 50-year peak discharge in Coon Valley by 57% between two periods of time: 1934/40 and 1978/81. The 14 flood control dams further reduced those early flows by an additional 7%.

Yes, land treatment improvements since the 1930’s have been the most influential factor in flood reduction at the lower end of the valleys. The big gains in land treatment have been attained by the agricultural community. Those gains must be maintained to protect the most substantial reductions in flood flows. Moving forward, practical improvements in land treatment are needed and encouraged, but the effect of those improvements would not be sufficient to replace the localized effects of a watershed dam.

The dams were designed to store and release runoff from a 50-year frequency rainfall event. The dams have minimal effect on larger storms because the runoff routes through open, vegetated spillways or overtops the dams. The influence of dams diminishes substantially moving downstream as more uncontrolled drainage basins contribute flow to the valleys.




People say we’re getting bigger and bigger floods, and we need all the protection we can get. Why would you take away the dams when things are getting worse?  I think the answer is “it will simply cost more to build safe dams, given the hydrogeology of the area, than the value of the benefits that will be achieved.”


Correct. On a watershed scale, the flood reduction benefits do not exceed the construction cost of replacing the dams due to the small acreage actually protected by the dams. The exception is Jersey Valley dam due to its recreational value.




But, that doesn’t answer the simple, common sense question above, “Why now when things are getting worse?”


We are recommending the decommissioning of dams now, because five dams recently failed, and the failures exposed serious vulnerabilities in the remaining dams. As we now evaluate the benefits and costs, the economic analysis does not support replacement of those failed dams.

We recommend the immediate decommission of high hazard dams due to the risk of failure and potential for loss of life. This recommendation is based on a total of 10 dam failures since 1978 in similar geography, and a detailed examination of the highly fractured, sandstone geology, and methodology used in the original dam construction. The remaining low hazard dams can be maintained until they fail or encounter repair costs that exceed their allocated benefits; then they should then be decommissioned.




An economic analysis doesn’t speak to the problem that people are experiencing on the ground, which is a combination of physical issues, but as or more importantly, emotional issues.


The economic analysis does not speak to the emotional issues and stories of personal property loss and recovery after extreme storms and flooding. However, it does provide an objective assessment for making important public works decisions on a watershed scale.




Our big problem is that the storms are simply getting bigger (many would say as a result of climate change) and essentially, the magnitude of the infrastructure that would be needed to “control” a 500 or 1,000-year flood event is massive, and massively expensive. And this is true all over the United States - not just in our little slide of paradise. But doesn't that mean that the old parameters for economic analysis need to shift to accommodate this new normal? We seem trapped in a legal framework, administrative rules and procedures, hopelessly outdated data, and analytic assumptions that don’t speak to this new normal.


A new normal in the frequency and magnitude of storms should not shift public policy toward funding projects that cost more than the benefits provided. However, a new normal does present environmental and social factors that must be evaluated prior to spending public funds, such as the potential threat to emergency access routes, the mitigation of flooded properties, the effect on fisheries, the effect on water quality and human health, etc. The public scoping process is a mechanism of the National Environmental Policy Act (NEPA) to uncover these types of factors under a new normal.

An economic analysis is the typical litmus test for a public works project, and it is embroidered into the legal framework, administrative rules, and assumptions for public funding. However, an economic analysis is only part of a Watershed Project Plan and environmental analysis which also evaluates potential non-monetary impacts, social concerns, and speaks to the new normal. We are squarely in the middle of this process and asking the public, “What are we missing?”




What about the dams in the Bad Axe Watershed? Six out of eight dams have gone over the 50-year mark, with Runge Hollow’s planned service life expiring in three years and Duck Egg in sixteen years. Of those eight, the Department of Natural Resources (DNR) characterizes three of them as high hazard, one as significant hazard, and four as low hazard. It’s likely that Sidie Hollow and Esofea may be similar to Jersey Valley in that their recreation value may be what causes them to be considered worth relocating. Also, the one and only PL-566 dam in Crawford County, the Blackhawk-Kickapoo Dam on Nederlo Creek, is four years out from its 50-year service life, and the NRCS is working with our LCD on a proposed tiling project to prevent seep. Are the dams in neighboring watersheds and counties impacted by this study?


Federal interest in the original construction of PL-566 watershed dams is complete after 50-years, at which time, the counties have autonomy in the final disposition of the dams. Fifty years is also the economic life or service life of these watershed dams. After 50 years, the materials used in dam construction will require more frequent inspection and repair. Accumulated sediment will start to fill space behind the dam reserved for flood water retention. At some point, the dams will encounter repair costs that exceed their allocated benefits.

The Watershed Project Plan and environmental analyses being developed for Coon Creek and West Fork Kickapoo watersheds is likely to set the standard for re-planning expired dams in Southwest Wisconsin. Market conditions, construction costs, and climate have changed substantially since the dams were constructed in the 1960’s. Re-planning dams collectively will be a precursor to repairing a single dam with Federal and state funding. Several watersheds have multiple dams that were planned and constructed collectively in order to move the needle with regard to flood control, and as such, decisions should not be made on a dam-by-dam basis.




What happens to the economic analysis if the rainfall depth for a 24-hour, 100-year storm, is 9.1 inches instead of 7.5? Going by this chart, the storm that breached the dams was actually a 1,000-year storm - what if you use that rainfall amount? Does the flood control become more or less valuable?


The benefit-cost ratio for watershed dams will increase if the storm precipitation depths increase. However, we do not want to bias the hydrologic model results by stacking conservatism upon conservatism, i.e. using the upper bounds of all data inputs.

A planning study focuses on a comparison of the mean predictions, and then develops a confidence interval based on the uncertainty in all of the factors used in the model. Uncertainty usually arises from errors during calibration, the design of model structure, and measurements of input data like precipitation amounts. We can also apply the upper and lower bounds of uncertainty to other factors in the flood model such as rainfall distribution patterns, soil infiltration rates, ground cover condition, flow path velocities, channel roughness, etc. The planning team has a quality assurance process that draws upon a number of techniques for determining the reliability of model predictions.

The question eludes to an uncertainty analysis which is an indispensable part of hydrologic study. It will be presented and discussed with the preferred alternative. An uncertainty analysis investigates the uncertainty of variables used in the decision-making process in order to assess confidence in the recommendations.

The table above is part of a large storm analysis by Daniel Wright, et. al. from the University of Wisconsin-Madison. To address the issue of climate change, the NRCS commissioned a special study to evaluate the intensity, duration, and frequency of large storms in the Coon Creek and West Fork Kickapoo watersheds over the last 18 years. Rainfall depths and distributions are typically sourced from NOAA Atlas 14 data, which are derived from historical rain gage records that span 30 years, but do not include data more recent than 2012.

This table compares the mean rainfall depths from Atlas 14 to the mean rainfall depths derived from the UW analysis. It shows that the 24-hour rainfall depths for the various storm probabilities have increased about 0.6 inch. Higher precipitation depths will increase the projected benefits of watershed dams, and the configuration of new dams.




By removing the dams and letting full flood flow pass, will the people already identified downstream in the watersheds be more or less safe? Is your conclusion that danger is greater from a breach of a dam than from the straight, uncontrolled flood waters?


The recommendation is to decommission the existing failed and remaining high hazard dams. A sudden breach of a high hazard dam presents a greater risk for loss of life than straight, uncontrolled floodwater. The response time is significantly less and the breach inundation area is larger than an uncontrolled floodplain.

Alternatively, homes and businesses within the breach inundation area could be removed to reclassify the dams as low hazard. The recommendation is to decommission low hazard dams after they fail or encounter a repair that exceeds the dam's allocated benefits. A sudden breach of a low hazard dam presents a greater risk for property damage than straight, uncontrolled floodwater. Breach inundation areas are zoned to prevent residential development so that dams can retain their low hazard classification.




We would like to understand the assumptions used to assign runoff curve numbers in the watersheds.  What data was used to determine the runoff curve number and land use? Can infiltration rate tests be done on the ground in order to calibrate/cross-check the runoff curve number assumptions?


Runoff curve numbers are used to model the volume of runoff from the land during a rainfall event. It is an important factor in storm hydrograph development. Runoff curve numbers were estimated using the EVAAL model developed by the DNR Water Quality Team (2014).

A runoff curve number is based on the following factors:

  • Land cover, includes crop rotation and canopy cover
  • Land treatment
  • Hydrologic soil group (average infiltration rates)
  • Antecedent moisture condition (soil moisture profile prior to a storm)
  • Initial abstraction (amount of precipitation held in micro storage)

Land cover is downloaded directly from the USDA-National Ag Statistics Service. The last 5 years of USDA-NASS National Cropland Data Layers (CDL) were used to determine generalized crop rotations within the watersheds. The CDL is a raster, geo-referenced, crop-specific land cover data layer created annually for the continental United States using moderate-resolution satellite imagery and extensive validation. The EVAAL model combines land cover with hydrologic soil group value, derived from the USDA-NRCS Soil Survey Geographic Database (SSURGO), to determine the associated curve number in the NRCS Runoff Curve Number table.

Land treatment is difficult to assess without direct field observations, so the model outputs both a high estimate of runoff potential (high curve number) and a low estimate (low curve number) representing both worst-case and best-case infiltration scenarios. Land treatment includes the following descriptions:

  • bare fallow,
  • crop residue,
  • straight row,
  • straight row + crop residue,
  • contoured,
  • contoured + crop residue,
  • contoured + terraced,
  • contoured + crop residue + terraced.

The median value in the curve number range was used to represent a “fair” or average hydrologic condition for hydrograph development and comparative floodplain analysis of the “dam” and “no dam” scenarios. The low end of the curve number range was used to model “good” or improved land treatment, the infiltration equivalent of meeting tolerable soil loss conditions everywhere in the watershed, which in some locations would be land that is contoured + crop residue + terraces (or water and sediment control basins). This low end curve number was evaluated by local conservationists, Bob Micheel and Sam Skemp. They determined that the most practical land management strategies and improvements that could be attained in the watersheds with concerted effort would likely be represented by a curve number between “fair” and “good.”

Typical conditions and values were assumed for antecedent moisture condition, initial abstraction, and canopy cover.

Infiltration rate tests, alone, would not be sufficient to calibrate or affirm the runoff curve numbers. Runoff curve numbers were derived (back-calculated) from research in gauged watersheds where the hydrologic conditions were well-characterized or known. Runoff curve numbers can only be verified or cross-checked by comparing modeled runoff hydrographs to gauged flow.




We would appreciate enhanced effort included in the Watershed Plan for upland practices, and how our region can plan for the work ahead whether rebuilding or decommissioning is selected.


The Watershed Project Plan will describe the flood control benefits of improving the hydrologic condition in the watersheds. The Plan will show the reduction in peak discharges, reduction in flooded acres, and increase in economic benefits that result from reducing the average runoff curve numbers. NRCS and county conservationists will need to prescribe practical land treatments and cover conditions (specific conservation practices), and target a percentage of watershed land use that is necessary to achieve those average runoff curve numbers.




Can decommissioning include assessment of making decommissioned dams a constructed wetland?


Wetland plants and wildlife exist in the saturated sediment pools behind the failed and existing dams. These sites largely are considered constructed or artificial wetlands. When the dams are decommissioned, a natural stream channel will develop through the sediment pool. These sites will drain and revert to riparian buffers or floodplain wetlands.

The present character of these wetlands can be maintained with a grade stabilization structure (e.g. sheet pile weir). The natural stream channel would need to be trained and hardened to flow over this structure. Recent sheet pile weir installations for wetland restorations work in Wisconsin cost about $250,000 for construction and $25,000 for geotechnical investigation.

Constructed, in-stream wetlands would alter the natural geomorphic process in the valley. A grade stabilization structure in the floodway would be a barrier to aquatic organism passage and require significant operation and maintenance. Most wetland restorations, enhancements, or creations are completed in uplands or off-channel (in the floodplain fringe) to reduce costs, better manage water levels, and avoid sedimentation.




It is imperative that at no point in this process we gloss over upland practices work. That is a significant matter folks want to discuss. For instance, the questions from our local reporter were rushed/cutoff on this topic, even though her questions greatly represent the concerns and pulse of this community.


Agreed. Upland land treatment practices are responsible for reducing peak discharges in the valley by 50% or more compared to 1934-1940. The Project Plan will model the effects of further reductions in runoff curve number based on practical changes in land treatment and cover conditions. Upland land management strategies are critical because decommissioning dams will expose valley properties to the full brunt of uncontrolled, upland runoff.




Another concern pertains to people who know the land around here, they know that most "no-till" fields are a joke. In other words, it is not uncommon for the phrase “no-till” to be used in instances where farmers do not actually do true no-till. This is particularly problematic if no-till is considered in the runoff curve number since it would does not reflect accurately what is happening in the field. Not all land use practices considered ‘conservation’ are uniform in the results they produce. It is my hope that runoff curve numbers are adjusted to consider actual land uses in place.


In order to model “with dam” and “no dam” conditions, runoff curve numbers were selected to represent typical land use under “fair” hydrologic conditions. In order to model the effect of land treatment improvements, runoff curve numbers were selected to represent “good” hydrologic conditions, i.e. the most practical, long-term, land treatment and cover conditions that could be attained in the watersheds with concerted effort.




I think we need a better and/or more precise definition of what we are identifying as conservation farming and how runoff curved numbers vary between say contour strips versus no-till versus grazing compared to woodland etc.


Runoff curve numbers were based on long-standing and accepted research values for the following land treatment combinations: straight row, contoured, terraced, crop residue, and fallow under “poor,” “fair,” or “good” conditions. Specific values were not assigned to no-till, cover crops, organic, or other soil health strategies.




No value seems to be correlated to true soil health practices. Is it possible to get real numbers on how infiltration rates are impacted by increased organic matter? And for those infiltration rates to be included in the analysis?


Specific values were not assigned to no-till, cover crops, organic, or other soil health strategies. Developing runoff curve numbers for these prescriptive land management strategies is beyond the scope of this study. Runoff curve numbers for these strategies must be inferred from the NRCS Runoff Curve Number table based on whether they are generally facilitating or hindering infiltration.




With upland treatment, we really don't need to hear that "existing local staff will use existing EQIP' to accomplish this. The community needs to hear how all of this work and a new watershed study is going to help lead to more capacity to address our flooding. Unfortunately, local landowners are far too familiar the limited resources and capacity to assist, so it is imperative to present more information to the public on how this study can leverage NEW RESOURCES.


A Watershed Project Plan and environmental analysis can leverage new resources by documenting a need that is environmentally, socially, and economically defensible. It does not include a financial plan for implementation of the preferred alternative for flood control, such as conservation workload analysis, real property acquisition, bond referendums, tax increment financing, Federal and State grants and cost-share, public-private partnerships, watershed appropriations and earmarks, recreational user fees, revenue taxes, etc.

The watershed planning study documents the first 7 Steps of Conservation Planning. Up to this point, we have been identifying problems and opportunities, evaluating resource data, and formulating alternatives. In Step 7, a decision will be made on the preferred alternative. The Watershed Project Plan document will provide a comparative analysis of the alternatives, and present the benefits, costs, environmental and socio-economic consequences of those alternatives.




During the meeting it was cited that the 100-year flood is x amount higher at the bridge in Coon Valley or Avalanche without with the dams versus with them in place. This current assessment seems to conflict the report done by SCS around 1980. Why do you consider these two measurements to be so different? More specifically, that report identifies a 17% reduction of peak flow in 100-year flood with working dams in place and now in 2021 we are talking a significantly less effect.


The latest hydrologic study in Coon Valley was prepared by William Krug, USGS, 1996. Krug presented a table with a comparison of peak discharges in Coon Valley, with and without land treatment, with and without dams, for the full range of storm probabilities.

The findings of this study do not conflict with Krug’s conclusions. Krug reported a 5-10% reduction in peak discharges at Coon Valley as a result of the dams. A reduction in peak discharge does not necessary result in a 5-10% reduction in flooded acres or economic damages.




A related concern pertains to how the model estimates runoff events compared to experiences on the ground, for instance: while the modeling uses NOAA’s Atlas 14 data (for instance, a 100-year storm occurring in 24-hour period), we observe more intense rainfall events over much shorter durations of time. In other words, we are getting more intense rainfalls, with higher amounts, in a shorter time period so is this discrepancy being captured?


Yes, we are modeling the results of increased storm precipitation depths in the Coon Creek and West Fork Kickapoo watersheds based on a large storm analysis by Daniel Wright, UW-Madison. The Wright study creates a hydrologic record over the last 18 years. Rainfall depths and distributions are typically sourced from NOAA Atlas 14 data, which are derived from historical rain gage records that span 30 years, but does not include data more recent than 2012.

The Wright study indicates that the 24-hour rainfall depths for the various storm probabilities have increased about 0.6 inch. Higher precipitation depths will increase the projected benefits of watershed dams, and the configuration of new dams.




With that being said, there is a strong sense from many in the community that flooding now seems notably worse from experiences with the dams working.


These observations are correct. The frequency and magnitude of flooding will increase without the dams, but the effect will diminish with distance down the valleys. However, the hydraulic and economic models show that the benefits of existing flood control protection do not exceed the costs of replacing the dams with safer construction.




When initial reflections came out, we heard the phrase “the economics look promising” meaning that we might meet the cost benefit ratio.  However, at some point the tune drastically changed with the exception of WF-01/Jersey Valley.  I am wondering why or how that it changed so quickly?


The benefit-cost ratio is significantly higher for Jersey Valley Dam than for remaining dams in the two watersheds.




The value derived from the acres protected seems like a poor way to quantify flood control benefits. The dams are silent sentinels working during every flood event for all acres in the floodplain.  They limit the flood damage downstream during a 5-year, 10-year….500-year event.  With Bob mentioning 20 large events that have occurred since 2007, it seems that the flood control benefits are underrated.  The values of township and county infrastructure repairs can be obtained from FEMA funded repairs over the years.  Some townships have poured millions of dollars into repetitively damaged roads, bridges, culverts since 2007. For examples, one township in Monroe County repaired a road in Ruland’s coulee with 250 loads of rock and gravel only to have it blown out a week later.  It is important to note that this damage and repair happened one week after the dams breached in 2018. Have these actual costs been included in the cost/benefit analysis?


“Floodplain acres protected” provide a baseline for comparison: with and without dams; with and without land treatment improvements; with and without floodway improvements. This measure helps us define and estimate losses in agricultural production, recreational opportunities, buildings, roads, and utilities. It does not represent random, unique, or extraordinary property damages such as hay storage on cropland, or storage units with a concentration of high value items. Consistent, high value losses should be brought to our attention.

Damage recovery that occurred since 2007 was factored into the retrospective economic analysis if the model results show that the dams would have prevented the damages. The retrospective analysis cumulated benefits by multiplying flood damages (prevented) by the actual number of storm occurrences over the last 60 years. Likewise, the projected economic analysis accounts for flood damages (prevented) and the probability of flood occurrence using the entire storm recurrence series, i.e. 2-year, 5-year, 10-year, etc.

Road crossing failure is assumed when the hydraulic model indicates that a road overtops. When a dam prevents that same storm from overtopping a road, flood benefits accrue based on a typical replacement cost of the road crossing.




I have seen the devastating effects of the hydrograph peaks in valleys without dams.  Ones that quickly pop to mind are Reads Creek, Tainter Creek, and Rush Creek.  The townships of Franklin, Sterling and (Freeman, and Utica in Crawford County) could attest to the damages.  However, when you look at the watersheds that have dams (Bad Axe, Coon Creek, and the West Fork) the streams “behave” themselves.  They are not constantly migrating across the valley floors exposing vertical banks of topsoil, fiber optic lines, culverts, and bridges. For instance, the County Road JJ bridge was blown out from the 2016 flooding along with a significant length of trout habitat improvements in Reads Creek watershed without a flood control structure. Bridge repairs were not completed for 3.5 years and significantly affecting local traffic roads.


The loss of land to “aggressive” channel migration was not included in the analysis, because it is a random and unpredictable process; and channel migration eventually sets up an equal amount property gain and loss on each side. The channel has moved back and forth across the valley for hundreds of years, but the channel width remains the same.

Property damages caused by “aggressive” channel migration, such as buried utilities and fence lines, which occur during ordinary high-water conditions, were not included in the analysis, largely because it is a random and unpredictable process.

Property damages caused by large storms, such as overtopped roads and bridges, and stream habitat improvement projects, were included in the cumulation of benefits if the dams could have prevented the occurrence.




I compared GIS layers for the 100-year floodplain with no dams, and the 100-year floodplain with 3 failed dams, that were developed for Coon Creek with the FEMA FIRM Flood Hazard Zones.  As would be expected, in most areas and especially in areas where a Floodway study was performed for the FEMA maps, the data layers are very consistent.  However, it is noticeable in the upper reaches that the FEMA Zone A in many areas has a significantly smaller footprint than the recent study results.  Given the nature of Zone A’s, I would assume that your data is much more reflective of the actual potential flood coverage than the FEMA Zone A delineations are.


I believe it would be useful to look at the online map data with the FEMA Flood Hazard Zones (and maybe a couple of the other layers that are available like the cross-sections) from the FEMA public/NFHL map service (https://hazards.fema.gov/gis/nfhl/rest/services/public/NFHL/MapServer).

FIRM map overlays may provide useful context, but they may also distract from the focus of this study. Although very similar, the floodplain models will not meet the current regulatory standard (i.e. level of effort and administrative conditions) for a Flood Insurance Study and Flood Insurance Rate Maps (FIRMs). FIRMs provide detailed information about the base flood, or areas that have a 1% annual chance of flooding and a 26% chance over the life of a 30-year mortgage, according to FEMA. FIRM maps are calibrated to gauged watershed and therefore retrospective in nature.

This study models the existing and future hydrologic conditions in the watershed. Sub-basin hydrographs are developed and routed through the watershed, with and without dams, which were designed to control runoff from the 50-year storm.




I believe it would be helpful for people to know how the regulated floodplain may change if certain dams are decommissioned and we have updated floodplain data.  That also brings up the question of how your final floodplain data would be used/provided to FEMA for FIRM map updates.  Is that part of this process in some way, or would that be a totally separate deal?


Separate. The disposition of watershed dams may not affect the FEMA Flood Insurance Study and the resulting Flood Insurance Rate Maps (FIRMs), because the dams do not significantly affect the base flood elevation (100-year floodplain). The dams were designed to control runoff from the 50-year storm.

A sudden dam failure results in a breach inundation area (dam shadow). These areas are regulated or zoned from development in order to maintain low hazard dam classifications. If these areas are encroached by development, dams may be re-classified as high hazard and require expensive rehabilitation projects to increase their level of protection and reduce the potential for loss of life. If dams are decommissioned, zoning would likely be changed in the breach inundation areas only.