factors controlling rates of bluff recession at two …davem/abstracts/05-8.pdffactors controlling...

Factors Controlling Rates of Bluff Recession atTwo Sites on Lake Michigan

Elizabeth A. Brown1, Chin H. Wu2, David M. Mickelson1,*, and Tuncer B. Edil1

1Geological Engineering ProgramUniversity of Wisconsin – Madison

1415 Engineering DriveMadison, Wisconsin 53706

2Civil and Environmental Engineering Department University of Wisconsin-Madison

1415 Engineering Drive Madison, Wisconsin 53706

ABSTRACT. Empirical relationships between recession rate of bluffs and precipitation, storm fre-quency, lake level, deep water wave power, and wave impact height are derived for two Lake Michiganshoreline reaches in Wisconsin. Recession rates are determined from digital orthophotos constructedusing historical aerial photographs at least once every decade from the 1940s to present. The recessionmeasurements represent spatial averages of rates measured at increments of 10-20 m along the shorelineover a distance of about 500–700 m. The temporal variations in recession rates over intervals rangingbetween 6 and 17 years were determined for the toe and crest at the sites with high (30–45 m) bluffs andfor the crest, but not the toe at the site with low (9–11 m) bluffs. Trends in precipitation, storm frequency,and deep water wave power show weak relationships with changes in bluff recession rate. Both the toe ofthe high bluffs and the crest of the low bluffs show temporal recession-rate patterns that closely matchthe changes in the average lake-water level. The crest of the high bluffs recedes at a rate that is relativelyinsensitive to lake level changes. The annual average of the peak monthly wave-impact height appears tobe the best predictor of bluff recession rate over the intervals studied.

INDEX WORDS: Lake Michigan, bluff recession, shore erosion, lake level, wave-impact height.

J. Great Lakes Res. 31:306–321Internat. Assoc. Great Lakes Res., 2005

INTRODUCTION

Bluff shorelines along the Great Lakes erode atdifferent rates depending upon a variety of externalcontrols. The rate of shoreline recession varies spa-tially, however, at a given shore reach, past reces-sion rate can be measured. Traditionally, the pastrecession rate has been used as a basis of estimatingfuture recession. However, recession rate variestemporally at a given shore reach in response a va-riety of external controls, so past recession rate maynot be a good predictor of future rates. In order toimprove our ability to predict long-term rates of fu-ture bluff recession, we developed empirical rela-tionships between historic recession rates of blufftoes and crests and environmental conditions in-cluding precipitation, storm frequency, deep water

*Corresponding author. E-mail: [email protected]

306

wave power, lake level, and wave-impact height attwo sites on the west shore of Lake Michigan (Fig.1), high (30–45 m) bluffs in Ozaukee County andlow (9–11 m) bluffs in Manitowoc County.

Overview of Bluff Processes

Because much of the Great Lakes shoreline hasbluffs of till or lake sediment above the beach(Pope et al. 2001), one component of shore erosionis bluff instability. While bluff material properties(including strength, i.e., angle of internal frictionand cohesion, and unit weight), slope geometry,stratigraphy, and groundwater level determine thestatic stability of a slope, natural time-varyingweathering processes including precipitation,freeze/thaw action (Wilcock et al. 1998, Sterrett1980), sheet wash (Sterrett 1980), seepage effects,

Factors Controlling Rates of Bluff Recession at Two Sites on Lake Michigan 307

and wave action can complicate bluff stability(Mickelson et al. 2004). Edil and Vallejo (1980) re-port a distinct reduction in the inclination of a sta-ble slope when the groundwater level, whetherperched or not, rises from 1/4 to 3/4 of the slopeheight. Recent studies (Montgomery 1998; Chase etal. 2001a, b) show that groundwater and bluffstratigraphy can affect short term bluff recessionrates. On the other hand, we believe that erosion bywaves is likely the main determinant of the long-term recession rate of bluffs because wave erosionprevents the bluff slopes from ever attaining equilib-rium. Wave action at the toe of the slope serves toweaken and remove exposed bluff material, therebyundercutting the toe of the overall slope and reduc-ing the stability, and ultimately causing failure.

The long-term response to wave erosion is fur-ther complicated by a changing slope geometry.Over time, the slopes evolve in response to the ef-fects listed above. Studies (Edil and Vallejo 1980,

Mickelson et al. 2004) have shown that the patternand rate of slope change depends upon bluff height,stratigraphy, soil type, and vegetative cover. Low(10 m or less) bluffs, such as those at the Mani-towoc County site, respond more rapidly to lakelevel, wave climate and precipitation patterns thanhigh bluffs. Generally, low bluffs that experienceerosion at the base have no trees and very little veg-etation due to the short “cycle” time of the slopefailures. The predominant slope processes of lowbluffs are shallow slumps, translational slides, andface degradation. In contrast, high (30–45 m)bluffs, such as those at the Ozaukee site, changeslowly because of the long “cycle” time to erodethe large mass of material at the base after failure.An episodic failure mode is usually exhibited bythe high bluffs (Edil and Mickelson 1995, Mickel-son et al. 2004). Figure 2 shows a typical sequence

FIG. 1. Map showing location of Ozaukee andManitowoc sites and detailed site maps.

FIG. 2. Phases of episodic changes for a highcohesive coastal bluff typical of the Ozaukee bluff(1) Steep unstable bluff; (2) large, deep-seatedslump takes place causing up to 50 feet of bluffrecession; (3) wave erosion of toe begins; (4) waveerosion continues lower bluff steepens; (5) waveerosion continues, lower steep segment of bluffgrows higher; (6) failure occurs again. Cycle maytake more than 50 years to be completed.

308 Brown et al.

through which these high bluffs pass. Large, deep-seated slumps occur locally at a rapid rate, deposit-ing the material at the base of bluff. The materialacts like a buttress for a number of years until thewaves erode the failed sediment. The waves thenresume their direct attack on the intact bluff faceand another large, deep-seated failure occurs even-tually.

Beach and Nearshore Processes

Change in lake level is commonly considered tobe one of major factors controlling bluff retreat rate(Bray and Hooke 1997, Kirk et al. 2000, Carter1976). Natural lake level variations can be subdi-vided into three categories: short term (changeswithin a few days or less), medium term (changeswithin a year), and long term (changes over a fewyears or more). Short-term changes are due towind-stress buildup, barometric pressure changes,and seiche (resonant oscillations) activity and donot represent a change in the volume of water. Astrong uni-directional wind lasting from hours toseveral days can cause water to rise significantlyfor a few tens of hours. These effects are more pro-nounced at the confined north and south ends ofLake Michigan than in mid-lake.

Medium-term (seasonal) changes affect an entirelake and are caused primarily by differences in ratesof runoff and evaporation. A typical seasonal cyclefor Lake Michigan shows a high in June-July andlow in January-February. Long-term changes arecaused by major variations in the climate within theGreat Lakes basin and glacio-isostatic rebound(Carter 1976). For the length of time considered inthis study, these long-term changes in lake level areprobably not important.

Coupled with sufficiently high water levels,storm generated waves have been considered aprincipal cause of recession along shorelines (Jib-son and Odum 1994, Dewberry and Davis 1994,Davidson-Arnott and Pollard 1980, Sunamura 1976,Sunamura 1977). Unlike lake level, which playsonly a passive role in coastal erosion or “sets thestage” for erosion to occur, wind generated waveson the Great Lakes are capable of eroding till bluffsboth directly and indirectly (Kamphuis 1987, Carter1976). Some waves in deepwater break under somecircumstance (Wu and Nepf 2002, Yao and Wu2004) and less energy would reach the beach. Oth-ers keep propagating to nearshore and reaching theshoreline. In cases where the beach is narrow (e.g.,times of high lake level or after erosion episode) a

much greater proportion of wave energy reaches thebluff, causing erosion (Sorensen 1997). The contin-uous onslaught of waves serves to erode and washaway the intact, exposed bluff face and to removeslumped material at the base of the bluff. Thus, theerosion at the base of the bluffs is a continuing, butnot a continuous process.

Davis et al. (1973) consider differences in thelocal rates of erosion along the eastern Lake Michi-gan shore to be more attributable to the presence orabsence of nearshore sand bars and man-madecoastal structures than to the relatively long-termfluctuations of lake level. In particular, nearshoredowncutting of cohesive shorelines impacts bluffstability indirectly. Nearshore downcutting is thegeneral planning down of the lakebed surface dueto the abrasive action of erosive waves in the pres-ence of sand. Kamphuis (1987) showed that thewave-related processes taking place on the fore-shore could actually control the long-term rate ofbluff erosion. This can occur where significantdowncutting produces a deeper water condition,and allows more wave energy to impact the bluffsthan previously would have taken place. A realthreat of nearshore downcutting exists for all tillshorelines where there is an insufficient sand sup-ply to provide adequate protection of the till lakebed from wave action, which is probably a majorfactor in the erosion of most till bluffs.

DESCRIPTION OF REACHES

The sites represent two common bluff types onthe Great Lakes. The Manitowoc site is a low bluffmade up mostly of clayey till and clayey lake sedi-ment. The Ozaukee reach has a very high bluff con-sisting of clayey till at the base, sandy and silty lakesediment in midslope, and clayey till at the top.Both sites have relatively wide beaches and littleartificial armoring of the shoreline.

The Manitowoc reach is located at the north edgeof Manitowoc County, Wisconsin. It extends fromthe Two Creeks Nuclear Power Plant in the south tothe Kewaunee Nuclear power plant in the north, adistance of about 7 km (Fig. 1). The low bluff(< 12 m) undergoes mostly parallel retreat by shal-low slides and face degradation. Although recessionhas been fairly rapid, the shapes of profiles WTR-1through WTR-2 show no noticeable change sincethe mid 1970s. A detailed description of bluff evo-lution is given in Chapman et al. 1997b. The landuse at the top of the bluff is agricultural.

The Ozaukee reach is located in southern Ozau-


kee County, Wisconsin and is about 4 km long. Thesite is characterized by high bluffs, generally be-tween 35 and 50 m. The erosion taking place here isdominated by large-scale slumps and is relativelysevere. The large, deep seated failures appear tooccur only rarely (on the order of 10s to 100s ofyears), but there is a substantial loss of bluff top(often 15 m or greater) when such failures occur.Wave attack at the base of the bluff slowly steepensthe overall slope of the bluff, leading to failure. Adetailed description of these bluffs is presented inChapman et al. 1997b. Profiles WPW-1 through 5were measured in 1999 (Brown 2000) and show nonoticeable change from those in Chapman et al.1997b. The land use at the top of the bluff is rela-tively new housing development, cleared land, andConcordia College.

METHODS AND DATA SOURCES

Recession Measurements

Given the complexity of the bluff-beach-nearshore system, it is still very difficult, if not im-possible, at this time to construct a numerical modelto predict rates of future bluff retreat. Therefore, wetook the empirical approach of comparing bluff re-cession rates during certain time intervals in thepast with several key variables. Recession rate at asmany time intervals as possible were obtained. Re-cession rates were determined from digital or-thophotos constructed using historical aerialphotographs taken at least once every decade fromthe 1940s to present. The temporal variation in re-cession rates over intervals ranging between 6 and17 years were determined for the toe and crest ofthe site with high (30–45 m) bluffs and for thecrest, but not the toe at the site with low (9–11 m)bluffs.

Historical aerial photographs were used to createdigital orthophotos. These raster images were usedto map bluff lines directly on the computer screen.Air photographs used to determine recession rateswere collected from a number of sources and camein a variety of forms. The details of the air pho-tographs used to measure the recession rates at theOzaukee site are summarized in Table 1. Two setsof photos (1956 and 1995) were in the form of digi-tal orthophotos at the outset of the study (fromS.E.H. Baker 1997). The historical airphotos usedto create orthophotos for the Manitowoc Site arelisted in Table 2. Kruepke (2000) had already cre-ated digital orthophotos from aerial photographs foryears 1938, 1961, and 1967. The 1952 and 1992

digital orthophotos were also available (fromS.E.H. Baker 1997). The digital orthophotos for theremaining years of record, 1975 and 1999, wereconstructed for this study using Orthomapper™.

For both sites 30-m resolution DEMs and USGSdigital orthophoto quadrangles (DOQQ) were usedto create orthophotos for the 3.2 km-long Ozaukeesite. Elevation data were stored as profiles with aspacing of 30 m. We modified the existing DEMs tooptimize the accuracy of the bluff crest and toe po-

TABLE 1. Air photography used for analysis atthe Ozaukee site.

Year Date Scale Type and Data form

1941 September 7 1:20,000 Black and WhitePhoto Prints

1950 October 3 1:20,000 Black and WhitePhoto Prints

1956 May 16 1:20,000 Black and WhitePhoto Prints andDigital Orthophoto

1964 June 4 1:20,000 Black and WhitePhoto Prints

1975 May 27 1:12,125 Black and White InfraredPhoto Prints

1988 August 16 1:20,000 Red InfraredScanned Photo

1995 April 1:20,000 Black and WhiteDigital Orthophoto

1999 September 22 1:6,000 ColorPhoto Prints

TABLE 2. Air photography used for analysis atthe Manitowoc site.

Year Date Scale Type and Data form

1938 July 4 1:20,000 Black and WhiteDigital Orthophoto

1952 September 1:20,000 Black and WhiteDigital Orthophoto

1961 October 6 1:20,000 Black and WhiteDigital Orthophoto

1967 August 28 1:20,000 Black and WhiteDigital Orthophoto

1975 June 2 1:12,125 Black and WhitePhoto Prints

1992 May 1:20,000 Black and WhiteDigital Orthophoto

1999 September 22 1:6,000 ColorPhoto Prints

310 Brown et al.

sition with a 2-foot contour interval topographicmap in the form of an Arcview™ shape file. Ar-cview™ and ArcInfo™ were used to map and ana-lyze the digital orthophotos. Digital polylines werecreated representing bluff crest and bluff toe linesfor each of the dates for which aerial photographywas taken. The original airphotos were viewed instereo beside the computer to confirm the toe andcrest locations using a three-dimensional view. De-tails of the procedure are given in Brown (2000).The recession measurements represent spatial aver-ages of rates measured at increments of 10–-20 malong the shoreline over a distance of about500–700 m for each of the measuring points. Thecenters of each of the profiles coincide with theshoreward extension of the measured offshore ba-thymetric profile locations. A script tool within ArcView™ developed by Joon Heo (pers. comm.2002) enables the average recession rate over alength of shoreline to be computed. The script pro-gram calculates the distance from the baseline totwo digitized historic bluff lines. The user specifiesthe number of equally spaced intervals along theshore to use for recession rate calculation. Thescript tool calculates the average recession rate bytaking an average of the difference in the distancesdetermined from the baseline to the two offset blufflines (Kruepke 2000).

Precipitation

Over periods of years, precipitation affects thelake level. Over periods of hours, days, and weeksprecipitation can increase groundwater levels andalter surface runoff. It is likely that periods of in-tense rainfall affect bluff retreat rates (Jibson andOdum 1994). For this study, the average annual pre-cipitation recorded at Milwaukee over each erosionepoch was compared to the average rate of bluff re-treat.

Water Level

Changes in bluff retreat rates commonly havebeen correlated with fluctuations in lake level (Jib-son and Odum 1994). The average, minimum, max-imum, and range of lake levels within each erosionepoch are compared to the recession magnitude ofeach erosion epoch. Historical lake level informa-tion from 1860 to present are on the NOAA websitein graphical form (NOAA) and digital monthly lakelevels dating from 1860 to present were obtainedfrom NOAA (Frank Quinn, pers. comm., 1999).

Wind

Wind data were obtained from two sources. TheUSACE (Hubertz et al. 1991) study used the histor-ical continuous wind observations made at sevenland stations and two anemometer-equipped buoysto estimate wind fields over Lake Michiganthroughout the period 1956 to 1997. Due to the lackof hourly data at all stations, the wind data weresampled every 3 hours (Hubertz et al. 1991). Thesecond wind data set consists of hourly anemometerreadings from the Milwaukee Airport dating backto the early 1900s. Gaps in the data are interpolatedto provide a near-continuous time series. All land-based wind data are corrected to 10 m height aboveground surface using the standard 1/7th power lawfor the wind speed profile (Davenport 1960). Be-fore the hourly land-wind data were applied to theshoreline study sites, a correction that accounts forthe difference between overland air temperature andwater temperature was applied (Schwab and Mor-ton 1984).

Wind Setup

Sustained high winds from one direction canpush the water level up at one end of the lake andmake the water level drop by a correspondingamount at the opposite end. This is referred to aswind setup. Changes in barometric pressure can addto this effect. The wind-induced setup is estimatedfrom a two-dimensional control volume approach(Sorensen 1997). To estimate this effect at the studysites, the fetch lengths associated with all wind di-rections impacting the sites were determined(Brown 2000).

Storm Events

Amin (1991) states that the generation of wavessufficiently large to cause erosion requires thatwind velocity from a constant direction exceedssome minimum value. These conditions may betermed storm events. The frequency of storm activ-ity is likely to influence the rate of bluff erosion.Powers (1958) and Carter (1976) identify storms asa key erosion agent. We interpreted the criteria usedby Amin (1991) and Davidson-Arnott and Pollard(1980) to define a storm event (Brown 2000). The3-hour intervals for which wind data are reportedfrom the ACE data are not considered to be smallenough time steps to examine storm activity. Sincethe Milwaukee wind data are available on an hourlybasis, they were used to determine the number of


storm events that occurred during each erosionepoch.

Wave Data

Wave data analyzed to determine deepwaterwave power and runup are products of the Michi-gan wind-wave hindcasting project conducted bythe Army Corps of Engineers (Hubertz et al. 1991).A numerical wind wave hindcasting model calledDWAVE (Resio and Perrie 1989) was used to simu-late wave growth, dissipation, and propagation indeep water. Wave parameters including waveheight, mean wave direction and wave period werecomputed at a number of virtual “stations” locatedat a selected number of points (Hubertz et al. 1991).

Deepwater Wave Power

Wave power is the wave energy per unit timetransmitted in the direction of wave propagation(Sorensen 1997). Amin (1991) found a significantpositive correlation between toe recession anddeepwater wave power based on weekly recessionmeasurements at peglines along a Lake Erie bluffshoreline site. By applying linear wave theory forsmall amplitude waves, the wave power P can beexpressed as the product of the wave group veloc-ity, Cg and wave energy density, E

—= ρgH2/8, where

H is the wave height, ρ and g are the density ofwater and gravitational constant, respectively. Inthis study, the deepwater wave power for all wavesdirected toward the shoreline at the study sites wasdetermined for each 3-hour time step using waveparameters from the WIS hindcast data set. An av-erage, minimum, and maximum deepwater wavepower was calculated for each of the erosion epochsfor which hindcast wave data are available (Brown2000). Since the two shoreline reaches under con-sideration are relatively short, variations in waveenergy among the profile locations due to wave re-fraction are expected to be negligible. The deepwa-ter wave height, therefore, is taken as an availablealternative to the wave energy equivalent at theshore.

Bathymetry

Beach and nearshore bathymetric profiles mea-sured at the study sites were used for the determina-tion of the elevation at the toe of the bluff and thecalculation of runup magnitude. These were pro-vided by Guy Meadows (pers. comm., 2000). Pro-files extend from the base of the bluff to 450 m or

more offshore and water depths between 5.5 and6.0 m.

Wave Runup and Wave-Impact Height

Once waves propagate from offshore tonearshore, energy may dissipate due to bottom fric-tion. The remainder of the energy is expressed byits ability to run up the face of the beach and possi-bly partially up the bluff slope. Runup is defined asthe maximum elevation above the still water levelto which the water from the breaking wave rises onthe beach. The runup magnitude is dependent onthe wave height and period of incident deepwaterwaves, the surface slope and profile of the shoreand the nearshore, the toe depth, and the roughnessand the permeability of the slope face.

Because the available nearshore slope informa-tion is much more detailed where soundings havebeen carried out, the bathymetric profiles were usedfor the runup calculations. Grain size of material inthe beach and swash zone were not available, so theequations by Ahrens and Seelig (1996) could not beapplied. Instead, the Army Corps of Engineers soft-ware ACES™ was used for calculating runup.Changes in beach width, beach volume, and sedi-ment type, all of which are known to vary some-what across the sites, were not considered in thisstudy. We assume beach and nearshore profileshave been similar in the past to what was measuredin summer of 1999 by Guy Meadows (pers. comm.,1999). The beach slope in the swash zone is aninput to the runup calculation. The runup is used,together with the lake level and base-of-bluff eleva-tion, to calculate the wave-impact height. Theshoreline orientation is considered for the computa-tion of storm activity, deepwater wave power andwave-impact height. Only waves and winds di-rected toward the shoreline when it was ice-freewere included in the storm frequency, wave power,and runup calculations.

Wave-impact height is defined as the heightabove the top of the beach to which the wave runupreaches. Wave-impact height is calculated using thefollowing relationship

Wave-Impact Height = {Still Lake Level} +{Wind Setup}+ {Runup} – {Base of Bluff Elevation}.

The wave–impact height was determined for each3-hour increment for which wave data were avail-able, from 1956 to 1997. Our hypothesis is that thegreater the incidence of high wave-impact heights,

312 Brown et al.

the more recession there is likely to have been. Tobetter quantify the amount of extreme wave activityon a temporal scale finer than the duration of eachepoch, the maximum wave-impact height for eachmonth of the entire hindcast period was computed.Annual averages based on these monthly peakswere then determined, which is relevant to the num-ber of storm events discussed in next section.

RESULTS

Recession Rates at the Two Sites

Tables 3 and 4 show time-interval recession ratesof bluff top and bluff toe for the two sites. At the

Ozaukee site (Table 3) the 1941 orthophoto only in-cludes profiles WPW1, 2, and 3 because the 1941aerial photographs did not extend far enough northto include two of the profiles (WPW-4 and WPW-5). Profile WPW-5 was not included in any of crestline mapping with the orthophotos created becausethe crest was not discernible on the airphotos whenviewed either in stereo or on the digital imageviewed on the computer screen because of densetree cover (Brown 2000).

Recession rates were measured at the crest, butnot at the toe of the bluff at the Manitowoc site.The Manitowoc bluffs are much lower and steeperthan at the Ozaukee site, so the toe was extremelydifficult to discern on ortho-rectified aerial images.Because the bluffs at the Manitowoc site have re-treated in a more or less parallel manner since the1970s (Mickelson et al. 1977, Chapman et al.1997a), the crest recession rate should reflect therecession rate at the toe. Any time lag between thewaves eroding the base of the bluff and the re-sponse at the crest, is expected to be significantlyshorter than the duration of the epochs considered.The results of crest-recession-rate measurements atthe Manitowoc site are summarized in Table 4.

Recession Rate and Precipitation

Annual precipitation rates and the average pre-cipitation rates per epoch are shown with the crest-recession rates for the Ozaukee site in Figure 3.There was a steady increase in the average precipi-tation starting after the 1956–64 period. With theexception of WPW-4, this appears to match thetrend of increasing crest-recession rates after the1964–75 period. However, the crest-recession ratevalues are relatively low (between 0 and 0.5 m/yr)and the change in precipitation over the span ofdecades does not vary appreciably. The precipita-

TABLE 3. Bluff-crest (top number) and bluff-toe(bottom number) recession rates at the Ozaukeesite.

WPW-1 WPW-2 WPW-3 WPW-4 WPW-5Profile [m/yr] [m/yr] [m/yr] [m/yr] [m/yr]

1941–56 0.32 0.22 0.47 * *0.36 0.54 0.44 * *

1956–64 0.02 0.24 –0.03 –0.21 *–0.01 0.06 0.30 0.07 0.11

1964–75 –0.05 0.08 –0.05 0.51 *2.19 1.78 0.97 1.31 1.11

1975–88 0.71 0.46 0.44 0.32 *1.50 0.75 0.19 0.44 0.36

1988–95 0.66 0.48 0.69 0.45 *–0.35 –0.14 0.34 0.13 0.94

Note: The positive rates indicate bluff retreat. The nega-tive toe recession rates indicate a near-zero toe retreat ora net deposition at the toe over the course of the erosionepoch. Negative crest recession rates are within the esti-mated error (Brown 2000) and are assumed to be zero.

TABLE 4. Bluff-crest recession rates at the Manitowoc site.

WTR-1 WTR-1-2 WTR-1-3 WTR-1-4 WTR-1-5Profile [m/yr] [m/yr] [m/yr] [m/yr] [m/yr]

1938–52 0.77 1.02 0.22 0.17 0.131952–61 1.40 1.27 0.95 0.76 0.491961–67 –0.09 –0.06 –0.23 –0.17 –0.051967–75 1.19 1.23 0.31 1.20 1.071975–92 1.03 1.12 0.72 0.45 0.291992–99 0.50 –0.21 0.54 0.57 0.56

Note: The positive rates indicate bluff retreat. Negative crest recession rates are within theestimated error (Brown 2000) and are assumed to be zero.


tion magnitudes on the scale of years, however,show some notable changes, particularly the maxi-mum yearly value of over 1,000 mm in 1959.

The precipitation rate and crest-recession rateover time for the Manitowoc site are shown in Fig-ure 4. The overall minimum of annualized precipi-tation averaged over the erosion epochs occurredduring the 1961–67 epoch. This is the same periodduring which there was virtually no crest recession.Higher recession rates occurred before and afterthis low-erosion epoch. Slightly increased precipi-tation rates also occurred before and after the low-precipitation epoch. However, the averageprecipitation and recession rate trends over time donot entirely agree. The epoch-averaged precipita-tion rates show a local maximum in the earliestepoch (1938–52), whereas recession rates for theearliest epoch were lower than the following(1952–61) epoch. The 1938–52 erosion epoch onlyoverlaps with precipitation data by 19%, so the av-erage precipitation rate may not be representative ofthe entire epoch.

Recession Rate and Lake Level

Bluff crest-recession rate is likely to be influ-enced by lake level at sites where the lag time be-tween undercutting at the base and the response inthe form of failure at the crest is short. At the Ozau-kee site, the response of the crest to the undercut-ting at the toe is expected to be on the order of 50to 100 years (Mickelson et al. 2004). In contrast,the crest of the lower bluff at the Manitowoc siteprobably responds to wave erosion at the toe inmonths, or at most, a few years. Figure 5 shows thecrest-recession rates at the Ozaukee site. Whilethere was little recession during the low water-levelperiod of the 1960s, the recession rate did not de-finitively rise thereafter as did the level of the lake.A slight rise in crest-recession rate took place be-tween the 1964–75 and 1975–88 epochs. This wascoincident with a further rise in lake level. How-ever, crest recession maintained its rate of about 0.5m per year through the final erosion epoch while

FIG. 3. Crest recession rate and precipitationrate at the Ozaukee site.

FIG. 4. Crest recession rate and precipitationrate at the Manitowoc site.

314 Brown et al.

FIG. 5. Crest recession rate and lake level at theOzaukee site. FIG. 6. Toe recession rate and lake level at the

Ozaukee site.

lake level dropped significantly, likely because ofthe long response time of high bluffs.

The bluff-toe recession rate at the Ozaukee siteand Lake Michigan level over time are shown inFigure 6. A reduced recession rate coincided with adrop in average lake level from the first epoch(1941–56) to the second epoch (1956–64). Themaximum recession rates recorded for this studyoccurred during the 1964–75 epoch. While thisdoes not represent the maximum epoch-based aver-age lake level, it represents a time period that expe-rienced the most dramatic rise in lake level. LakeMichigan rose by more than a meter during thisepoch. Following the 1964–1975 period, the reces-sion rates for nearly all the profiles continued to fallthrough to the most recent epoch considered(1988–95). During the 1975–88 epoch, a time dur-ing which the recession rates had already begun tofall, the average lake level over the scale of epochsreached a maximum.

The recession rate of the bluff crest at the Mani-towoc site is shown together with historic lake levelin Figure 7. The crest experienced equally high

(1–1.5 m/yr) recession rates before and after the1961–67 epoch for profiles WTR-1 and WTR-2.For all the profiles, the low-erosion period of1961–67 corresponds with a low lake-level period.The lake-level high after the 1961–67 period was ofgreater magnitude and duration than the lake-levelhigh before this. Profiles WTR-4 and WTR-5 hadless recession per year before the low lake level pe-riod than after (i.e., good agreement with the lakelevel trends). The bluffs at WTR-3 show more re-cession per year before the low lake level periodthan after (i.e., poor agreement with the lake leveltrends). The variations in lake level, therefore, arematched extremely well by changes in crest-reces-sion rates of 2 of the 5 profiles (WTR-4 and WTR-5), moderately well by crest-recession changes in 2profiles (WTR-1 and WTR-2) and moderatelypoorly in profile WTR-3.

Recession Rate and Storm Events

The number of storms per year is shown on anannual basis and on an epoch-averaged basis for the


Ozaukee site in Figure 8. The crest and toe-reces-sion rates are shown on these figures for one repre-sentative profile. The highest average rate of stormsoccurred during the most recent erosion epoch(1988–95), a time during which the crest-recessionrate was at or near its highest level. However, thecrest recession rate magnitudes were not greaterthan 0.7m/year and were generally between 0 and0.5m/year across all the erosion epochs. A numberof consecutive years of elevated storm activity oc-curred in the mid-1960s and the early 1970s. Bothof these periods are associated with an epoch overwhich high toe-recession rates were measured forthe Ozaukee site. Toe-recession data are availableonly for the Ozaukee site. For four of the five pro-files, the lowest bluff-toe recession rate at theOzaukee site occurred during the most recent epoch(Brown 2000). Indeed, the recession rates for thesefour profiles are negative numbers, potentially dueto the presence of mass wasting material buildingup at the toe of the bluff that had not yet been re-moved by wave action.

At the Manitowoc site (Fig. 9) a relatively largeaverage number of storms per year, on the timescale of the erosion epochs, occurred during the pe-riod (1961–67) when the least recession took place.The maximum average number of storms per yeartook place during the most recent epoch, 1992–99.At the time scales being considered, crest-recessionrate does not appear to have been strongly influ-enced by storm frequency alone at either of thestudy sites.

Recession Rate and Deepwater Wave Power

The trend of deepwater wave power over time isshown in Figures 8 and 9 for the Ozaukee and Man-itowoc sites respectively. The deepwater wavepower reached its peak during the 1964-75 epoch, atime when the highest recession rate occurred at thetoe at the Ozaukee site. The annual mean deepwaterwave power stayed at high levels until 1980. Before

FIG. 7. Crest recession rate and lake level at theManitowoc site.

FIG. 8. Lake level, storm count, precipitation,wave power and representative toe and crest reces-sion rates at profile WPW-2 at the Ozaukee site.

316 Brown et al.

and after the high values exhibited between 1970and 1980, the deepwater wave power seems to havevaried about a relatively constant mean value. Theconsistently high annual mean deepwater wavepower between 1970 and 1980 is also observed atthe Manitowoc site (Fig. 9). The wave power roseand stayed high beginning in the 1967–75 erosionepoch, a time at which the highest recession ratewas observed for 3 of the 5 profiles (WTR-3 toWTR-5). For the remaining two profiles (WTR-1and WTR-2), the maximum recession rates werenot attained between 1967 and 1975, but the mea-sured rates still showed a dramatic increase from1961–67 period to 1967–75 period.

Recession Rate and Wave Impact Height

The annual average of maximum monthly wave-impact heights is shown together with the toe-reces-

sion rate for the Ozaukee site in Figure 10. The an-nual average of peak monthly wave impact heightsvaries from 2.5 m to 5.0 m. A large jump in peakwave-impact heights occurred during the epoch as-sociated with a dramatic rise in toe-recession rates(1964–75) at the Ozaukee site. The small differ-ences in beach slope and bluff toe elevations for thedifferent profiles account for differences in magni-tude of the wave-impact heights, but do not appearto alter the trends in wave-impact height over time.

The trend of annual average peak monthly maxi-mum wave-impact height varies greatly over timeand compares more favorably to the trends of reces-sion rate than do the epoch minimum, maximum ormean wave-impact height Brown (2000). The an-nual average of peak monthly wave-impact heightsand crest-recession rate at the Manitowoc site areshown in Figure 11. The annual average of peakmonthly wave impact varied from 1.75 m to 4.5 m.At the Manitowoc site, the wave-impact heightsrose dramatically early in the 1967–75 epoch, thetime period during which the highest recession rateoccurred.

DISCUSSION

One of the main objectives of this study was tofind empirical relationships between bluff recessionrates and one or more factors that appear to influ-ence them. The influence of storm counts and pre-cipitation magnitude pale in comparison to theeffect that lake level and wave-impact height havehad on recession rate of the bluffs. By applying thestorm definitions used in this study (Brown 2000)to the Milwaukee wind data we have determinedthat there is approximately 1 storm every 3 daysbased on annual rates. Since we know storms do notoccur this frequently, our storm definition is clearlynot isolating the extreme events that are capable oferoding bluff material. A more restrictive storm de-finition would yield a storm count over each epochthat might relate better to observed recession rates.Since the epoch-averaged precipitation trend is sosimilar to that of the lake level, it is difficult to de-couple the effect of precipitation from the effectthat lake level has on the recession rate.

Lake level has an unquestionable influence onthe toe-recession rate of the high bluffs at the Ozau-kee site and the crest-recession rate of the lowbluffs at the Manitowoc site. They show very simi-lar relationships to the epoch-averaged lake levels(Figs. 6 and 7). Recession of the crest of the Ozau-kee bluffs shows a relatively poor relationship with

FIG. 9. Lake level, storm count, precipitation,wave power, and crest recession rate for profileWTR-4 at the Manitowoc site.


lake level trends because consistently low recessionrates are observed throughout the time period ofrecord. This is likely because the 5 to 20 year pe-riod of this study is shorter than the cycle time forwaves to erode the base of the high bluffs suffi-ciently to cause a failure that extends to the crest asshown in Figure 2. While recession rates at the toeof the Manitowoc bluffs were not quantified for thisstudy, visual observations made of these low bluffsover the past 25 years suggest that the retreat at thecrest follows closely behind the retreat at the toeand the slopes of the bluff doesn’t change apprecia-bly. These data suggest that the bluff failure cycletime at the Manitowoc site is considerably shorterthan the 5-15 year intervals considered for thisstudy. The difference in recession rate between thetoe and the crest at the high-bluff site and the paral-lel retreat at the low-bluff site has been observedfor a number of years (Mickelson et al. 2004).

Berg and Collinson (1976) argued that high lakelevels are a primary cause of bluff erosion. Blufftop recession of bluffs varying in height between 6m and 27 m was considered for their study. Theysuggested that bluff top recession rate rises a fewyears after lake level rises, but the elevated rate canbe maintained or even accelerated as the lake leveldecreases from a high level. According to Berg andCollinson, as lake level rises, well-developedbeaches delay the onset of maximum erosion ratesbecause it takes time to deplete the beach. They at-tributed the observed time lag between a fall in lakelevel and a decreased erosion rate to the time re-quired to vegetate the scarified bluffs. Other possi-bilities to explain the time lag between water levelfall and decreased bluff top recession rate may bedue to groundwater destabilization (Montgomery1998), or simply that the upper bluff slope re-mained steep and unstable.

We find that peak bluff-toe recession rates, mea-sured over the scale of decades, actually precedethe maximum epoch-averaged lake level for theOzaukee site. The maximum toe-recession rate oc-curs during an epoch (1964–75) that experiencesthe most dramatic rise in lake level. The Manitowocsite bluff-crest recession rates generally follow thepattern of average lake level with no temporal shiftin either direction. The recession intervals consid-ered for the Berg and Collinson (1976) study wereshort enough (1-year intervals) to observe a timedelay before erosion rates increased after a rise inlake level, and a sustained high erosion rate wassustained while lake level fell. These time lags ob-served by Berg and Collinson are on the order of

3–4 years long. The erosion epochs considered forthis study (5–15 year intervals) are too long to mea-sure the delay in the onset of high recession ratedue to high water level or the delay in the decreaseof recession rate following a drop in lake level.

Amin (1991) observed a strong correlation be-tween deepwater wave power and bluff-toe reces-sion rates. Amin collected recession data bymeasuring exposed horizontal peg lengths everyone or two weeks over a period of two years. Datafrom the present study do not show nearly as stronga relationship between deepwater wave power andrecession rate as Amin’s study. The time resolutionused by Amin was much finer and the amount of re-cession data generated was much greater than thepresent study. Accurate recession-rate measure-ments recorded on the scale of weeks may be thekey requirement to establish a strong statistical cor-relation between recession rate and wave power.However, a study such as Amin’s that only spans atotal of 2 years’ time cannot consider the influenceof lake level, since the cycle time of lake level vari-ation is on the scale of 10–15 years for the GreatLakes.

Another important factor may be the beachwidth. Since the bluffs at Amin’s site had very littleprotective beach, shoreward waves were regularlyreaching the base and the face of the bluffs. Thetwo Wisconsin bluff sites adopted for this studyhave relatively wide beaches. During low lake levelperiods, the waves rarely reach the bluffs and whenthe lake level rises the bluff is reached more fre-quently. The reduced frequency with which thewaves impact the bluffs at the Wisconsin study sitesmay be part of the reason for the relatively pooragreement with deepwater wave power and bluff re-cession.

Davis et al. (1973) suggest that lake level merelyplays a passive role by providing the appropriateconditions for bluff recession to occur. The actualremoval of material, they contend, is due in part tohigh waves associated with intense storms. In an at-tempt to better predict recession rate, they introduceconsideration of the annual average of maximummonthly wave impact heights. Figure 11 (Mani-towoc site) shows that when lake level reaches apeak value, the crest recession is at or near its high-est level, and the wave-impact height rises dramati-cally during the 1967–75 epoch. Because of thecoincidence of these and the time intervals availableat the Manitowoc site, the results for the Manitowocsite do not allow us to interpret a difference betweenthe effectiveness oflake level and wave-impact

318 Brown et al.

height as predictors of the rate of bluff erosion. Onthe other hand, the Ozaukee site (Fig. 10) shows aclear, sharp rise in wave-impact height and a peak inthe toe-recession rate during the 1964–75 epoch, atime period over which the epoch-averaged lakelevel did not reach a maximum (although the lakelevel shows a dramatic rise over this epoch). Thewave impact height shows considerable promise forexplaining trends in bluff-recession rates over time.Because the wave-impact height combines the ef-

fects of lake level and waves, it can be expected toshow a better agreement with bluff-recession rateover time than lake level alone, and its usefulnessshould be tested on other shorelines where shorttime-interval recession data are available.

CONCLUSIONS

1. There is not a strong relationship betweendeepwater wave power alone and recession rate

FIG. 10. Average annual monthly peak wave impact height and toe reces-sion at the Ozaukee site.


over the time scales considered at the two studysites.

2. Storm frequency alone has little impact on re-cession rates over the time intervals studied.

3. Because the epoch-averaged precipitationtrend is so similar to that of the lake level, it is dif-ficult to decouple the effect of precipitation fromthe effect that lake level has on the recession rate.

4. Lake level appears to have a major influenceon recession rate. Both the toe-recession rate of thehigh bluffs at the Ozaukee site and the crest-reces-sion rate of the low bluffs at the Manitowoc site

show remarkably similar patterns over time to theepoch-averaged lake levels.

5. The crest-recession rate trends observed at thehigh bluffs of the Ozaukee site do not correspondwell with lake level changes, but the toe–recessionrate shows trends similar to that of lake level overtime. This is because the 5-to-20-year period of thisstudy is shorter than the cycle time for waves toerode the base of the high bluffs sufficiently tocause a failure that extends to the crest (probably 50to 100 years).

6. While recession rates at the toe of the Mani-

FIG. 11. Average annual monthly peak wave impact height and crestrecession at the Manitowoc site.

320 Brown et al.

towoc bluffs were not quantified for this study, vi-sual observations made of these low bluffs over thepast 25 years suggest that the retreat at the crest fol-lows closely behind the retreat at the toe. Thesedata suggest that the bluff failure cycle time at theManitowoc site is considerably shorter than the 5-20 year intervals considered for this study. Signifi-cant correlation of crest recession rates with lakelevel changes at this site provides support for thisinference.

7. Peak recession rates, measured over the scaleof decades, actually preceded the maximum epoch-averaged lake level for the bluff toe at the Ozaukeesite. The maximum toe-recession rate occurred dur-ing an epoch (1964–75) that experienced the mostdramatic rise in lake level. The Manitowoc sitebluff-crest recession rates generally follow the pat-tern of average lake level with no temporal shift ineither direction.

8. The present study introduces consideration ofthe annual average of maximum monthly wave im-pact heights. This is a composite factor that showsconsiderable promise for explaining trends in bluff-recession rates over time. The wave-impact heightcombines the lake level with the wave effects andshows a promising relationship to bluff-recessionrate over time. Our results suggest that wave-impactheight may be more strongly linked to recessionrate than lake level alone.

ACKNOWLEDGMENTS

We wish to thank Mr. Phillip Keillor and Dr.Dave Hart of the University of Wisconsin SeaGrant. Dr. Alberto Vargas of Wisconsin CoastalManagement provided ideas, advice, and DEM andDOQQ files and airphotos. Dr. Frank Scarpace andseveral others at the University of Wisconsin Envi-ronmental Remote Sensing Center, the creator ofOrthomapper™, provided technical advice. Mr.Allan Luloff and Dr. Richard Gleason provided ad-ditional recession data. This study was funded bythe Great Lakes Protection Fund Grant 470, Univer-sity of Wisconsin fund number 133-CK86.

REFERENCESAhrens, J.P., and Seelig, W.N. 1996. Wave Runup on

Beaches, In Proc. Internat. Conf. on Coastal Engi-neering, pp. 981–992. Orlando, Florida, AmericanSociety of Civil Engineers.

Amin, S.M.N. 1991. Bluff toe erosion along a section ofthe Lake Erie South shore, Ph.D. thesis, Kent StateUniversity, Kent, Ohio, U.S.A. Kent, Ohio.

Berg, R.C., and Collinson, C. 1976. Bluff erosion, reces-sion rates and volumetric losses on the Lake Michiganshore in Illinois. IL State Geol. Sur. EnvironmentalGeology Notes 76.

Bray, M.J. and Hooke, J.M. 1997. Prediction of Soft-Cliff Retreat with Accelerating Sea Level Rise. J.Coastal Res. 13(2):453–467.

Brown, E.A. 2000, Influence of water level, wave cli-mate, and weather on coastal recession along a GreatLakes shoreline. M.S. thesis, University of Wisconsin,Madison, WI.

Carter, C.H. 1976. Lake shore erosion, Lake County,Ohio: Setting, processes, and recession rates from1876–1973. Ohio Div. Geol. Surv. Report No. 99.

Chapman, J.A., Edil, T.B., Mickelson, D.M., and BayLake Staff 1997a. Lake Michigan shoreline recessionand bluff stability in northeastern Wisconsin:1996.Bay-Lake Regional Planning Comm. Tech. Rpt.

———, Edil, T.B., and Mickelson, D.M. 1997b. LakeMichigan shoreline recession and bluff stability insoutheastern Wisconsin: 1995. Southeast WisconsinRegional Planning Comm. Tech. Rpt. 36.

Chase, R.B., Kehew, A.E., and Montgomery, W.W.2001a. Determination of slope displacement mecha-nism and causes: an approach using new geometricmodeling techniques and climate data. In LandscapeErosion and Evolution Modeling, Ed. R.S. Harmonand W.W Doe III, pp. 57–87. New York: KluwerAcademic.

———, Kehew, A.E., and Montgomery, W.W., 2001b.Determination of kinematics of slope movementsusing low-cost monitoring and cross-section balanc-ing. Envir. and Eng. Geosci. VII (2):193–203.

Davenport, A.G. 1960. Rationale for determining designwind velocities. Structural Division J., Am. Soc. CivilEngineers, 86:39–68.

Davidson-Arnott, R.G.D., and Pollard, W.H. 1980. Waveclimate and potential longshore sediment transportpatterns, Nottawasaga Bay, Ontario. J. Great LakesRes. 6:54–67.

Davis, R.A., Seibel, E., and Fox, W.T. 1973. Coastalerosion in eastern Lake Michigan—causes andeffects. In Proc. 16th Conf. Great Lakes Res., pp.404–412. Internat. Assoc. Great Lakes Res.

Dewberry and Davis 1994. Basic analyses of waveaction and erosion with extreme floods on GreatLakes shores. Draft Report for Federal EmergencyManagement Agency, Mitigation Directorate,National Flood Insurance. Prepared by Dewberry &Davis, Program Management Engineering and Tech-nical Services Division, Fairfax, Virginia.

Edil, T.B., and Mickelson, D.M. 1995, Overconsolidatedtills in eastern Wisconsin. Nat. Res. Council, Trans-portation Res. Bd. Record no. 1479: 99–106.

———, and Vallejo, L.E. 1980. Mechanics of coastallandslides and the influence of slope parameters. Eng.Geol. 16:83–96.


Hubertz, J.M., Driver, D.B., and Reinhard, R.D. 1991.Hindcast wave information for the Great Lakes. LakeMichigan, Wave Information Studies (WIS) of U.S.Coastlines Rpt. 24, U.S. Army Corps of Engineers,Washington, D.C.

Jibson, R.W., and Odum, J.K. 1994. Rates and processesof bluff recession along the Lake Michigan shorelinein Illinois, J. Great Lakes Res. 20:135–152.

Kamphuis, J.W. 1987. Recession rate of glacial tillbluffs. J. of Waterway, Port, Coastal and Ocean Eng.113:60–73.

Kirk, R.M., Komar, P.D., Allan, J.D., and Stephenson,W.J. 2000. Shoreline erosion on Lake Hawea, NewZealand, caused by high lake levels and storm-waverunup. J. Coastal Res. 16:346–356.

Kruepke, J.J. 2000. Understanding and monitoringcoastal bluff erosion using aerial photography, a casestudy: western Lake Michigan shoreline, Two Creeksshoreline, Two Creeks, Wisconsin. M.S. thesis, Univ.Wisconsin, Madison, Wisconsin.

Mickelson, D.M., Acomb, L., Brouwer, N., Edil, T.B.,Fricke, C., Haas, B., Hadley, D., Hess, C., Klauk, R.,Lasca, N., and Schneider, A.F. 1977. Shoreline erosionand bluff stability along Lake Michigan and Lake Supe-rior shorelines of Wisconsin: Shore Erosion Tech. Rpt.,Coastal Management Program, Madison, Wisconsin,U.S.A.

———, Edil, T.B., and Guy, D.E. 2004. Erosion ofcoastal bluffs in the Great Lakes. In Formation, Evo-lution, and Stability of Coastal Cliffs—Status andTrends , ed. M. Hampton and G.B. Griggs, pp.107–123. U.S. Geol. Surv. Prof. Paper 1693.

Montgomery, W.W. 1998. Groundwater hydraulics andslope instability analysis: elements for predictionshoreline recession. Ph.D. dissertation, WesternMichigan University, Kalamazoo, Michigan.

Pope, J., Stewart, C., Dolan, R., Peatross, J., and Thomp-son, C., 2001, The Great Lakes shoreline type, ero-

sion, and accretion. U.S. Army Corps of Eng., Open-File Map.

Powers, W.E. 1958. Geomorphology of the Lake Michi-gan Shoreline. Final Report, Northwestern University,College of Liberal Arts, Dept. of Geography.

Resio, D., and Perrie, W. 1989. Implications of an F4equilibrium range for wind generated waves, J. Phys.Oceanography 19 (2):193–204.

Schwab, D.J., and Morton, J.A. 1984. Estimation ofoverlake wind speed from overland wind speed: Acomparison of three methods. J. Great Lakes Res.10:68–72.

Sorensen, R.M. 1997. Basic Coastal Engineering. NewYork: Chapman & Hall.

Sterrett, R.J. 1980. Factors and Mechanics of Bluff Ero-sion on Wisconsin’s Great Lakes Shorelines. Ph.D.thesis University of Wisconsin, Madison, Wisconsin.

Sunamura, T. 1976. Feedback Relationship in Wave Ero-sion of Laboratory Rocky Coast. J. Geol. 84:427–437.

———. 1977. A Relationship between Wave-InducedCliff Erosion and Erosive Force of Waves. J. Geol.85:613–618.

Wilcock, P.R., Miller, D.S., Shea, R.H., and Kerkin, R.T.1998. Frequency of Effective Wave Activity and theRecession of Coastal Bluffs: Calvert Cliffs, Maryland.J. Coastal Res. 14(1): 256–268.

Wu, C.H., and Nepf, H.M. 2002. Breaking wave criteriaand energy losses for three-dimensional breakingwaves. Journal of Geophysical Research—Oceans,C10, 3177, 2001JC001077, 41:1–18.

Yao, A., and Wu, C.H. 2004. Energy dissipation ofunsteady wave breaking on currents, J. PhysicalOceanography 34(N10):2288–2304.

Submitted: 8 May 2004Accepted: 6 June 2005Editorial handling: Gerald Matisoff

factors controlling rates of bluff recession at two …davem/abstracts/05-8.pdffactors controlling...

Documents