Collection of New Recession Rate Data
In 1994 the U.S. Army Corps of Engineers - Detroit District prepared a comprehensive kilometer-by-kilometer database of recession rate data for all of the U.S. Great Lakes shorelines including Lake Michigan (Stewart, 1994). This database provides mean, median, maximum and minimum recession rates for each kilometer reach and was based on all available shoreline recession rate data as of March 1994.
In order to utilize this as a baseline recession rate database for the 1997 Lake Michigan Potential Damages Study, any new recession rate data (i.e., anything updated, calculated or created subsequent to the 1994 database) for the Lake Michigan shoreline needed to be incorporated to this database.
To acquire this new data, inquiries were made with various individuals and State agencies. New recession rate data was acquired primarily for the Michigan and Wisconsin shorelines through the following data sources:
- State of Michigan, Department of Environmental Quality: New or updated data was obtained for the majority of counties (excluding Berrien), along the eastern shoreline of Lake Michigan;
- Wisconsin Recession Rate Study - The State of Wisconsin contracted S.E.H. and Baker Engineering to develop a new recession rate methodology for the Wisconsin coast. Three 10-mile test sites in three counties (Manitowoc, Ozaukee, Racine) were completed and "new" recession rate data was determined for these sites;
- SEWRPC Lake Michigan Recession and Bluff Stability in Southeastern Wisconsin - This 1997 report by SEWRPC provides updated recession rate data for Kenosha, Racine, Milwaukee, and Ozaukee Counties. Calculations are provided for the periods 1963-1995, 1970-1995 and 1975-1995.
- Bay/Lakes Regional Planning Commission, Lake Michigan Recession and Bluff Stability in Southeastern Wisconsin - This is a report similar in nature to the above and covers some of the northeastern counties including Sheboygan, Manitowoc, Kewaunee and Door. Calculations are provided for the period 1978-1992.
Detailed descriptions of the new data sets are found in Stewart (1997). The comprehensive updated database is also available as an Appendix to the Stewart (1997) report. These may both be obtained from USACE Detroit District.
Creation Of A Single-Value Recession Rate Database, Lake Michigan Shoreline
As a first step in the analysis of recession rates and land use for the Lake Michigan Potential damages study, the revised recession rate database created above had to be reviewed in order to select a single, most appropriate recession rate value for each kilometer reach along the shoreline.
To a degree, this exercise was performed during the creation of a recession rate data set that was prepared for the USGS and University of Virginia during the creation of a yet unpublished map of Great Lakes erosion and accretion (Stewart et al., 1997). However, the logic and assumptions used in creating this data set were never explicitly recorded, and since then, changes in our understanding of recession and the use of recession rate data in the establishment of hazard lands along the shoreline have occurred. As such, a review of the data was again necessary to establish the best value of recession for each reach.
For the USGS exercise described above a number of general assumptions were used in selecting the best value of recession for each reach. Generally speaking, the best value of recession was that which:
- Had the longest period of record;
- Was the most accurate in terms of it?s "Confidence" ranking (this usually correlated to the method of calculation used);
- Had a fair number of data records within the reach;
- Was used by the State in the determination of formal erosion hazard setbacks; and
- Was the only estimate of recession available for the reach
In re-visiting the database for this project, we had to consider a number of other issues. For example, many agencies (State of Ohio for example, Mr. Scudder Mackey, personal communication) are beginning to use shorter term recession rate information to establish setbacks as they feel it is more representative of coastal and development conditions that have existed in the recent past. Historic data (e.g., comparing circa 1800 shorelines to 1990s shorelines) may be less representative of reality than comparing 1960s or 1970s shorelines with 1990s shorelines. Given this, we looked for such short-term data and where appropriate given other factors such as its confidence, number of data points, etc., selected it for use here.
Second, some of the State CZM recession rate data used in determining single value recession rates for the USGS project were fairly low in the confidence ranking. Where these occurred and where other more confident data existed, the more confident data was substituted.
Third, an assumption was made in the original database that recession rates of all bedrock shorelines was zero, even if the shoreline was classified as "non-resistant" bedrock. Recent discussions as part of the shoreline classification exercise have indicated that the erosion of soft bedrock areas can be a critical component of profile retreat in these areas. In light of this, the classification information for the Lake Michigan shoreline was consulted and where "non-resistant" bedrock was identified, the recession rate was changed from "0.00" to "no data." This occurred primarily in portions of Emmett County as well as in areas of the Upper Peninsula.
Fourth, the new data sets entered into the master database (described above) were used in many places as new "best" values for many reaches. This is particularly true of the new Michigan DNR and SEWRPC (1997) data in Michigan and Wisconsin respectively. Despite it?s lower confidence, some of the SEH and Baker (1997) values were used for some reaches, as they replaced data that was of an even lower confidence, or where data was based on only a single data point.
Once all data had been selected, a single value recession rate database was created. A hardcopy of this database is presented in Appendix 2 of Stewart (1997). Unlike the master database, only mean recession rate values are listed, along with the period of record, source and comments where available.
Short-Term Trends in Water Levels and Recession Rates
This task involved a closer examination of all known recession rate data for Lake Michigan in an attempt to determine relationships between water level trends and recession rates (i.e., can we establish with any accuracy, recession rates during high water periods?, low water periods?, falling water level periods?, rising water level periods?), as well as to determine if recent recession rates have been significantly changed (relative to long-term historic rates), as a result of shoreline modification, development, changes in sediment supply, etc.
Existing recession rate data was examined with reference to periods of higher or lower water levels on the Great Lakes in order to determine if there were any data sets that were calculated over a high water, or low water level period. The theory here was that by pulling these data sets out, we might be able to get a better handle on whether or not the data reflects increases in erosion rates during high water and decreases in erosion rates during low water, a case commonly observed by shore property owners.
In reviewing the data it became apparent that there were generally 2 high water periods that could be extracted. These are loosely defined as the 1970s high water period and the 1980s high water period. No data records existed that were suitable to cover the 1950s high water period. Periods of record within each of these vary, and in some cases cover the time when levels were rising to the peak levels, or only the very short period when levels were at their peak.
Data covering low water level periods is scarce. Only one data set was available covering the 1960s low water period (1955-1967) and it is of low confidence. Other data sets tended to cover the "low" water period between the 1970 peak and the 1980s peaks. Data sets used here have periods of record of: 1973-1984, 1978-1984, and 1975-1980.
For both high water recession and low water recession, available data sets were found for the eastern shoreline of Lake Michigan from about Reach 695 - 940 (Michigan and Indiana shoreline), and for the Wisconsin shoreline from about Reach 1073-1340. This was to be expected, as these are areas that have traditionally had erosion problems, and thus areas where researchers have focussed their efforts.
In order to better visualize the relationship between long term recession rates and short-term trends, a series of simple graphs were created with the data. An example is found on the following page for the section of shoreline between St. Joseph and New Buffalo, Michigan.
Examination of these graphs shows some simple but interesting relationships. In the majority of cases, recession rate data calculated over the high water level periods is significantly higher than the long-term average rate of recession utilized. There are a few cases where this does not hold, and a few others where data calculated by Wood in the mid-80s shows significant increases in accretion during the high water level period. This occurs along portions of the Indiana shoreline and bears further investigation to determine if it is a measurement error, or due to some other factor (e.g., nourishment?).
Where data for both the 1970s and the 1980s is available, 1980s high water recession rates tend to be higher than the 1970s high water recession rates. This may indicate a
direct correlation between water level height and erosion rate, however, there does not appear to be enough data available to really verify this any further.
Where "Low Water" recession rate data is available, it was found to be significantly lower in all cases versus both long-term data and 1970 s or 1980s high water data.
A graphical representation of this nature makes it easy to identify recession "hot spots", whether one is looking at the long-term data or the short-term trend information. It certainly seems to raise a red flag as to those areas that may experience severe problems during a high water period. Granted, the data is a bit biased based on where researchers have conducted studies, but it seems to serve as a good starting point for hazard identification.
Generation of New Recession Rate Data
The objectives of this work include enhancing the completeness of existing recession rate data contained in the Detroit District, USACE databases. This task is focused on filling data gaps in the current record that were identified in the above tasks. New short-term recession rate data is also being generated for critically sensitive shorelines (primarily a series of 8 detailed study sites) for recent periods of high (1970s, 1980s, and 1990s) or low (1960s) Great Lakes water levels. Long-term average annual rates will also be generated for shorelines which were either recently developed or are expected to be developed over the study period. Data from previously undeveloped areas will be important where the erosion contributes significantly to sediment budget in downdrift developed areas.
There are three key activities currently underway for the collection of new recession rate data.
First, Stewart (1997) identified that there is a general lack of recession rate data for the Green Bay portion of the Lake Michigan shoreline (Marinette, Oconto, Brown and Door counties) in Wisconsin. To rectify this and also to develop a GIS methodology for use in conducting future calculations and determinations of recession rates along the Great Lakes, VGI Vision Group International Inc. has been retained to provide the Lake Michigan Potential Damages Study with recession rates for this section of Lake Michigan shoreline and conduct a proof of concept study to demonstrate the applicability of GIS and viewer technology in automating the calculation of recession rates. To accomplish this, shoreline positions for two different time periods are being digitized from available digital raster topographic files and digital Orthophoto files. An application is being developed using ArcView and AVENUE scripting to compute shoreline recession rates from digitized shoreline data. In general, the GIS will allow users to select a reach of shoreline and period of record for which recession rates will be calculated. The selected reach will then be displayed and shore normal (orthogonal) transects will automatically be added at 50 meter intervals. The user may then query the GIS to calculate the annual rate of recession for any of the displayed transects. An additional function will allow users to calculate the mean annual recession rate for an entire reach.
Second, the University of Wisconsin is utilizing softcopy photogrammetric methods to calculate bluff recession rates for two sites along the Lake Michigan shoreline, the first, a two mile stretch of Lake Michigan coastline near Two Rivers, Wisconsin and the second, a two mile stretch of Lake Michigan coastline near St. Joseph, Michigan. The objective of this task is to generate accuracy and cost comparisons of softcopy photogrammetric methods with those of LIDAR and standard photogrammetric analytical stereo plotter techniques for bluff mapping and recession rate calculations. Part of this work will also include researching DEM differences methods to include development of specifications that address resolution differences, resampling techniques, accuracy estimates and error propagation, metadata, file structures, file formats, and file compression. The Contractor will also produce DEM?s of the Michigan site using the Government supplied data and scanned images. An accuracy study of these data will not me made. Using similar data (scanned photographs and control with the same approximate accuracy as supplied for the Michigan site), DEM?s will be produced for the Wisconsin site. An accuracy assessment of these data will be made comparing the results from the "Low Cost Alternative" to the results of the main study. The accuracy of the DEM at the Michigan site will be inferred from this study.
Third, detailed calculations of short-term recession rates were carried out at each of the 8 detailed site studies (see Section 4.3) that have been selected for detailed analysis. This work is being conducted by Baird & Associates and the results will feed directly into the detailed Coastal Process Model that is being developed (see Section 4.4). A key product of this Phase 2 investigation by Baird was the development of a new approach for projecting future recession which incorporates an uncertainty band around the projected future shoreline. The basis of this approach stems from the fact that the calculated spatial variability in recession rate (based on two snap shots of bluff crest position in time) is related to the spatial variability in geologic conditions and to a temporal factor associated with the evolution cycle of bluff failure. Therefore, it is important to utilize this information to create bands of uncertainty around future shoreline positions projected either by conventional extrapolation of historic recession rates or by predictions of future shoreline position using the deterministic coastal processes model COSMOS (see Section 4.4). Baird are working closely with Dave Mickelson of the University of Wisconsin to further develop, test and refine this hypothesis. This key finding has two important implications for the LMPDS system: 1) the projected future shoreline position will have a band of uncertainty which will have a significant influence on damage potential projections; and 2) modifications may be required to the shoreline classification system based on the findings of Mickelson to allow lakewide estimates of the uncertainty band width around future projections of shoreline (or bluff crest) position based on bluff type classification.
It is anticipated that the generation of all new recession rate data as described above will be completed by the end of 1998.
Click to return to the GIS Analyses Home Page.
