Additional Datasets: Additional datasets were used when available as gap fillers and or to map new hydro-geomorphic coastal wetland extents. The IJC's Lake Ontario Wetland Inventory (LOWI) was used to fill gaps along the St. Lawrence only. It was received to late in the project to be utilized for all of Lake Ontario.
Canadian Evaluated Wetland Standards: This project builds off of existing NRVIS Evaluated Wetland data which was derived from Wetland Evaluation records. These are the most detailed surveys currently available for wetlands in Ontario. The Wetland Evaluation reports were collected from OMNR District Offices and range from 1983 to 1997. The data consists of polygon features designated as wetland through the Ontario Wetland Evaluation System. All data in the evaluations were interpreted and field verified. The NRVIS data standards for horizontal accuracy in this dataset is +/-5 meters In the Ontario Wetland Evaluation Process, wetlands smaller than 2 hectares will not be evaluated, the wetland boundary was drawn where 50% of the plant community consists of upland species and a 2 meter depth contour (at low water) was used to define the deep water boundary between wetland and open water. (Ontario Ministry of Natural Resources."Ontario Wetland Evaluation System, Southern Ontario Manual. 3rd Edition" March 1993.)
Digitizing and Data Registration Standards: Where Canadian coastal wetlands have been identified but do not have suitable digital polygon data, the wetland boundary and corresponding area will be generated through delineation. The preferred method due to time and project constraints was to complete this digitally. Accuracy performance criteria are essential when digitizing to reduce the error introduced during conversion of 3-D real world objects into 2-D map objects. The peer group and project leads have accepted the accuracy criteria for this conversion as follows: 1) Air photos for identified wetlands will be scanned into digital image format that produces an obtainable minimum resolution. For the 1:10 000 and 1:20 000 scanned photos must produce an acceptable sub meter pixel resolution. For 1:10 000 photos, scanning the image at 600dpi will produce a pixel of approximately 0.4m. The 1:20 000 photos will also be scanned at 600dpi, due to data storage constraints, and will produce a pixel resolution of 0.8m. 2) The quality of the scanned imagery, including the evenness of contrast and brightness ranges, should be radiometrically colour-balanced across the wetland area to assist in the photo mosaic. 3) The georegistration of image-to-ground coordinates will be done using ArcGIS 8.2. 4) All images will be compiled using the 6 Degree Universal Transverse Mercator (UTM) projection expressed in meters, with appropriate UTM zone specified. The horizontal datum will be North American Datum Adjustment 1983(NAD 83). 5) For the purpose of establishing ground control points (GCP), high precision network data will be derived from the OMNR's Natural Resources Values Information System (NRVIS), provided in ARC/INFO export interchange format. Coverages of permanent positions, including roads, railways and utility lines, are most effective for use in determining GCPs. Wherever possible transport features (e.g. road intersections) should be selected. 6) The accuracy of GCPs is absolutely critical. The images must have a Root Mean Square (RMS) error within a measured positional accuracy of +/- 5 meters, with the corresponding RMS text files saved to confirm this result. 7) As georeferencing accuracy is contingent to the base data and to the scale of the photo, the RMS standard has been set to best meet the areas within the investigating extent. 8) The GCPs should be well distributed throughout the photo rather than clustered together, with a minimum of 5 points collected. Where photos are mosaiced together for complete wetland coverage, there should be at least 3 tie points per adjacent photo. Each GCP must be selected and referenced at a scale of 1:500 9) All georeferenced photos will be saved to CD in a .tiff or .sid format and be accompanied by the tiff world file (.twf) or sid world file (.swf) accordingly, and the RMS text file. These criteria are consistent with the protocol utilized by the OMNR in the creation of the evaluated wetland polygons, and assisted the maintenance of data integrity though the developing coastal wetland dataset. For more detailed information on NRVIS spatial accuracy please see the NRVIS Guide for End Users.
Each coastal wetland identified was classified hydrogeomorphologically. The classification schema was decided and agreed upon by the Great Lakes Coastal Wetlands Consortium Working Group. In conducting this classification on a wetland-by-wetland basis, certain rules and/or assumptions were made: 1) It was assumed that all coastal wetlands of the Great Lakes will meet the criteria of at least one of the hydrogeomorphic classifications listed. 2) If a wetland was complexed with more than one hydrogeomorphological type, the existing wetland polygon was split to best represent each individual. A new wetland name was given to each hydrogeomorphic type. In a complexed and or evaluated wetland system, each hydrogeomophically typed wetland was given the name of the wetland complex followed by a number, sequentially from west to east 3) In cases, where anthropogenic alteration has disrupted the hydrology of the system, the wetland hydrogeomorphic classification was to best represent its original connection to the lake, before alteration occurred. 4) Coastal wetlands must reside within the lake specific, historic high water level contour, as recorded by the Canadian Hydrological Service. If a continuous wetland extends outside of this boundary, it will also be included in the Inventory. A basin flood plain provides a reference to the upper extent of the coastal wetland located in that basin. The flood plain is a maximum average of a fluctuating boundary and is therefore, by nature not very accurate. It was not used as a definitive boundary but as a guide of reference to a possible upper extent. The extent delineation will occur after air photo interpretation and will be based on the natural wetland continuum. A continuous wetland was included in this Coastal Wetland Inventory in its entirety but where it was not continuous, the flood plain provides the upper limit of what was included with the wetland complex.
Wetland Boundary Delineation : The polygon extents in the coastal wetland database were provided by the OMNR and were mostly accepted as true. Spatial editing and/or the creation of new data, only occurred in 3 cases: 1) the absence of a digital wetland boundary 2) splitting/removing the non-continuous upper extents of an existing wetland boundary because it was not considered coastal and 3) complexed wetlands whose current boundaries need to be split into hydrogeomorphological entities.
For the Canadian portion of this dataset, remote images were necessary for the classification and were needed, boundary deliniation. The OMNR has colour IR photos at a scale of 1:10 000, taken in the summer months between the years of 1994 and 2000. They were available for the entire southern basin of the Great Lake's extending to the southern half of Lake Huron.The CIR coverage ended just north of the Parry Sound district border The northern basin had to be compensated with an alternative image source. Forest Resource Inventory (FRI) black and white contact prints provided this alternative. They were available for the remaining coastal wetlands of the basin at a scale of 1:20 000. These photos were also all taken during summer months, but their date range is a littler older, from 1986 to 1994. The scale and quality of the FRI's still allowed for proper geomorphic classification of the wetland and between these two sources, there was full coverage of the Great Lake's basin with relative consistency. All acquired photos were analyzed in analog form with the assistance of a stereoscope. In cases where the digital wetland boundary polygons did not exist for identified coastal wetlands, or significant spatial alterations needed to occur, the aerial photos were scanned into digital format and georeferenced to OMNR NRVIS data. This provided a digital tablet from which to on-screen digitize the coastal wetland boundary in ArcGIS 8.2. Delineation of the wetland boundary was generated using standardized air photo interpretation techniques (Owens and Hop 1995) and was limited to wetland areas greater than 2 hectares. The 2 hectare minimum is consistent with the OMNR's evaluated wetland polygon data. In areas where large wetland complexes exist it was more efficient and thus, cost-effective to obtain satellite imagery than aerial photos. Such areas include St. Clair Marsh Wetland Complex, Wapole Island in Lake St.Clair and Long Point Wetland Complex in Lake Erie. For these areas, digital 5m panchromatic imagery and 15m multispectral imagery was obtained through the OMNR. The Coastal Wetland Inventory is considered complete for the entire Great Lakes basin.
Hydrogeomorphic types and descriptions are outlined in the classification schema developed by D. A. Albert, J. Ingram, T. Thompson and D. Wilcox, on behalf of the Great Lakes Coastal Wetland Consortium (GLCWC). See "Great Lakes Hydrogeomorphic Classification Schema.doc". A workshop held in partnership with U.S. project leads and the GLCWC, created this peer accepted classification. It is to be submitted for peer review into the Journal of Great Lakes Research (International Association for Great Lakes Research (IAGLR))
There is potential for updates in the upper Great Lakes. Data gaps were very extensive for Lake Superior and north Lake Huron on the Canadian side. This dataset accounted for all data gaps outlined in the Environment Canada and OMNR's March 2003 publication "Coastal Wetland Atlas, A Summary of Information (1983-1997)" and photo coverage was received for all these areas. However, if a potential wetland lay outside of the obtained photo coverage, it would not be included in the dataset. Due to the extensiveness of island archipelagos and rocky outcroppings found in these areas, it is most likely that coastal wetlands greater than 2 hectares are missing from the final dataset. Future updates, may want to give priority to these areas. The lower Great Lakes have extensive datasets available and photo coverage extended for the entire shoreline. The Coastal Wetland Inventory is very comprehensive from Lake St Clair to the Cornwall Dam on the St Lawrence River.
Horizontal accuracy for the Evaluated Wetland polygon layer is estimated to be on the order of +/- 5 meters (NRVIS, Technical Reference Guide for End-Users, Ontario Digital Geographic Database(ODGD) Natural Resources Values and Information (NRVIS) Guide. April 2000) The horizontal accuracy of all newly digitized polygons is based on the control data and methodology used to extract and position control points on the image. The OMNR base data used in the rectification process has a horizontal accuracy of +/- 5 meters. All photo registration RMS error was maintained to less tha 0.5m. Text files have been saved for verification.
The positional accuracy of the data set has not been tested under the National Standards for Spatial Data Accuracy.
Data collection a preparation: Arc View 3.3 projects were set up containing all of the existing digital datasets described in the citation section of this metadata. Datasets were classified on the basis of land-use, to assist in locating, delineating, and classifying coastal wetlands. The working projection for most lakes was Albers Equal Area. However, to save processing time, the working projection for wetlands in Michigan and Wisconsin was Michigan GEOREF and Wisonsin Transverse Mercator (WTM), respectively, and projected to Albers upon lakewide coverage completion. Michigan's data was projected with the help of a projection tool offered on the Michigan Center for Technology's website at <http://www.michigan.gov/cgi/>, Wisconsin coastal wetland data was projected with ArcINFO 8.3 and ArcToolbox 8.3. Additional Data Sources: Doug Wilcox and Martha Carlson's International Joint Commission work, Lake Ontario Wetland Inventory, LOWI, was utilized for the St. Lawrence River when it became available. The LOWI data could be utilized for all of Lake Ontario; however, the scope of this project does not allow us to go back incorporate this data for areas of the lake that were complete before receiving the data. MNFI provided a polygon wetland coverage for Lake Michigan, MNFI Wetland Communities inventory. The original Herdendorf hardcopy topographic maps are missing for the Isle Royal area in Michigan (Herdendorf wetlands 320-348), all of Lake Michigan, and locally in the Erie Island and Sandusky area of Lake Erie. Thus, all hydro-geomorphic wetland extents for the HWI on Lake Michigan in Michigan were based on MNFI's hard-copy maps of field-verified wetland communities. On Lake Superior, MNFI made interpretations of coastal wetland complexes based on the 1978 color infra-red aerial photography and the Herdendorf USFWS reports (Herdendorf, C. E., S. M. Hartley, and M. D. Barnes (Eds). 1981a). For these areas, the NWI was still considered the most accurate source of data for the digital mapping of hydro-geomorphic coastal wetland complexes. Also a digital inventory, MNFI Wetland Communities Inventory was selected to create new geomorphic complexes over the NWI coverage when it was deemed a more accurate representation of the hand-drawn hydro-geomorphic wetland extents (Dennis A. Albert, MNFI, personal commun., 2004).. Lastly, some additional wetland locations from the Regional Environmental Monitoring and Assessment Program (REMAP) projects by USFWS and USEPA were included with each Herdendorf Wetland Inventory point coverage (HWI).
The published document entitled "The Ontario Great Lakes Coastal Wetland Atlas: A summary of information (1983 - 1997) consolidated and evaluated all available coastal wetland data. It identifies a UTM zone specific centroid position for these wetlands. Supplementary digital spreadsheets of this data were available with the publication. The MS Excel zone specific spreadsheets are imported into ArcGIS 8.2. The 'display XY data' function in ArcMap was used for the import. This digital point data was saved as a zone specific shapefile. Each shapefile was reprojected into zone 18 and merged into one coverage using ArcMap's Geoprocessing Tool. The resultant dataset are starting point locations of known coastal wetlands of the Great Lakes.