Updating the National Wetland Inventory (NWI) for Southern Lower Peninsula of Michigan

Transcription

1 Updating the National Wetland Inventory (NWI) for Southern Lower Peninsula of Michigan FINAL REPORT Submitted to: U.S. Fish and Wildlife Service Great Lakes Coastal Program Office Agreement Number: J010 Task Order Number: T010 & U.S. Fish and Wildlife Service Upper Mississippi River - Great Lakes Joint Venture Agreement Number: J010 Task Order Number: T008 Submitted by:

2 TABLE OF CONTENTS Acknowledgements... 2 Disclaimer... 2 Abstract Introduction Project Justification NWI Background Project Area Methods Data Sources Drained Wetlands New Wetlands Field Verification Accuracy Assessment Results Wetland Loss by Total Acreage Wetland Loss by Number of Wetlands Wetland Loss by Drainage Class Wetland Loss by NWI Type Emergent Wetlands New Wetlands Net Change in Wetland Acreage Accuracy Assessment Project Summary Literature Cited Web Sites Appendix A: Wetland loss per county (Acres) Appendix B: Wetland loss per county (# of wetlands) Appendix C: Wetland loss by drainage class per county Appendix D: Wetland loss by NWI Type per county Appendix E: Emergent wetland loss by county Appendix F: New wetlands per county Appendix G: Net Change in wetland acreage per county Appendix H: Accuracy Assessment Error Matrix

3 Acknowledgements Funding for this project was made available from the USFWS: Great Lakes Coastal Program, USFWS: Upper Mississippi River - Great Lakes Joint Venture, and Ducks Unlimited. Disclaimer This project was designed to develop an updated wetland inventory for: 1) input into habitat models, and 2) planning and targeting habitat restoration and protection efforts. It was not designed, and should not be used, for wetland regulatory issues. Ducks Unlimited, Inc. makes no representation or warranty of any kind regarding this material, data and information, including, but not limited to, the accuracy of the material, data and information or its suitability for any purpose. All use of the material, data and information is at the user s sole risk. By using any of this material, data and information, the user agrees that Ducks Unlimited, Inc. is not responsible for their use of the material, data and information or the results thereof. For additional information about the project, report, or maps, please contact: Ducks Unlimited, Inc., Great Lakes/Atlantic Regional Office, 331 Metty Drive, Suite 4, Ann Arbor, MI 48103, (734)

4 Abstract It is important to use data that is both temporally and spatially relevant when developing waterfowl habitat models and planning wetland restoration and protection activities. The U.S. Fish and Wildlife Service s National Wetlands Inventory (NWI) has been used for many modeling and planning activities due to its areal extent and spatial relevance. However, much of this inventory is between 15 and 20 years old. As part of a pilot project to test methodology for updating NWI throughout the Great Lakes, Ducks Unlimited has updated the NWI for the southern portion of Michigan s Lower Peninsula by identifying both losses and gains of wetlands. The update was accomplished by viewing the NWI data over spring 1998 series color infrared (CIR) digital orthophotos to determine the status of the wetlands. Drained wetlands were given a modifier as to the drainage type (agriculture, developed, recreation, or other). Wetlands that were identified on the 1998 series digital orthophotos and not part of the NWI were digitized into a new layer. A stratified random sample of the NWI update was selected and field verified to assess the accuracy of the interpretation. This project identified 42,876 acres of drained wetlands and 12,665 acres of new wetlands in the project area. The overall accuracy of the photo interpretation was 84%, however, there was a five to seven year time difference in the aerial photos and field verification. The data layers produced in this project will increase the accuracy of habitat models and planning activities in Southern Michigan. 3

5 1.1 Project Justification 1. Introduction A current and comprehensive understanding of wetlands (including type, abundance, and geographical distribution) can be an important limiting factor when developing habitat models and planning wetland restoration and protection efforts. The U.S. Fish and Wildlife Service s National Wetlands Inventory (NWI) has been used for many modeling and planning activities due to its areal extent and spatial relevance. However, much of this inventory is between 15 and 20 years old (Figure 1). Furthermore, in some areas of the NWI, small scale black and white aerial photos were used for the classification. This type of aerial photography is not suited for identifying small wetlands. Over the past years the landscape in Michigan has changed dramatically (Michigan Land Use Leadership Council, 2003). The use of outdated information in the modeling or planning process can have dramatic effects on the modeling and planning outcomes. Having an understanding of current wetland status will improve the planning and biological foundation of the Great Lakes Coastal Program, Upper Mississippi River - Great Lakes Joint Venture, Ducks Unlimited, and their partners. A current assessment of wetlands in Michigan and the Great Lakes will allow these partnerships to refine their modeling efforts and, more importantly, target critical areas for restoration, enhancement, and protection. Current habitat assessment models such as the Habitat Evaluation Network (HEN) and Great Lakes mallard distribution rely on NWI data for modeling. This updated wetlands layer, in conjunction with other habitat layers, will enable partnerships involved in wetland restoration to plan and manage their activities with a higher degree of accuracy and efficiency. From: Figure 1. Era of Digital Data Photography used for Original NWI 4

6 1.2 NWI Background The NWI was initiated by the U.S. Fish and Wildlife Service in 1975 in response to a dramatic loss of wetlands across the United States. The program provides detailed information on wetlands throughout the United States. Coastal wetlands and wetlands along major river systems were the first areas where wetlands were mapped because of their biological and economic importance. To date, over 90 percent of the lower 48 states and over 50 percent of Alaska has been inventoried. For more information, see the Web Site section of this document. NWI uses the Cowardin et al. (1979) definition and classification scheme for identification of wetlands. The Cowardin system is the national standard for wetland mapping, monitoring, and data reporting. A wetland is defined as having one or more of the following three attributes: (1) at least periodically the land supports predominantly hydrophytes, (2) the substrate is predominantly undrained hydric soil, (3) the substrate is nonsoil and is saturated with water or covered by shallow water at sometime during the growing season of each year (Cowardin et al. 1979, Dahl 2000). In Michigan, the three categories of wetlands are: Riverine (river systems), Palustrine (freshwater marshes), and Lacustrine (lakes and reservoirs) (Figure 2). For more detailed information on the classification scheme, see the Web Site section of this document. Riverine (may be tidal or non-tidal) Fresh Water Habitats Paulstrine Forested Palustrine Shrub Palustrine Emergents Palustrine Unconsolidated Shore Palustrine Unconsolidated Bottom Palustrine Aquatic Bed Lacustrine River Systems Forested Swamps Shrub Wetlands Inland Marshes/Wet Meadows Shore Beaches/Bars Open Water Ponds Floating Aquatic/Submerged Vegetation Lakes and Reservoirs Figure 2. NWI Wetland Classification Scheme The NWI classification was compiled through manual photo interpretation of NHAP and NAPP photography (see Web Site section of document) and supplemented by soil surveys to identify hydric soils. The delineation of the wetland boundaries were then transferred to USGS 7.5 minute topographic quadrangle maps and manually labeled according to the Cowardin classification scheme. The Michigan Department of Natural Resources contracted with a private vendor to convert the hardcopy NWI maps to a GIS format. The US Fish and Wildlife Service performed some additional attribute editing of this GIS layer, which can be downloaded at their web site. For additional information on the NWI classification process, see the Web Site section of this document. 5

7 2. Project Area The project area encompasses approximately 3/4 of the lower peninsula of MI. The boundary is derived from a combination of both the Upper Mississippi River - Great Lakes Joint Venture Primary Focus area as well as the Saginaw Bay watershed (Figure 3). Figure 3. Project Boundary 6

8 3. Methods 3.1 Data Sources The primary data source used to update the NWI data was the 1998 series color infrared (CIR) digital orthophotos. The 1998 series digital orthophotos were acquired in the spring (March, April, May) of using color infrared (CIR) film (see Web Site section for a link to the metadata and acquisition dates). These photos were used because they were the most current and complete photos for the entire project area. The photos have a resolution of 1 meter, and are available at the State of Michigan s Center for Geographic Information (CGI) web site (see Web Site section). The NWI data used in this project is available from both the Michigan s Center for Geographic Information and the U.S. Fish and Wildlife s NWI web sites. 3.2 Drained Wetlands The procedure for identifying drained wetlands in the NWI data consisted of viewing the NWI wetlands on top of the spring 1998 series digital orthophotos and interpreting the status (unchanged, drained, partially drained, or modified) of the wetlands (Figure 4). The project area was interpreted systematically by viewing a section of land starting in the upper left corner of a county and progressing throughout the county on a section by section basis. The status of an NWI wetland was considered unchanged when the 1998 series photo was interpreted as the same NWI Class as the NWI wetland. Differences in the NWI Subclasses were not considered in this project. The status of an NWI wetland was considered drained when the 1998 series photo was interpreted as a different landuse class (ie: agriculture, development, recreation). The status of an NWI wetland was considered partially drained when part of the NWI wetland was interpreted as a different landuse class in the 1998 series photo and part of the NWI wetland was considered unchanged. The partially drained NWI wetland polygons were split into a drained and unchanged polygon. The status of an NWI wetland was considered modified when the 1998 series photo interpretation classified the NWI wetland in a different class then what the NWI wetlands was labeled (ie: shrub/scrub to unconsolidated bottom). 7

9 Partially Drained Wetland Drained Wetland Modified Wetland Figure 4. Types of Wetland Drainage Unchanged Wetland Six attributes (image date, drained, drain type, modified, partial, and comments) were added to the existing NWI attribute table to track and report the drained wetlands. The image date field was attributed with the date of the aerial photo used to interpret the status of the wetland (in this case it was 1998). The drained field was attributed with either a Y if it was drained or N if the wetland was not drained. For wetlands that were drained, the drain type field was attributed with one of the following four categories: agriculture, recreation, development, and other (Figure 5). If the wetland was not drained this attribute was left black. Table 1 displays the land uses that were interpreted from the air photos with their associated drainage class. The modified field was attributed with either a Y if it was modified or N if the wetland was not modified. In some cases, only part of the wetland was drained so the partial attribute was filled in with a Y. This attribute was recorded so that the partially drained wetlands could be separated from the fully drained wetlands when calculating the total number of wetlands drained. The comments field was filled in with any additional comments unique to the individual wetland. 8

10 Drained due to Agriculture Drained due to Development Drained due to Recreation (golf course) Figure 5. Drained due to Other (logging) Drainage Classes Table 1. Land Use Interpreted and Their Associated Drainage Classes Agriculture Development Recreation Other orchards urban housing golf courses logging crops rural housing race tracks airports pasture industry sports fields highways farm houses/ structures commercial unknown/unidentified 9

11 3.3 New Wetlands In addition to updating the NWI layer any new wetlands not part of the original NWI were digitized into a separate GIS layer. These new wetlands may be the result of inadequate detail (due to the scale of the photography) when the NWI was originally classified. The boundaries of the new wetlands were identified and digitized from the 1998 series digital orthophoto and attributed with the appropriate NWI code (Figure 6). New Wetlands NWI Wetlands Figure 6. An Example of the New Wetlands Digitized. 3.4 Field Verification During the interpretation process, there were some wetlands that could not be positively identified as being drained. This may have been a result of a poor aerial photo for that particular area, or it was too difficult to determine the drainage class from the photo. In these cases, the wetlands were field inspected to determine if they are drained and their drainage class. 10

12 3.5 Accuracy Assessment A stratified random sample of the drained and un-changed wetlands was chosen to verify the accuracy of the interpretation in the field. To ensure a representative sample of wetland types, the samples chosen reflected the proportion of class types (forested, emergent, scrub-shrub, unconsolidated bottom.) present in the study area. Due to funding and accessibility issues, all of the field samples were selected within 300 feet of roads in an eight county area (Figure 7). Once the samples were generated, a field inspection was conducted by driving to each site using a GPS unit and hand-held PDA. The existing conditions of the wetland were then entered into a database. An error matrix (Story and Congalton, 1986) was then compiled and analyzed by comparing the photo interpretation to the ground conditions. Figure 7. Counties Included in Field Inspection 11

13 4. Results The results reported in this document are for planning purposes only and are not intended for regulatory purposes. Before using any of the reported results, the user should have a through understanding of the methods. 4.1 Wetland Loss by Total Acreage The total amount of NWI wetlands in the project area was 2.5 million acres, of which 42,876 acres were drained (Table 2). Therefore, there was a 1.7% loss in wetland acreage from the date of the NWI classification (late 1970 s to early 1980 s) to the 1998 series digital orthophotos ( ). Of the drained wetland acreage, 68% was completely drained while 32% was partially drained. The highest loss based on the percent of wetland acres drained occurred on the eastern side of Michigan (Figure 8). Macomb county had the highest loss at 10.21%. For individual county results, see Appendix A. Total NWI Wetland Acreage Table 2. NWI Wetland Acreage Drained Partially Drained Acres Completely Drained Acres Total Drained Acres % Drained 2,532, , , , % Figure 8. Percent Area Drained per County Based on Acres 12

14 4.2 Wetland Loss by Number of Wetlands There were a total of 404,325 NWI wetlands in the project area, of which 21,851 wetlands were drained or partially drained (Table 3). About 86% of the drained wetlands were completely drained while the other 14% were partially drained. Based on the number of totally drained wetlands, there was a 4.7% loss in the number of wetlands for the project area. Figure 9 displays the percent loss of the number of wetlands by county. The percentage is based upon the number of completely drained wetlands in the county divided by the total number of wetlands in that county. Huron County had the highest percentage loss with 17.1%. For individual county results, see Appendix B. Table 3. Number of Wetlands Lost Total Number of NWI Wetlands Completely Drained Partially Drained Total Drained % Drained * 404,325 18,863 2,988 21, % * = based on number of completely drained wetlands/total number of wetlands in study area Figure 9. Percent Wetland Loss Based on Number of Wetlands 13

15 4.3 Wetland Loss by Drainage Class The majority of wetlands in the project area were drained due to agriculture (60.6%), followed by development (22.8%), other (11.1%), and recreation (5.4%) (Table 4 and Figure 10). Agriculture was the dominant drainage class in most counties except for the Southeastern part of Michigan, where development was the leading cause of wetland loss (Figure 11). For individual county results, see Appendix C. Table 4. Wetland Loss (acres) by Drainage Class Agriculture Ag Recreation Rec. Development Dev. Other Oth. Acres % Acres % Acres % Acres % % % % % 11% 23% Agriculture Recreation Development Other 61% 5% Figure 10. Percent Wetlands Drained by Drainage Class 14

16 Figure 11. Leading Factors Causing Wetland Drainage per County (# Acres) 15

17 4.4 Wetland Loss by NWI Type Emergent and forested wetlands had the biggest losses in acreage with over 30,000 acres drained (Table 5). Overall 38.3% of wetlands drained were classified by NWI as emergent, 36.3% as forested, 22.7% as scrub shrub, and 2.7% as unconsolidated bottom (Figure 12). There is a spatial trend that shows emergent wetlands were the dominant NWI type drained in the central part of the study area (Figure 13). For individual county results, see Appendix D. Table 5. Wetland Loss (acres) by NWI Type Emergent Emer. Forested For Scrub-Shrub ScSh Unconsolidated Bottom UBot Acres % Acres % Acres % Acres % 16, % 15, % 9, % 1, % 23% 3% 38% Emergent Forested Scrub-Shrub Unconsolidated Bottom 36% Figure 12. Percent Wetlands Drained by NWI Type 16

18 Figure 13. NWI Type Drained per County Based on Acres Drained 17

19 4.5 Emergent Wetlands Emergent wetlands are an important wetland type for waterfowl in Michigan. When compared to the percentage drained of all wetlands in the study area (1.7%), the emergent wetlands have been drained at a considerable higher rate (3.8%) (Table 6). In some counties, the percent of drained emergent wetlands is over 20% (Figure 14). For individual county results, see Appendix E. Table 6. Emergent Wetland Loss Total Emergent Drained % Acres Emergent Acres Lost 427, , % Figure 14. Percent Emergent Wetland Loss per County 18

20 4.6 New Wetlands There were 17,293 new wetlands identified in the project area for a total of 12, acres (Table 7). St. Clair County had the most new wetlands with 2,943. The counties with the highest amount of development seemed to be associated with high amounts of new wetlands (Figure 15). For individual county results, see Appendix F. Table 7. New Wetlands Identified # New Wetlands Acres Avg. Size (acres) 17,299 12, Figure 15. New Wetlands per Square Mile. 19

21 4.7 Net Change in Wetland Acreage The net change in wetland acreage from the original NWI (late 1970 s to early 1980 s) to the 1998 series digital orthophotos (1997 to 2000) was a negative 30,211 acres (Table 7). It is important to note that some of the new acres created may have been present at the time of the original NWI classification, but were not included due to the scale of the aerial photography. Most counties had a decrease in wetland acreage (net loss) while 7 counties had an increase (Figure 16). Appendix G lists the results by county. Table 7. Net Change in Wetland Acreage Acres Drained Acres Created Net Change 42,876 12,665-30,211 Figure 16. Change (acres created minus acres drained) in Wetland Acreage 20

22 4.8 Accuracy Assessment Based on the stratified random sample, there were 46 emergent wetlands, 35 forested wetlands, 15 Scrub-shrub wetlands, and 5 unconsolidated bottom wetlands that were field inspected. A total of 85 wetlands were correctly interpreted out of the 101 that were examined. The overall accuracy of the photo interpretation was 84.2 (see Appendix H for the error matrix). When performing an accuracy assessment it is important to have the field verification (reference data) collected as close to the date of the imagery being interpreted as possible so that the errors reported are not an artifact of temporal differences in the reference and classification data. For this project, it was not possible to conduct the field verification at the same time as the 1998 series digital orthophotos. There was a five to seven year difference in the acquisition of the aerial photography and the field verification, which may have caused some of the errors in the interpretation of the un-drained wetlands. 5. Project Summary The USFWS: Upper Mississippi River Great Lakes Joint Venture, USFWS: Great Lakes Coastal Program, Ducks Unlimited and their partners are in the process of creating habitat models and planning efforts for the Great Lakes. The best wetland data set for input into models and planning efforts for this area is the National Wetlands Inventory. The biggest concern in using the NWI layer in these efforts is the amount of change that has occurred since the NWI was completed (late 1970 s to early 1980 s). It was assumed that these changes could have a large impact on both the models and planning process, but, until now, there was no validation of these assumptions. With the creation of this updated NWI for part of Michigan and using the same methodology to expand it to five of the Great Lake States, the partners will be able to more accurately plan for habitat restoration and protection activities throughout the Great Lakes. 21

23 6. Literature Cited Cowardin, L.M., V. Carter, F.C. Golet, and E.T. LaRoe Classification of Wetlands and Deepwater Habitats of the United States. U.S. Department of the Interior, Fish and Wildlife Service, Washington, DC. FWS/OBS-79/31. Dahl, T.E Status and trends of wetlands in the conterminous United States from U.S. Department of the Interior, Fish and Wildlife Service, Washington, D.C. 82pp. Michigan Land Use Leadership Council Michigan s Land, Michigan s Future: Final Report of the Michigan Land Use Leadership Council. August 15, pp. (see web sites section for link to document) Story, M and R.G. Congalton Accuracy Assessment: A User s Perspective. Photogrammetric Engineering and Remote Sensing. Vol. 52, No. 3, pp Web Sites Michigan Land Use Leadership Council s Final Report located at: State of Michigan s Center for Geographic Information located at Michigan Spring CIR Digital Orthophotos metadata located at: USFWS National Wetland Inventory website located at: USFWS National Wetland Inventory map codes can be found at: Metadata for the USFWS National Wetland Inventory can be found at: Information on NAPP and NHAP Photography can be found at:

24 Appendix A: Wetland loss per county (Acres) County Acres of wetlands Completely drained (acres) Partially drained (acres) Total drained (acres) % area drained Alcona % Allegan % Arenac % Barry % Bay % Berrien % Branch % Calhoun % Cass % Clare % Clinton % Eaton % Genesee % Gladwin % Gratiot % Hillsdale % Huron % Ingham % Ionia % Iosco % Isabella % Jackson % Kalamazoo % Kent % Lapeer % Lenawee % Livingston % Macomb % Mecosta % Midland % Missaukee % Monroe % Montcalm % Muskegon % Newaygo % Oakland % Oceana % Ogemaw % Osceola % Oscoda % Ottawa % Roscommon % Saginaw % 23

25 Sanilac % Shiawassee % St. Clair % St. Joseph % Tuscola % Van Buren % Washtenaw % Wayne % Totals % 24

26 Appendix B: Wetland loss per county (# of wetlands) County wetlands in county completely drained partially drained total drained % drained Alcona % Allegan % Arenac % Barry % Bay % Berrien % Branch % Calhoun % Cass % Clare % Clinton % Eaton % Genesee % Gladwin % Gratiot % Hillsdale % Huron % Ingham % Ionia % Iosco % Isabella % Jackson % Kalamazoo % Kent % Lapeer % Lenawee % Livingston % Macomb % Mecosta % Midland % Missaukee % Monroe % Montcalm % Muskegon % Newaygo % Oakland % Oceana % Ogemaw % Osceola % Oscoda % Ottawa % Roscommon % Saginaw % Sanilac % 25

27 Shiawassee % St. Clair % St. Joseph % Tuscola % Van Buren % Washtenaw % Wayne % 0 Totals % * percent drained is based on completely drained wetlands/total wetlands in county 26

28 Appendix C: Wetland loss by drainage class per county County Ag Acres Ag % Rec Acres Rec % Dev Acres Dev % Oth Acres Oth % Total Acres Alcona % % % % Allegan % % % % Arenac % % % % Barry % % % % Bay % % % % Berrien % % % % Branch % % % % Calhoun % % % % Cass % % % % Clare % % % % Clinton % % % % Eaton % % % % Genesee % % % % Gladwin % % % % Gratiot % % % % Hillsdale % % % % Huron % % % % Ingham % % % % Ionia % % % % Iosco % % % % Isabella % % % % Jackson % % % % Kalamazoo % % % % Kent % % % % Lapeer % % % % Lenawee % % % % Livingston % % % % Macomb % % % % Mecosta % % % % Midland % % % % Missaukee 0.00 n/a 0.00 n/a 0.00 n/a 0.00 n/a 0.00 Monroe % % % % Montcalm % % % % Muskegon % % % % Newaygo % % % % Oakland % % % % Oceana % % % % Ogemaw % % % % Osceola % % % % Oscoda % % % % Ottawa % % % % Roscommon % % % % Saginaw % % % %

29 Sanilac % % % % Shiawassee % % % % St. Clair % % % % St. Joseph % % % % Tuscola % % % % Van Buren % % % % Washtenaw % % % % Wayne % % % % Totals % % % %

30 Appendix D: Wetland loss by NWI Type per county County Emerggent Acres Emer % Forested Acres For % Scrub Shrub Acres ScSh % Unconsolidated Bottom Acres UBot % Total Acres Drained Alcona % % % % Allegan % % % % Arenac % % % % Barry % % % % Bay % % % % Berrien % % % % Branch % % % % Calhoun % % % % Cass % % % % Clare % % % % Clinton % % % % Eaton % % % % Genesee % % % % Gladwin % % % % Gratiot % % % % Hillsdale % % % % Huron % % % % Ingham % % % % Ionia % % % % Iosco % % % % Isabella % % % % Jackson % % % % Kalamazoo % % % % Kent % % % % Lapeer % % % % Lenawee % % % % Livingston % % % % Macomb % % % % Mecosta % % % % Midland % % % % Missaukee 0 n/a 0 n/a 0 n/a 0 n/a 0 Monroe % % % % Montcalm % % % % Muskegon % % % % Newaygo % % % % Oakland % % % % Oceana % % % % Ogemaw % % % % Osceola % 0 0.0% % % Oscoda % % % 0 0.0% Ottawa % % % % Roscommon % % % 0 0.0% Saginaw % % % %

31 Sanilac % % % % Shiawassee % % % % St. Clair % % % % 2692 St. Joseph % % % % Tuscola % % % % Van Buren % % % % Washtenaw % % % % Wayne % % % % Totals % % % %

32 Appendix E: Emergent wetland loss by county County Emergent Acres Drained Emer. Acres % lost Alcona % Allegan % Arenac % Barry % Bay % Berrien % Branch % Calhoun % Cass % Clare % Clinton % Eaton % Genesee % Gladwin % Gratiot % Hillsdale % Huron % Ingham % Ionia % Iosco % Isabella % Jackson % Kalamazoo % Kent % Lapeer % Lenawee % Livingston % Macomb % Mecosta % Midland % Missaukee % Monroe % Montcalm % Muskegon % Newaygo % Oakland % Oceana % Ogemaw % Osceola % Oscoda % Ottawa % Roscommon % Saginaw % Sanilac % 31

33 Shiawassee % St. Clair % St. Joseph % Tuscola % Van Buren % Washtenaw % Wayne % Totals % 32

34 Appendix F: New wetlands per county County # new wetlands Acres Avg. size Acres Alcona Allegan Arenac Barry Bay Berrien Branch Calhoun Cass Clare Clinton Eaton Genesee Gladwin Gratiot Hillsdale Huron Ingham Ionia Iosco Isabella Jackson Kalamazoo Kent Lapeer Lenawee Livingston Macomb Mecosta Midland Missaukee n/a Monroe Montcalm Muskegon Newaygo Oakland Oceana Ogemaw Osceola Oscoda Ottawa Roscommon Saginaw

35 Sanilac Shiawassee St. Clair St. Joseph Tuscola Van Buren Washtenaw Wayne Totals

36 Appendix G: Net Change in wetland acreage per county County Drained Acres New Acres Net Change Alcona Allegan Arenac Barry Bay Berrien Branch Calhoun Cass Clare Clinton Eaton Genesee Gladwin Gratiot Hillsdale Huron Ingham Ionia Iosco Isabella Jackson Kalamazoo Kent Lapeer Lenawee Livingston Macomb Mecosta Midland Missaukee Monroe Montcalm Muskegon Newaygo Oakland Oceana Ogemaw Osceola Oscoda Ottawa Roscommon Saginaw

37 Sanilac Shiawassee St. Clair St. Joseph Tuscola Van Buren Washtenaw Wayne Totals

38 Appendix H: Accuracy Assessment Error Matrix C L A S S I F I E Emergent Drained Emergent Not drained Forested Drained Forested Not drained Field Scrubshrub Drained Scrubshrub Not drained Un_bottom Drained Un_bottom Not Drained User's Accuracy Emergent Drained % Emergent Not drained % Forested Drained % Forested Not drained % Scrub-shrub drained % Scrub-shrub Not drained % Un_bottom Drained % Unbottom Not drained % D TOTAL Producer's Accuracy 88.00% 95.24% 77.78% 70.59% 71.43% 87.50% % % *The total accuracy was 85/101 (84.2%) 37