The Status of Land Use/Land Cover Mapping

The Status of Land Use/Land Cover Mapping

In Wisconsin’s Lake Michigan Coastal Counties

______

-- Draft --

June 23, 1998

David Blough

Student Hourly

I.  Inventory of Land Use

Overview

Land use data for the 11 Lake Michigan Coastal counties is available primarily from two sources: (1) the Bay-Lake Regional Planning Commission (Bay-Lake RPC), encompassing the northeastern coastal counties, and (2) the South East Wisconsin Regional Planning Commission (SEWRPC), the southeastern coastal counties. In addition, Brown County has its own land use data available.

Bay-Lake RPC has two sets of land use data. The first is part of a Bluff Stability/Bluff Erosion Study done in cooperation with Dr. David Mickelson of UW-Madison and others. The Bluff Stability study area is 1000 feet inland from the Lake Michigan/Green Bay coastlines, and draws on data from 1996. Other land use data compiled by RPC covers various townships, cities and villages during the period 1995-present.

SEWRPC compiles digital land use data from aerial photos for all counties in its jurisdiction every 5 years. Land use is classified at the parcel level with a 3-digit code into approximately 15 general categories, which are in turn subdivided by specific uses.

Brown County has countywide land use, vintage 1990 for the county and 1994 for the City of Green Bay. Land use was compiled at the parcel level, but aggregated where there are contiguous land uses. There are some subparcel designations. Brown County Land Use Codes (R1,RA, R2, …MF, C50, C60, C70,...I20, I40, I72…) are the classification system, a hybrid land use/land cover system.

Bay-Lake RPC Land Use Data

Sources/Agency contacts

Mark Walter, GIS coordinator

Aaron Schutte, Community Planner

Bay-Lake Regional Planning Commission

920-448-2820

Date(s) of interview/source material

Phone conversation with Mark Walter, June 10, 1998. Phone conversation with Aaron Schutte, June 15, 1998.

General Description

Bay-Lake RPC has two sets of land use data. The first is part of a Bluff Stability/Bluff Erosion Study done in cooperation with Dr. David Mickelson of UW-Madison and others. The Bluff Stability study area is 1000 feet inland from the Lake Michigan/Green Bay coastlines, and draws on data from 1996. Other land use data compiled by RPC covers various townships, cities and villages during the period 1995-present.

Attribute information

Land use in the RPC’s data is classified according to a 3-digit code developed by SEWRPC, Bay-Lake RPC, and others several decades ago. There are nine broad development categories:

100s residential

200s commercial

300s industrial

400s transportation

500s communication/utilities

600s institutional/governmental facilities

700s outdoor recreation

800s agriculture/silviculture

900s natural areas

The 9 categories are further divided into roughly 40-50 subcategories.

Method of Classification, Scale, & Accuracy

Land use was classified based on field notes taken on unrectified air photos. The most recent air photos available were used:

County / Date of Photos / Scale of Photos
Sheboygan County / 1995 / 1” = 400’
Brown County / 1994 / 1” = 200’
Door County / 1994 / 1” = 400’
Manitowoc & Kewaunee Counties / 1990 / 1” = 800’
Marinette & Oconto / NAPP 1992 / Enlarged to 1” = 400’

After notes were taken on these photos in the field, they were placed on a light table and overlayed with a vellum base map of parcels with building footprints. Field notes on the photos were traced onto the parcel maps and then digitized. Distortion from relief displacement, tilt, and scale variation on the photos was minimal and corrected during the tracing process.

Based on the source data and the method of classification, the geographic information should be useable at scales of 1:2400 to 1:9600 with reasonable accuracy.

Date(s) compiled

The Bluff Stability study draws on field observations from 1996. Other land use studies draw on field observations from 1995 (Sheboygan County) to the present.

Datum & Coordinate System

Bay-Lake RPC converts all county-level data into WTM27. Some counties within Bay-Lake RPC use Wisconsin County Coordinates (based on NAD 83). Door and Kewaunee counties use State Plane Central.

Bay-Lake has an AML in PC Arc/Info for converting between datums and projections. Schutte reports that converting back and forth has not caused any noticeable degradation of accuracy, for instance, when data sets are edge-matched.

Geographic coverage & gaps

Taken together, the RPC’s land use data covers most of the Lake Michigan/Green Bay coast inland 1000 feet, plus several townships, cities and villages within the Bay-Lake region. Bay-Lake can provide a map showing the areas which have been mapped for land use. A verbal description follows, based on Mark Walter and Aaron Schutte’s descriptions:

The Bluff Stability study runs from the Sheboygan-Ozaukee border north to the Sturgeon Bay shipping canal. Some urban areas along the Lake Michigan coast – such as Manitowoc and Two Rivers -- are not classified by land use because no bluff profile exists in those areas. There is a gap in coverage at the village of Cleveland, and the Lake Michigan side of Door county is not covered. However, Green Bay is covered to Marinette county.

The Green Bay coast, not included in the Bluff Stability study, is mostly classified through a variety of local land use studies conducted at different times since 1995. These other land use studies include:

1. City of Marinette

2.  Town of Peshtigo

3.  Town of Little River

4.  Town of Oconto

5.  City of Oconto

6.  Town of Pensaukee

7.  Town of Little Suamico

8.  4 miles inland along Green Bay through Town of Scott, City of Green Bay, and UW-Green Bay

9.  Town of Red River

10.  Town of Gardner

11.  Town of Nesauwapee

12.  The west ½ of the Town of Sebastapol

13.  Unincorporated areas south of Clay Banks, Door County, to the city of Sheboygan

14.  An area around the city of Kewaunee

15.  The Manitowoc-Two Rivers sewer service area, but not at level of individual housing units (single family, multi-family); instead, a more generalized classification.

16.  Town of Manitowoc Rapids

17.  Town of Newton

18.  Town of Mosel

19.  Town of Sheboygan

20.  Town of Wilson

21.  City of Sheboygan

Gaps in coverage include:

1. Town of Suamico

2. Village of Howard

3.  Sturgeon Bay

4.  Most of Door County

5.  Town of Centerville

6.  Town of Cleveland

7.  Town of Holland

Resource estimates for completion of mapping

The Town of Holland is currently in progress and may be done in about a month. Sturgeon Bay is working on a land use plan, but the status and timetable of this plan is unknown.

Completing the 1000 meter buffer along the Lake Michigan/Green Bay shoreline would require extending the current 1000-foot buffer included in the Bluff Stability study and filling in other gaps along the shoreline, including much of Door County. Schutte estimates the following time would be required if Bay-Lake were to do this:

2.5 to 3 weeks to inventory the remainder of the 1000 meter buffer, and to include Door County and other gaps

+ one month to clean and digitize base maps

= approximately 2 months of 40-hour weeks total to get the finished product.

Data access and sharing policies.

Bay-Lake RPC makes their data available for distribution, and provides metadata with each release. Data requests should be in the form of a letter specifying the area, coordinate system, and datum desired. Requests take at least 2 weeks to process. Mark Walter can accommodate ZIP, JAZ, and CD media, but large file transfers across the internet take hours with Bay-Lake’s current equipment.

Data Format & File Size

Bay-Lake RPC runs Arc/Info NT, ArcView, and AutoCad. The RPC uses AutoCad as the base drawings for most of their geographic records. Most geographic data are in the form of PC/ArcInfo coverages or AutoCad drawings. As a rough estimate of file size, the entire Bluff Stability study, in compressed format, is 7.5 Megabytes.

SEWRPC Land Use Data

Sources/Agency contacts

“Appendix A. Spatial Data Inventory for Lake Michigan Coastal Counties in Wisconsin”

David Hart.

Date(s) of interview/source material

Interviews by David Hart of various county officials:

Ozaukee County: March 13, 1998

Milwaukee County: March 12, 1998

Racine County: March 17, 1998

Kenosha County: March 17, 1998

General Description of land use/land cover resources

Attribute information

Land use is classified according to a SEWRPC-derived system, comprised of a 3 digit code with a trailing code for density. Approximately 111 categories can be generalized to approximately 15 categories.

Method of Classification, Scale, & Accuracy

SEWRPC compiles digital land use mapping at 5 year intervals (1963, 1970, 1975, 1980, 1985, 1990, 1995 in progress) from 1”=400’ aerial photos. Land use is field checked, delineated on photos, and board digitized. The inventory is at the parcel level, but is not snapped to the cadastral data.

Date(s) compiled

[above]

Datum & Coordinate System

[unknown]

Geographic coverage & gaps

SEWRPC may have digital land use mapping for the entire coastal area in Ozaukee, Milwaukee, Racine, and Kenosha Counties. Milwaukee county reports 100% coverage of coastal areas. The City of Milwaukee has a separate data set. Racine county reports 100% of coastal areas, with SEWRPC currently compiling 1995 land use data for the Town of Mount Pleasant. Kenosha county has 1995 SEWRPC data, and may have 1990 SEWRPC data as well.

Resource estimates for completion of mapping

[Unknown]

Data access and sharing policies.

[unknown. SEWRPC sells its data at a non-trivial price.]

Data Format & File Size

[unknown]

II. Inventory of Land Cover

Overview

A variety of digital land cover data is available through Wisconsin’s Department of Natural Resources (WIDNR), but little is usable at scales larger than 1:24,000 for the counties bordering Lake Michigan. Resources which come close to the scale requirement are:

Wetlands data from 1:24,000 rectified photo base maps, begun 1990.

Hydrography from USGS 7.5’ quad separates, currently in progress.

DEMs from USGS 7.5’ quads, at 30 meter resolution.

DLGs from USGS 7.5’ quads for various themes such as Boundaries, Hydrography, Transportation Status, and Manmade Features.

Rural Land Cover Vegetation Mapping from Landsat Thematic Mapper satellite images taken primarily in 1992, minimum mapping unit of 5 acres, usable at 1:40,000 to 1:500,000.

Of the above list, the Rural Land Cover Vegetation Mapping data, commonly referred to as the WISCLAND dataset, is the most unique and current source of land cover data for the state of Wisconsin, despite its small scale.

WISCLAND Land Cover Data

Sources/Agency contacts

(1)  Ted Koch and Bob Gurda, State Cartographer’s Office

(2)  WIDNR Geo Services’ web site, http://www.dnr.state.wi.us/org/at/et/geo/

(3)  Metadata provided with an early release of WISCLAND data (wlc_0198.txt).

Date of interview/source material

Conversations with Ted Koch and Bob Gurda were on June 10, 1998. Pages on the WIDNR web site generally give a “last modified” date in early June 1998, but the content seems to refer to March 1998 or in some cases earlier. The WISCLAND metadata is dated Feb 9, 1998.

General Description

WISCLAND, a voluntary partnership of public and private entities, seeks to develop and analyze statewide landscape GIS data. One WISCLAND project is the creation of a seamless, statewide rural land cover data layer from the analysis of satellite imagery. This project is made possible through cooperation among the WIDNR, the National Biological Service's Gap Analysis Program (GAP), and the USDA Natural Resources Conservation Service, among others.

Attribute information

The WISCLAND Rural Land Cover Vegetation Mapping dataset classifies land cover into approximately 14 classes, based on images taken by the Landsat TM satellite on two different dates in 1992 . Land cover categories, designed to be compatible with existing classification schemes such as Anderson’s and UNESCO’s, include:

URBAN/DEVELOPED

High Intensity

Low Intensity

Golf Course

AGRICULTURE

General Agriculture

Herbaceous/Field Crops

Row Crops

Forage Crops

Cranberry Bog

GRASSLAND

FOREST

Coniferous

Jack Pine

Red Pine

White Spruce

Mixed/Other Coniferous

Broad-leaved Deciduous

Aspen

Oak

Maple

Mixed/Other Broad-leaved Deciduous

Mixed Deciduous/Coniferous

OPEN WATER

WETLAND

Emergent/Wet Meadow

Lowland Shrub

Forested: Broad-leaved Deciduous

Forested: Coniferous

Forested: Mixed Decidous/Coniferous

BARREN

SHRUBLAND

CLOUD COVER

Method of Classification, Scale, and Accuracy

The WISCLAND classification system involves comparing each pixel of an image with the spectral properties of “ground truthed” pixels, or pixels with known land cover. This comparison is done through a clustering algorithm which is “trained” by image analysts. The resulting classification of an image is then checked for accuracy using a second sample of pixels with known land cover.

The classification scheme is hierarchical, allowing land cover types to be aggregated together if specific types can not be distinguished with sufficient accuracy. For example, if Corn cannot be distinguished as a specific land cover, it can be aggregated into the more general category of Row Crops. Similarly, if Red Pine and Jack Pine both exist, but only Red Pine can be accurately distinguished, Red Pine will be coded as Red Pine, but Jack Pine with be coded Other Coniferous.

The resulting dataset comes with user’s and producer’s accuracies for each land cover class in each image tile covering the state of Wisconsin. Because the dataset is limited by the pixel size of Landsat imagery, it is usable at 1:40,000 to 1:500,000.

Date(s) compiled

The WISCLAND classification draws on Landsat TM images taken on two different dates, primarily in 1992. The final classification should be available for release by June 30, 1998.

Datum & Coordinate System

The WISCLAND dataset is referenced to the 1991 adjustment to the North American Datum of 1983, and is projected in the Wisconsin Transverse Mercator coordinate system (WTM83/1991).

Geographic coverage & gaps

The WISCLAND dataset covers the entire state of Wisconsin.

Resource estimates for completion of mapping

The WISCLAND dataset is scheduled for completion on June 30, 1998. There is high demand for repeating the project with more recent satellite imagery in order to study land cover change over time, if resources of approximately $1 million can be found.

Data access and sharing policies.

WIDNR currently licenses the GIS datasets it produces or co-produces. Licenses are sold on a cost-recovery basis, and users are prohibited from distributing the data to other users. According to Bob Gurda of the State Cartographer’s Office, the WIDNR may soon liberalize its data sharing policies to reduce administrative overhead. If this occurs, WISCLAND data may be available with few restrictions.