Alabama

Organizing

<%Case "AK"%>

Alaska

Not started

<%Case "AZ"%>

Arizona

The AZ Web site is currently not available.

Land cover: The Arizona GAP land cover map, having 54 classes, is finished. It is available at http://www.srnr.arizona.edu/nbs/gap. The accuracy assessment of this map is being completed through a partnership of University of Arizona and Northern Arizona University. The final figures for the assessment are now being tabulated and will be available early in 1999 at the above address.

Animal modeling: Animal distribution models are being refined. These should be ready for the final GAP report by the end of December 1998.

Land stewardship mapping: The land stewardship map is completed and ready to be incorporated into the final GAP report.

Analysis: Gap analysis has been completed for Arizona; the tables and graphs are now being finalized for the final report.

Reporting and data distribution: The final AZ-GAP report is in the process of completion, with a few pieces yet to be finished. Our anticipated date of delivery is the end of February 1999. Some of the data are currently available on our web site (listed above). We will be adding data as they become available and anticipate that all data will be on the web site by the end of March 1999.

<%Case "AR"%>

Arkansas

Arkansas-GAP was completed in summer 1998 and delivered to National GAP as a 5-CD volume funded as a data distribution research project. Data are available from the AR-GAP home page (http://www.cast.uark.edu/gap/). National GAP is currently working with AR-GAP on finalizing the CD product for distribution, expected in early spring 1999.

<%Case "CA"%>

California

Land cover: Mapping is complete, and maps are available for the entire state or for any of the 10 regions of the state through the CA-GAP web site (see below).

Animal modeling: The California Department of Fish and Game helped us improve the crosswalk from plant community types to wildlife habitat types. Predictive modeling has been completed for 455 native terrestrial vertebrates, using a rating rather than a yes/no suitability mapping. We are collaborating with Dr. Jim Quinn from the University of California-Davis and his staff on a validation of the modeling by comparing predicted species lists with lists of observed species for a large sample of managed areas throughout the state.

Land stewardship mapping: Mapping has been completed. Because of redistribution restrictions on the 1:100,000-scale stewardship used for the analysis, we also compiled a 1:2,000,000-scale version being distributed with the final CA-GAP database.

Analysis: All analyses have been completed. In addition to the standard analyses for GAP, we developed a "threat index" based on roadedness and projected population growth to further prioritize plant communities appearing to be most vulnerable based on management status.

Reporting and data distribution: The CA-GAP report has been completed and is posted on our web site (see below). We are currently arranging to print a large number of copies of the executive summary for wide distribution. The GIS database is also available on-line. A CD-ROM will be published by the end of 1998 that will contain the report in Windows Help format, the database as ARC/INFO export files or image and text formats, and an interactive atlas. The interactive atlas is designed to make the CA-GAP data useful to persons without access to or experience with GIS software. It incorporates the ArcView Data Publisher software, along with a customized user interface we created, to assist users in querying and displaying GAP data in meaningful ways. The database and the CD-ROMs will be transferred to the California Department of Fish and Game for long-term maintenance and user support. Check the CA-GAP home page (see below) for links to their web site.

Other accomplishments and innovations: Besides the completion of the CA-GAP report and publication of the database, most of the progress in the past year has been in the area of applications of GAP data for other purposes. We highlight three of these applications here.

One of the problems in analyzing the potential effects of development at the county- or finer scale is determining the biological importance of the ecosystems and habitats being impacted. There may be very detailed data for the local area, but how should the regional context be measured? Is a 20% loss of a habitat in any given county significant? It depends on the county’s share of the overall extent of the habitat type. Chris Cogan, a Ph.D. student from the University of California, Santa Cruz, has linked several models and spatial analyses to evaluate the loss and fragmentation of the biota in a county. GAP status and land cover data are used in determining regional importance. Local scale data are then used to report on potential loss of locally important features.

In a second application, emissions of volatile organic compounds from vegetation can exceed those from human-caused sources, such as cars. Thus they need to be accurately estimated before we can develop realistic and effective strategies for controlling human-caused sources. Models exist to calculate spatially explicit, short time-step emissions. The data required include weather, species-specific emission rates, and the total mass of leaves that are emitting. Knowing the composition of the land cover is critical because emission rates can vary by three orders of magnitude among species. For the Southern California Ozone Study, Arthur Winer from UCLA and colleagues are testing how well the CA-GAP database predicts both the composition and structure of various cover types in modeling emissions. The CA-GAP land cover database includes data on dominant canopy species and, in southern California, on cover density classes. Dr. Winer’s group randomly sampled a small number of GAP polygons in San Diego with transects to measure composition and cover/volume and compared both the estimated composition, classification, and predicted emissions. Although far from perfect agreement in any of these parameters, the preliminary results appear promising, especially given the assumptions about cover that must be made in interpreting the GAP data. Much of the disagreement about species was between taxonomically related species with similar emissions rates.

The third application developed a systematic protocol to identify potential sites for research natural areas (RNAs) for the U.S. Forest Service. The current methodology typically uses a screening approach to delete areas that do not meet the selection criteria and look at what is left. We felt this approach was too subjective and did not help choose among sites when there were many that passed the screening step. We developed a procedure that relied on GAP land cover and land ownership data. The first step is to identify the cover types that should be represented in RNAs. GAP data provides a regional context that determines which types are most dependent on national forest lands. For the types identified as target types on this basis, the second step is to measure the variability of the environments in which they occur. The third step calculates a site suitability for each target type to identify which sites are most typical for that type. Then a site selection model is used to select alternative sets of sites based on different weighting of the importance of the suitability rankings. This model is designed to find efficient solutions to represent the most cover types in a given number of sites while considering the suitability of the sites as well. This "weighted-benefits maximal covering location problem" model has also been applied using the CA-GAP wildlife data by 7.5’ quadrangle for the entire state of California.

The URL for the CA-GAP home page is http://www.biogeog.ucsb.edu/projects/gap/gap_home.html. The web site contains links to the final report, the GIS database for downloading, and information about the CD-ROM.

New Publications:

Stoms, D.M., M.I. Borchert, M.A. Moritz, F.W. Davis, and R.L. Church. 1998. A systematic process for selecting representative research natural areas. Natural Areas Journal 18:338-349.

Stoms, D.M., M.J. Bueno, F.W. Davis, K.M. Cassidy, K.L. Driese, and J.S. Kagan. 1998. Map-guided classification of regional land-cover with multi-temporal AVHRR data. Photogrammetric Engineering and Remote Sensing 64: 831-838.

Stoms, D.M., F.W. Davis, K.L. Driese, K.M. Cassidy, and M.P. Murray. 1998. Gap analysis of the vegetation of the Intermountain Semi-Desert Ecoregion. Great Basin Naturalist 58: 199-216.

<%Case "CO"%>

Colorado

Land cover: The land cover map is undergoing accuracy assessment under direction of Co-Principal Investigator Dr. William Reiners through work of Ken Driese of the botany department, University of Wyoming. Sixty percent of the initial air videography sample frames for Colorado have been interpreted by Tom Owens. Work on Wyoming frames is anticipated to begin in early 1999. Copies of the land cover base line with draft metadata have been provided to CO-GAP and National GAP cooperators for use, review, and comments.

Animal modeling: Vertebrate modeling outputs were distributed to CO-GAP cooperators for review from July through November 1998. Vertebrate models were finalized in November 1998, and final stewardship/species cross-tabulations were generated for use in completion of the CO-GAP chapter on "Analysis Based on Management Status." Data models (Habitat Affinities Modules) and model map outputs will be moved to the web in early 1999.

Land stewardship mapping: Land stewardship mapping was finalized in September 1998 by CO-GAP team member Larry Satcowitz by incorporating stewardship/status information on proposed Bureau of Land Management (BLM) Wilderness Study Areas. Status categories were reviewed with BLM personnel. Draft metadata were prepared for the land stewardship coverage and distributed for review by cooperators.

Analysis: Gap analysis was conducted on ArcView Spatial Analyst by Lee O’Brien of the CO-GAP team. Vertebrate models for 597 species were generated and cross-tabulated against the land stewardship coverage. Tables reformatted through Excel spreadsheet macros were generated for the final report.

Reporting and data distribution: The final report for CO-GAP has been developed through fall 1998, with draft chapters being routed to CO-GAP team members for review. Final chapters were sent to cooperators for review in December 1998. CO-GAP land cover and land status information has already been integrated into the Natural Diversity Information Source (NDIS), being developed by the Colorado Division of Wildlife (CDOW) in collaboration with the Natural Resource Ecology Laboratory at Colorado State University, Colorado Natural Heritage Program (CNHP), and others. The development of NDIS (http://blueberry.nrel.colostate.edu/ndis) is jointly funded by CDOW and Great Outdoors Colorado with matching contributions from the Denver Museum of Natural History, Rocky Mountain Elk Foundation, CNHP, and others. Internet access to CO-GAP coverages and analysis outputs will also be made available in winter and spring 1998/99.

Other accomplishments and innovations: The vertebrate modeling parameters for CO-GAP were developed into Habitat Affinities Modules in a Visual dBase environment for distribution to cooperators for their review and comments. These modules will be moved to a CO-GAP home page for access and download to encourage ongoing review and refinement of the modeling efforts. CO-GAP models will be provided to NDIS for consideration in continuing development of web site content. Colorado is looking forward to National GAP-inspired efforts to "regionalize" GAP modeling.

<%Case "CT","MA","RI"%>

Massachusetts, Connecticut, & Rhode Island

Land cover: An accuracy assessment of the land cover map completed in 1997 is now under way. Preliminary results indicated that there were significant classification errors, especially in the Cape Cod region. An error model is being developed from the accuracy assessment project, and a revised land cover map will be developed from this error model. Additionally, development of a new land cover map is planned within the next 18 months as part of an NSF-funded project of the Department of Computer Science at the University of Massachusetts (UMass).

Animal modeling: With completion of the expert review of mammal range maps during summer 1998, all vertebrate models are complete. Predicted habitats for all 273 vertebrate species modeled in the Southern New England region (MA, CT, RI) were identified and coarse species richness maps developed for each taxa group. Additionally, habitats were identified throughout the Connecticut River watershed for priority species of neotropical migratory birds. We plan to redo the vertebrate habitat maps once the revised land cover map is available.

Land stewardship mapping: All conservation lands in the region are mapped and classified according to conservation status. The database for Connecticut was not as well developed as for Massachusetts and Rhode Island. Over 15% of Southern New England is classified as conservation lands, and about 7% of the land area was classed in categories 1 & 2.

Analysis: Species richness analyses from the general maps have been completed. The final gap analysis should be complete by January 1999.

Reporting and data distribution: All data layers are currently available on the Southern New England Gap Analysis home page (http://tove.fnr.umass.edu/gaphome.html). We plan to distribute the vertebrate models and predicted habitat data layers as an ArcView project. The final vegetation map will not be distributed on CD until revisions are complete. A manual for incorporating GPS-logged aerial videography into land cover mapping efforts is under development and will be distributed on CD-ROM.

Other accomplishments and innovations: The National Science Foundation funded a $1.6 million collaborative project with the UMass Computer Science Department to develop an automated process for producing 3-D terrain maps. The project utilizes context-sensitive processing and machine learning via incremental real-time classification. This is a three-year project that will develop new approaches for using a variety of digital data platforms to produce high-resolution land cover maps.

<%Case "DE", "MD", "NJ"%>

Maryland, Delaware, & New Jersey

Land cover: During 1998, significant progress was made on land cover mapping. All of the TM scenes for Maryland and Delaware have been re-registered; all ancillary data have been collected (land cover, STATSGO soils, National Wetlands Inventory, slope, elevation, and aspect), rasterized, and registered to the TM scenes. The decision rules have been written for two scenes, and interpretation is under way for the remainder of Maryland and Delaware.

This project is using MicroImages’ TNTmips for image processing. In the fall of 1998, MicroImages was contracted to write a script in their native programming language to facilitate the image classification following the methods of Slaymaker and others (1996). The script is completed, and classification is under way. The Delmarva Peninsula will be completed shortly after the first of the year. The remainder of the Maryland-Delaware portion will be completed in spring of 1999, and accuracy assessment will begin at that point.

Late in 1998 an agreement with New Jersey’s Department of Environmental Protection, Natural and Historic Resources Group, and The Nature Conservancy was finalized to facilitate the GAP land cover and stewardship efforts in New Jersey. Work is expected to begin following completion of the draft work from Maryland and Delaware in late spring and be completed late in 1999.

Contact: Ann Rasberry (see State Projects for contact information)

Animal modeling: Considerable progress was made in 1998 in modeling vertebrate distributions for the mid-Atlantic GAP Project. A previously completed database of species’ distributional limits has been incorporated into the modeling process for Maryland and Delaware, and significant progress has been made toward the completion of a similar database for New Jersey. Investigators are nearly finished compiling literature reviews and summary documents of species-habitat relationships for approximately 200 species of breeding birds, 70 species of mammals, and 110 species of reptiles and amphibians. A relational database has been developed to store range information and species-habitat relationships. This database also has query and modeling tools to extract this information and apply it to spatial GIS layers in generating distribution models. A currently available general land cover data set is being used in preliminary modeling efforts and will be replaced with the final GAP land cover data set once it is completed. The modeling process has been effectively automated, with the database containing adjustable parameters for controlling habitat suitabilities and weighting of habitat layers. It is anticipated that all the bird and mammal models will be completed by the end of 1998, and the reptile and amphibian models completed by February 1999.

One significant challenge in the modeling process was to effectively model forest-interior bird species. This subclass of species has a strong negative response to forest fragmentation; distribution models based on habitat alone, irrespective of landscape-level parameters measuring forest fragmentation, considerably overestimate species distribution. This subclass includes many species in decline in this region, including a number of neotropical migratory songbirds (see Robbins et al. 1989). GAP investigators developed a number of GIS layers to measure landscape-level metrics of forest fragmentation and, in consultation with research personnel at the Patuxent Wildlife Research Center, applied these layers to the modeling process. The accuracy of models has been greatly improved. In addition, further analysis has enabled investigators to identify important forest stands for forest-interior birds, priority restoration sites, and potential additional factors limiting populations (see below).

Additional data tables have been developed with information on wildlife habitats and plant community alliances for the mid-Atlantic region. GAP investigators updated the list of alliances for the three-state region from documents developed by The Nature Conservancy and state Natural Heritage Program ecologists; these are all in an ongoing improvement mode. A list of wildlife habitats, which includes closely related alliances or aggregations thereof (and may better represent habitats perceived by wildlife species), has also been developed from the wildlife literature. These classifications, in addition to the FGDC classification standard, have been fully cross-linked in the database. Species distribution models can be applied at any classification level based on these cross-links.

Contact: Rick McCorkle or James Gorham,
Fish & Wildlife Biologists
Delaware Bay Estuary Project,
U.S. Fish & Wildlife Service
richard_mccorkle@fws.gov, (302) 653-9152 x17
james_gorham@fws.gov, (302) 653-9152 x12

Land stewardship mapping: GAP investigators conducted a land stewardship data development meeting during the summer of 1998 at which a process was outlined for completing this data layer. Management plans were obtained for several of Delaware’s Wildlife Management Areas, and contacts were identified for future interviews regarding management. An updated GIS layer of boundaries was obtained for all federal, state, and NGO lands as well as conservation easements in Delaware.
Most of the conservation area boundaries in Maryland have been collected. A letter has been sent to all managers of federal and county properties in the state requesting a copy of the area management plan. Maryland Department of Natural Resources (DNR) is responsible for developing area plans for all state lands, which are being obtained. Staff are developing a check list for reviewing the area plans, which will also be used to interview area managers where no management plan exists. Once completed, this process will expand to Delaware. The land stewardship layer is expected to be completed by spring 1999 for Maryland and Delaware and by late fall 1999 for New Jersey.

Analysis: Considerable analysis is under way for identifying areas with high potential for supporting forest-interior neotropical migrant songbirds on the Delmarva Peninsula, as well as priority sites for forest restoration and/or acquisition. Preliminary analyses indicate that few remnant forest stands exist which meet area and habitat requirements to support an intact forest avifauna. These stands have been identified and are currently being evaluated for ownership and land management status. In addition, analysis has been performed to identify clearings, which, if reforested, would substantially increase suitability for forest-interior birds. Maps and presentations of this material are currently being compiled for various audiences, including the Delaware Department of Natural Resources and Environmental Control, the Delaware Open Space Council, the Delmarva Conservation Corridor Initiative, and various land conservation groups.
GAP investigators expect to complete analysis of the protection and management status of biodiversity in Maryland and Delaware by fall 1999 and expect the New Jersey analysis to be completed by December 1999.

Reporting and data distribution: The final report writing for Maryland, Delaware, and New Jersey has begun and should be completed by December 1999. The GIS layers for Maryland and Delaware will be distributed in ARC/INFO ArcView format, with associated data tables in Microsoft Access format, on CD-ROM. A similar approach will probably be used in New Jersey.

Other accomplishments and innovations: By developing our vertebrate modeling scripts in ArcView’s Avenue, we have provided both a user-friendly interface for accessing data tables and a custom control over the modeling process. Users can query for species-habitat relationships or maps of species distributions, or survey the base GIS layers used in the modeling; lists of species occupancy can also be generated from input of habitat information. In addition, users can query the community alliance and wildlife habitat tables, generating a list of potential alliances/habitat by input of plant species, location, or physical characteristics of the site. Efforts are under way to make this software package portable for eventual distribution of the GAP data sets.

Literature Cited:

Robbins, C.S., D.K. Dawson and B.A. Dowell. 1989. Habitat area requirements of breeding forest birds of the middle Atlantic States. Wildlife Monographs 103:1-34.

Slaymaker, D.M., K.M.L. Jones, C.R. Griffin and J.T. Finn. 1996. Mapping deciduous forests in southern New England using aerial videography and hyperclustered multi-temporal Landsat TM imagery. Pages 87-100 in J.M. Scott, T.H. Tear, and F.W. Davis, editors. Gap Analysis: a landscape approach to biodiversity planning. American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland.

<%Case "FL"%>

Florida

Land cover: Land cover classification from 1993/94 Landsat Thematic Mapper imagery is complete for peninsula Florida from the keys up to and overlaying Georgia. There are currently 102 thematic classes, although some of these classes will be aggregated in the final product. The two Landsat scenes of the Florida panhandle will be completed in winter/early spring 1999. These scenes will complete the state.

Animal modeling: Habitat affinity matrices have been completed, and all the taxonomic groups except birds, reptiles, and amphibians have been reviewed. Review of the latter will be completed in the early spring of 1999. During the next 12 months the habitat matrices will be updated with the current standard taxonomic codes and matched to the final land cover classification. Previously completed and tested models are awaiting the final animal matrices and land cover for the production of a spatial habitat database and biodiversity models.

Land stewardship mapping: The Florida Natural Areas Inventory (The Nature Conservancy Heritage Program) and the Geoplan Center at the University of Florida have independently compiled GIS coverages of conservation land for the state of Florida. Our analysis of the two coverages has shown very good overall agreement. In the next 12 months we will combine the two coverages to include areas missing in one or the other and coding each of the conservation areas to GAP standard rankings based on the level of conservation protection.

Analysis: Data production will be completed in spring 1999, and final analysis will follow GAP standards. Final analysis will be completed in the next 12 months.

Reporting and data distribution: Reporting and data distribution will follow GAP standards. A final report will be prepared within the next 12 months.

Other accomplishments and innovations: Aerial land cover sampling with a digital color infrared camera has been used to assist the labeling of unsupervised classifications. Noncontinuous transects have been flown for all of north Florida at a spatial resolution of between 3 and 5 cm. A navigation system directs the pilot and triggers the camera while collecting a real-time corrected GPS position and pitch, roll, and heading information. Positional error of the images is typically less than 30 m; however, recent improvements in the camera setup have resulted in errors of less than 20 m. Doctoral research is under way to examine the potential for automated computer recognition of invasive plant species using the digital camera imagery.

<%Case "GA"%>

Georgia

Land cover: We are beginning our mapping this year. We have recently received statewide TM coverage of Georgia for the years 1996-1998. Our goal is to create a general land cover map of the state, similar to an Anderson level II classification, before generating an alliance-level map during the next year. We are in the process of gathering existing vegetation point data from cooperating governmental agencies and private organizations.

Animal modeling: We will be starting to collect data and literature this winter. We anticipate hiring a graduate student in the fall of 1999 to begin working on these models.
Land stewardship mapping: We anticipate completing our land stewardship map this winter. We have included Army Corps of Engineers lands, wetland mitigation sites, and conservation easements held by land trust organizations. This work was partially funded by EPA Region 4.

<%Case "HI"%>

Hawaii

The Hawaii Ecosystem Data Group (EDG) will be conducting HI-GAP. The group includes the Fish and Wildlife Service, The Nature Conservancy, the Biological Resources Division of the USGS, the University of Hawaii’s Center for Conservation Research and Training, Hawaii’s Department of Land and Natural Resources, and the Hawaii Ecosystem at Risk program. A workshop is planned for March 1999.

<%Case "ID"%>

Idaho

Land cover: The latest Idaho land cover classification was completed in December 1998. It was created through the joint efforts of the Wildlife Spatial Analysis Lab at the University of Montana, the remote sensing/GIS labs at Utah State University, and the Idaho Gap Analysis Lab. Vital statistics: 80 land cover types, 30-m resolution, 2-ha minimum mapping unit. Estimated accuracy ranges from 53.4% to 93.39% in North Idaho and 63.6% to 79.3% in southern Idaho. The vegetation raster grid, the accuracy data, and a draft of the vegetation section of the Idaho GAP final report can be downloaded from www.wildlife.uidaho.edu/idgap.htm.

Animal modeling: Wildlife habitat relationship (WHR) models have been created for breeding birds and mammals of Idaho and sent out for expert review. We are currently incorporating reviewer comments and revising the models. Dr. Charles Peterson of Idaho State University is creating WHR models for Idaho’s amphibians and reptiles. The estimated date of model completion for all species is March 1, 1999. The accuracy assessment of our models should be completed by April 1, 1999.

Land stewardship mapping: The updated Idaho land stewardship coverage is nearing completion. Although it is based on the original Idaho GAP stewardship map, this coverage shows substantial improvement in detail for land with protection status 1 and 2. The expected date of completion is February 1, 1999.

Analysis: The ARC/INFO AML programs to conduct the gap analysis of Idaho have been written. We are awaiting completion of the Idaho land stewardship coverage to begin analysis of the land cover types. Analysis of the WHR models will be conducted as the models are completed. Expected date of completion for the analysis phase is April 1, 1999.

Reporting and data distribution: Section 1 of the final report, the land cover of Idaho, has been completed. It is available from the Idaho GAP web site at www.wildlife.uidaho.edu/idgap.htm. Other chapters are in progress. As products are completed, they will be available on the Idaho GAP web site.

<%Case "IL"%>

Illinois

Land cover: A general land cover classification for the state was completed in October 1995. This classification identified 19 land cover classes: four urban, three forest and woodland, three agriculture, two grassland, five wetland, and two other categories (water and barren areas). The GAP vegetation classification is being performed for separate natural cover elements of the original imagery using a Boolean mask for specific classes (forests, forested wetland, and rural grassland). Classification protocols are similar to protocols for UM-GAP (see GAP Bulletin 5:35). Classification to the community/alliance level has been completed for southern Illinois. Our efforts will next concentrate on the western side of the state along the Mississippi River and then move on to the remainder of the state.

Animal modeling: We have created a list of amphibian, reptile, and mammal species to be mapped. We are using specimens collected by the Illinois Natural History Survey (INHS) and the University of Illinois Museum of Natural History to obtain locational records for each species. There are approximately 27,000 specimens in the amphibian and reptile collection for Illinois of which we have mapped about 20,000. The mammal collection contains about 12,000 specimens of which we have mapped 95%. Once we finish mapping, we will delineate ranges for each species and conduct an expert review of the range maps. Furthermore, we will start building habitat association information for each species. Information gathered previously for the Illinois Fish and Wildlife Information System will be helpful in developing habitat associations. We will use the breeding bird survey for Illinois to create a list of bird species to be mapped and begin delineating ranges for those species.

Land stewardship mapping: We have developed a land stewardship map for Illinois, attributed general ownership categories, and assigned management status levels. The GAP coding scheme for land units needs to be determined and assigned for each property.

Analysis: We have completed some preliminary analyses using locational data for amphibians, reptiles, and mammals to create species richness maps using the EMAP hexagons. We will continue to do more analyses as our species and vegetation mapping progresses.

Reporting and data distribution: We hope to start writing segments of the report at the end of 1999. The INHS has a substantial amount of information on plants and animals in Illinois available on its web pages. Efforts are under way to develop an IL-GAP web site and link it to related information already on the web. In particular, web pages identifying the vegetation alliances in Illinois and associated distribution maps for southern Illinois have been developed.

Other accomplishments and innovations: Referenced below are three projects that have been completed by GAP participants using the land cover database of Illinois in the analysis.

McKinney, L.B., M.G. Joselyn, J.L. Aycrigg, and P.W. Brown. 1998. Identification of large grassland ecosystems in Illinois. Final report. Illinois Natural History Survey, Champaign, Illinois. 169 pp.

Suloway, L., M. Joselyn, P.W. Brown. 1996. Inventory of resource-rich areas in Illinois: An evaluation of ecological resources. Illinois Department of Natural Resources/EAA-96/08. 167 pp.

Van Deelen, T.R., L.B. McKinney, M.G. Joselyn, and J.E. Buhnerkempe. 1997. Can we restore elk to southern Illinois? The use of existing digital land-cover data to evaluate potential habitat. Wildlife Society Bulletin 25(4):886-894.

<%Case "IN"%>

Indiana

Land cover: Land cover mapping and accuracy assessment for the Indiana project are complete. We are finalizing metadata. Draft data are available by contacting the Principal Investigator. Goals for the next 12 months are to work with partners to improve the thematic detail and accuracy of the Indiana land cover product.

Animal modeling: The vertebrate models are complete, including detailed models for some edge-associated, riparian-associated, and wetland-associated species. Vertebrate models are running as of November 1998. Goals for the next 12 months include making the models widely available for evaluation by interested users in Indiana.

Land stewardship mapping: Land stewardship mapping is complete. Goals for the next 12 months include working with the Indiana Department of Natural Resources and other agencies to update and add additional categories (e.g., Wetland Reserve Program lands) to the coverage.

Analysis: Analysis is under way as of November 1998. Goals for the next 12 months include using the results of the analysis and conducting additional analyses to facilitate biodiversity conservation in Indiana. The primary mechanism for this will be the Indiana Biodiversity Initiative.

Reporting and data distribution: These are under discussion by the Steering Committee in Indiana. The standard report will be issued, and data will likely be available on CD-ROM and on the web. Preliminary data are already being made available.

The Indiana GAP project continues to conduct metaprojects using the GAP methodologies and infrastructure. Three projects in progress are: development of the environmental assessment for the proposed Grand Kankakee Marsh National Wildlife Refuge; support for the U.S. Fish and Wildlife Service Ecosystem Management Initiative through development of a GIS to support the Migratory Bird Resource Priority of the Ohio River Valley Ecosystem Team; and support for the Indiana Biodiversity Initiative, a multi-organization effort to protect biodiversity in the context of an economically productive Indiana landscape.

<%Case "IA"%>

Iowa

The Iowa Gap Analysis Project (IA-GAP) is in its second year. An IA-GAP home page is accessible at http://www.ag.iastate.edu/centers/cfwru/iowagap/

Land cover: The Iowa Geological Bureau completed the first phase of mapping the land cover of Iowa into six basic classes: trees, grassland, cropland, water, urban, and barren. Mapping was facilitated by ancillary data from the Natural Resources Inventory provided by USDA-NRCS. (Contact: Jim Giglierano, Iowa DNR, Geological Survey Bureau, 109 Trowbridge Hall, Iowa City, IA 52242, Phone: 319-335-1594, jgiglierano@igsb.uiowa.edu).

Following the FGDC’s standard vegetation classification system, a list of 75 natural and seminatural vegetation alliances were identified as existing in Iowa. This list was reduced to about 35 map classes, including agricultural categories, that potentially can be mapped in phase 2 of the land cover classification. Extensive field data were collected in summer 1998 for use in cluster identification and labeling. Tom Rosburg, Drake University, submitted more than 500 sites with rapid vegetation assessments. Additional assessments were received from county and state biologists. We continue to seek and acquire ancillary data sets that will assist in image processing and land cover mapping.

A program developed by the U.S. Fish and Wildlife Service for aggregating National Wetland Inventory codes into five major classes (temporary, seasonal, semipermanent, permanent, and open water) was modified and used to prepare a GIS layer for Iowa wetlands. The aggregation process was completed in December 1998 for all 99 Iowa counties, and the resulting layer will be used as the primary ancillary data set for wetland mapping. Currently all data are being merged, and protocols are developed and tested for the satellite scene that covers the northeastern corner of the state.

The IA-GAP staff continues to meet biannually with Kansas, Missouri, Nebraska, South Dakota and North Dakota to share data and information on polygon edge-matching, legend compatibility, accuracy assessment, and other problems that are common to GAP in the Great Plains Region. GIS coordinators from Iowa, Kansas, Missouri, and Nebraska met with Iowa State University statisticians in July 1998 to discuss a uniform plan for accuracy assessment of the final land cover maps in the four-state region. General consensus was reached on protocols and data analysis.

In October 1998 IA-GAP held a one-day workshop for approximately 80 participants to inform cooperators of the status of the project and receive feedback.

Vertebrate modeling: Lists of vertebrate species for which distribution models will be developed have been completed and are available on the IA-GAP web site. The lists include 178 species of birds, 65 mammals, 48 reptiles, and 21 amphibians. A digital copy of the recently completed Iowa Breeding Bird Atlas was obtained. Museum records and other ancillary data for mammals, reptiles, and amphibians are currently being compiled and consolidated into a single database. Geographic ranges, habitat associations, and modeling criteria are in early draft stages.

Land stewardship: Federal land coverages have been acquired but not checked. Data for state lands in Iowa are mostly complete and digitized. Boundaries for about one-half of county lands have been obtained. Stewardship status codes have been assigned for approximately 20% of the areas.

NatureMapping: Iowa State University has received an educational grant from the Iowa Resource Enhancement and Protection program to start a statewide NatureMapping project. A steering committee was formed, and workshops are being planned for winter 1998 and spring 1999.

<%Case "KS"%>

Kansas

Land cover: Currently, 85% of the state has been mapped to the cropland/natural vegetation stage, and 12% has been classed to the alliance level. During the next 12 months, three tasks will be undertaken by the land cover mapping group. First, field data collected during the summer of 1998 will be entered into a relational database to serve as input to the alliance vegetation mapping effort. Second, the 15% of the state remaining to be classed to the cropland/natural vegetation stage will be completed by January 1999. Lastly, a substantial portion of the remaining 88% of the state needing to be classed to the alliance level will be completed. A graphic demonstrating current progress is available on the web version of the Bulletin.

Animal modeling: Currently, species lists of terrestrial vertebrates to be mapped are being finalized following expert review. Range distribution maps for EMAP hexagons are being generated, based on museum collections, publications, and current observations. We are beginning to build wildlife habitat relationship (WHR) models. Ancillary databases also are being located to help in building these predicted habitat relationship models. During the next 12 months, all WHR models will be completed. Drafts of predicted species distributions will be completed and sent out for expert review as portions of the land cover map become available. Following expert review, final species distribution maps will be generated.

Land stewardship mapping: Currently, 90% of the land stewardship layer has been updated using 1-m digital orthophotographs; editing and correcting protected areas coverages will continue over the next 12 months.

Analysis: We do not expect to start analysis within the next 12 months.

Reporting and data distribution: Reporting and data distribution will not be initiated within the next 12 months.

<%Case "KY"%>

Kentucky

Contacts:

Keith Wethington (land stewardship mapping)
Kentucky Department of Fish and
Wildlife Resources, Frankfort
Keith.Wethington@mail.state.ky.us,
(502) 564-7109 x405

Tom Kind (land cover)
Mid-America Remote Sensing Center, Murray State University
tckind@bach.murraystate.edu, (502) 762-3110

Terry Derting (vertebrate modeling)
Department of Biology, Murray State University
terry.derting@murraystate.edu, (502) 762-6327

Land cover: To map the land cover of Kentucky we are using an iterated unsupervised classification procedure to produce a base layer for use in land cover modeling. Satellite scenes have been mosaicked into pathwise coverages that are the fundamental classification and mapping elements. A spectral classification has been completed on one of the four paths covering the state, with work well under way on the second path. In keeping with a consensus reached by the Southeast GAP working group, scenes were stratified according to Bailey’s ecoregions, and a separate spectral classification was produced for the deciduous vegetation in each region. A complete set of 1:24,000 DEM coverages have been acquired from the USGS and mosaicked to match our pathwise classifications.

A literature review was done to establish vegetation/landform relationships with ancillary data sets, including DEMs and derivatives, a concavity/convexity model, ecoregional analysis using Bailey’s ecoregions, National Wetlands Inventory (NWI) digital data, and NAPP color infrared photography. In addition, new aerial videography was flown in fall 1998 to supplement the existing flightlines acquired in 1996. A draft of potential alliances and aggregations for the first satellite path has been completed and is available for inspection on our web site.

Major goals for the land cover mapping in year two include: developing a consistent land cover modeling procedure based on the available ancillary data and the results of the spectral classification, completion of a vegetation cover-type layer for our first two satellite paths, and continuation of the path-by-path classifications.

Animal modeling: The vertebrate component of KY-GAP is officially under way. A wildlife biologist was hired in September 1998 to assist with this component of the project. Initial work on the vertebrate layer has focused on the development of species lists for mammals, birds, reptiles, and amphibians that include species element codes, scientific names, common names, state and federal ranks and status, and references. Data sets made available from the Kentucky Department of Fish and Wildlife Resources (KDFWR) and the Kentucky Nature Preserves Commission (KYNPC) will be major sources of information for the vertebrate layer. Experts at the KDFWR, the KYNPC, and other state and regional institutions have made commitments to serve as reviewers for the vertebrate mapping. We anticipate completion of the species lists, a database of locational records, and delineation of geographic ranges for mappable species during the first 12 months of the vertebrate mapping component of the project. During 1999 we will also begin work on development of habitat relationships for species in each vertebrate group with the assistance of undergraduate research students. We are working closely with the staff of Tennessee and other GAP projects in the southeastern region to ensure compatibility of products for use in regional maps.

Other accomplishments and innovations: Major accomplishments during the first year of the program include: acquisition and processing of three TM data sets to be used in the land cover mapping; development of a methodology for organizing and classifying the satellite data and related ancillary data sets; and establishing cooperative partnerships with private, state, and federal organizations within Kentucky to help with the acquisition of reliable ground truth.

To continue to promote GAP to a wider audience, a presentation on Gap Analysis was given at the Kentucky GIS conference in May 1998. Yearly participation in this conference is anticipated as a way of keeping the Kentucky GIS community informed of our mission and our progress. Also, a web site has been developed to provide a general educational resource relating to Gap Analysis in Kentucky. In addition to providing a way of keeping abreast with our progress, the web site will provide an on-line method for expert reviewers to give their feedback. Development of this web site has been accomplished with the cooperation of the KDFWR.

<%Case "LA"%>

Louisiana

Contacts:

Jimmy Johnston (project leader)
USGS/National Wetlands
Research Center, Lafayette
jimmy_johnston@usgs.gov, (318) 266-8556

Steve Hartley (land cover, analysis)
USGS/National Wetlands
Research Center, Lafayette
steve_hartley@usgs.gov, (318) 266-8543

Pat O’Neil (land cover, analysis)
USGS/National Wetlands
Research Center, Lafayette
pat_oneil@usgs.gov, (318) 266-8699

Richard M. Pace, III (animal modeling)
Louisiana Cooperative Fish and Wildlife
Research Unit, Baton Rouge
rpace@lsu.edu, (504) 388-5747

Mark Swan (land stewardship mapping)
The Nature Conservancy, Baton Rouge
mswan@tnc.org, (504) 388-1040

Land cover: LA-GAP has completed the final vegetation map layer. The original 25-m vegetation layer has been resampled to a 30-m resolution to meet the National GAP standards. The layer was then aggregated to a minimum mapping unit of eight continuous pixels, using the GigaMerge software (see Ford et al. 1997, also see www.cs.umt.edu/MERGE/). The vegetation layer has been clipped to the 43 USGS 30 x 60-minute map series that makes up the state. The compilation of the ground-truth data into a database for accuracy assessment validation is complete. Accuracy assessment has been delayed due to errors found in the Imagine software for calculating accuracy assessments (for more on this problem see "Accuracy Problems in Imagine Software" on the GAP Bulletin Board/Notices). An ArcView Avenue program is being written to produce an accuracy statement. Ancillary data sets (STATSGO soil data, geology 1:500,000 scale, USGS [7.5-minute quadrangles, 30 x 60-minute tiles, roads, railroads, pipelines, hydrology], TM coverage, parish boundaries, state boundary, Wetland Reserve Program parcels, and National Wetlands Inventory data) are complete. The goal for winter 1998/99 is to complete the metadata for all ancillary data sets, complete the accuracy assessment for the land cover/land use layer, and complete the land cover section for the final report.

Animal modeling: Species distribution maps were resolved to EPA hexagons from faunal guides (mammals and herps) and survey data (birds). Location data for museum specimens collected in Louisiana were not compiled expressly for GAP. Wildlife habitat relationships (WHR) were extracted from literature, natural history guides, and expert advice. Vegetation classes used in WHR tables were cross-walked to LA-GAP cover classes, and first generation predicted distribution maps were developed. Due to low appeal of predicted distributions of herps and water birds, second-generation maps are currently being developed by incorporating a more extensive hydrology-related data layer into the modeling process.

Land stewardship mapping: In 1998 we added recently acquired properties, completed the attribution of all polygons to GAP standards, and merged several coverages into one coverage, as required by GAP. The latter presented difficulties in that many state wildlife management areas (WMAs), national wildlife refuges (NWRs), and national forests share boundaries, requiring the elimination of sliver polygons. By mid-1999 we will have:

1. updated our current polygons consisting of state parks, military bases, WMAs, NWRs, national forests, natural areas, and one national park;
2. added a large set of state lands (not parks or WMAs), Army Corps of Engineers properties, and some additional private lands widely advertised as preserves or wildlife areas, and scenic streams (either as lines or very narrow polygons);
3. completed metadata for the stewardship coverage;
4. completed the written chapter on Louisiana’s land stewardship and management.

Analysis: The analysis began in December 1998 and should be finished in late February 1999.

Reporting and data distribution: The land cover chapter is in progress and waiting for the accuracy assessment statement. Currently the National Wetlands Research Center is in the process of building a web site for the LA-GAP data.

Other accomplishments and innovations: see "Using GAP for nonpoint water pollution management".

Literature Cited:

Ford, R., R. Redmond, and S. Barsness. 1997. "Merge": Breakthrough software for user-defined MMUs. Gap Analysis Bulletin 6:22-23.

<%Case "ME"%>

Maine

Land cover: Completed. A document detailing the methods used, "Development and testing of a vegetation and land cover map of Maine," has been drafted and will soon be sent out for review.

Animal modeling: Completed. An M.S. student is conducting more detailed research on testing the predicted vertebrate occurrences, going beyond the review presented in the final ME-GAP contract report.

Land stewardship mapping: Completed. Another M.S. student is working with this and other GIS data layers to test whether or not Maine’s conservation lands have captured a fair representation of the state’s natural variability in selected abiotic and biotic factors.

Analysis: Completed.

Reporting and data distribution: A final report documenting the Maine Gap Analysis was sent out for review. All reviews were completed, and the final contract report was written. The first gap analysis of Maine was completed at the end of 1998.

<%Case "MI", "MN", "WI"%>

Michigan, Minnesota, & Wisconsin

(Upper Midwest Gap Analysis Project)

Contacts:

Daniel J. Fitzpatrick,
UM-GAP Coordinator
USGS Upper Midwest Environmental
Sciences Center
daniel_fitzpatrick@usgs.gov, (608) 783-7550 x12

Mike Donovan
Michigan Department of Natural Resources
Wildlife Division, Waterloo Office
donovanm@state.mi.us, (517) 522-4097

Dave Heinzen
Minnesota Department of Natural Resources
Resource Assessment Unit
david.heinzen@dnr.state.mn.us,
(218) 327-4449 x222

Tim Weiss
Wisconsin Department of Natural Resources
Bureau of Wildlife Management
weissn@dnr.state.wi.us, (608) 267-9428

Land cover: The Michigan Department of Natural Resources’ Wildlife Division is responsible for the land cover classification of the Lower Peninsula (LP) and is currently working with a contractor to classify the northern half of the LP. Completion is expected in early 1999; work on the remainder of the LP will continue in 1999. The vegetation map for the Upper Peninsula (UP) is currently being developed at the U.S. Geological Survey’s Upper Midwest Environmental Sciences Center (UMESC); completion of the eastern UP is expected in 1999; work in the western UP will continue in 1999. The lack of ground-truth information in the form of georeferenced digital data, including NWI, hinders more rapid progress in Michigan.

The Minnesota Department of Natural Resources (DNR), Division of Forestry, Resources Assessment Unit, is developing the land cover map for Minnesota. The DNR has contributed in-kind services, including the facilities, hardware and software, and field work. Most of the state classification is now completed-all but the accuracy assessment. With increased department support, the land cover mapping effort is expected to be completed by late 1999.

Wisconsin has been very successful in developing support for the current land cover mapping from various local, state, and federal agencies and organizations under the Wisconsin Initiative for Statewide Cooperation on Landscape Analysis and Data (WISCLAND), of which the Gap Analysis project has been a part. The image processing protocol was developed specifically for Wisconsin, though it has undergone modifications after piloting and input from Minnesota and Michigan. The Wisconsin land cover data layer is now completed. The remote sensing derived classification scheme will be cross-walked to the NCVS in early 1999 with assistance from TNC.

Animal modeling: Vertebrate modeling for MN, WI, and MI will begin in 1999 with UMESC coordinating the effort among the state partners. Dr. Nancy Mathews at the University of Wisconsin-Madison has developed a research protocol to investigate the effects of scale on species distribution modeling in the Upper Midwest. The three-year effort is to be funded through a cooperative multiagency agreement with the NRCS and the USFS. This study is scheduled to be completed in 1999.

Land stewardship mapping: The Minnesota DNR has completed and delivered a stewardship map for the state. Based on the Public Land Survey, every 40-acre polygon has been attributed for public owner, manager, unit name, and stewardship status. Tribal, private conservancy, and private land parcels in excess of 1,000 acres are also identified. Wisconsin and Michigan are in the planning stages of developing similar databases with work beginning in 1999.

Analysis: Gap analysis planning is scheduled to begin in late 1999.

Reporting and data distribution: UMESC will begin distributing Wisconsin land cover data and Minnesota stewardship data via the WWW in early 1999.

Other accomplishments and innovations: The Upper Midwest Image Processing Protocol is published and available on-line at: ftp://ftp.emtc.usgs.gov/pub/misc/umgap/98-g001.pdf. The Upper Midwest Gap Analysis Project will host the 1999 National GAP Meeting in Duluth, Minnesota, on August 2-6.

<%Case "MS"%>

Mississippi

Land cover: Much of the computationally intensive part of the vegetation mapping process was concluded during 1998. A large amount of time was spent creating and organizing image databases, transferring bitmaps, reprojecting imagery, preclustering analysis of imagery, and clustering image data. To handle and keep track of multiple image data sets, a GIS database was created using PCI for each Landsat scene. Within each database, image channels were created to store not only the spectral imagery but several working channels as well.

Several processes were used on each Landsat scene prior to statistical clustering procedures. In order to reduce spectral variability found among similar vegetation types across large areas (as found in Landsat imagery), we subset each scene into its representative physiographic provinces. The next step involved manually delineating recent timber harvest locations on the imagery. Sample classifications of the imagery revealed that these areas are spectrally similar to agricultural areas, and simple clustering algorithms could not separate the two classes. All imagery was then reprojected to align with our ancillary data sets.

As is common with Landsat imagery, there were clouds present in a few scenes. For one coastal scene it was decided to manually mask out the cloud cover prior to clustering. These areas were then patched using the data from the cloud-free date. Heavy cloud cover and haze were present in a second scene that covered the northwest corner of the state. For this scene we decided to only use the cloud-free date.

Once the data sets were organized and subset, K-means clustering was performed. Clustering was applied to bitmaps that met the following criteria: classified as pine, hardwood, mixed, agriculture, barren/grassland, or other in the land use/land cover map created by the Space Remote Sensing Center (SRSC) and not a recent harvest. This clustering was repeated for each physiographic province within each scene.

We recently received color infrared stereo aerial photography that will be used in the identification and labeling of the clusters. Having stereo triplets of photography will allow us to make the distinction among differing age and density classes of vegetation. Once the clusters are labeled, we will "burn in" a road/transportation network that was created. By burning in this layer and a marginal buffer, we hope to reduce the error of classification that is common along road edges. The road layer was created by buffering existing primary and city/county road vectors by a distance determined from measurements taken from aerial photography.

Approximately 25% of the stereo triplets are being scanned for registration to the Landsat classifications. These will be used for final verification of the cluster classifications of each scene. The remaining photography will be used prior to verification to aid in identification of cluster labels. These procedures have been documented in a master’s thesis by Susan Batten, "A Procedure for Creation of a Land Use/Land Cover Map of Mississippi."

Animal modeling: Range maps have been developed and reviewed by the guidance committees. Models have been developed for nearly all of the terrestrial species. The models have been tested on an Anderson level II land classification map for accuracy and reliability. Models are under review from the guidance committees for better application of ancillary data to achieve greater refinement and accuracy. Procedures have been developed to apply the current models to the final land cover product. All models will be finalized and run on the final land cover product within the next 12 months.

Land stewardship mapping: The Mississippi Department of Wildlife, Fisheries, and Parks (DWFP) has been instrumental in developing this layer. All public land holdings have been compiled from existing digital data or digitized by the DWFP using a GPS receiver. Pertinent information regarding land holdings are entered and checked for accuracy. Management status will be added after accuracy assessments are complete. This data set will be finished within the next 12 months. Private timberlands are available but will not be included because of the frequent shifts in owners.

Analysis: Analysis has not been started. All analyses will be conducted in the next 12 months.

Reporting and data distribution: Relevant parts of the final report have been continually updated since the project got under way. Currently, the methodology sections are being updated to correspond to the final work conducted on both the land cover and vertebrate modeling. The report will be updated for all sections, including analysis, within the next 12 months.

<%Case "MO"%>

Missouri

Land cover: Phase I land cover mapping was completed by the Missouri Resource Assessment Partnership (MoRAP) and is currently being used for vertebrate modeling. Accuracy assessment was completed by MoRAP for the Phase I product based on 15,000+ verification sites distributed statewide. MoRAP, through separate state funding, is pursuing a Phase II product utilizing the information gained from this assessment to build a better land cover base. This land cover database has been integral to the regional planning efforts of the Missouri Department of Conservation as well as activities of the U.S. Forest Service and other cooperators.

Animal modeling: Modeling has been completed by the Geographic Resources Center (GRC) for 321 terrestrial vertebrate species (148 birds, 64 mammals, 44 amphibians, and 65 reptiles). Modeling was conducted for each species using base land cover classes as well as derivative and ancillary databases. These included such databases as ecotones, interiors, contiguity, distance relations, special natural features, watershed breaks, soils, geology, temperature, rainfall, population density, road density, elevation, and wetlands. These were arithmetically combined to form a continuum of suitability across the state for each species, based on modeled inputs.

Land stewardship mapping: Stewardship mapping and coding was completed by MoRAP. Some maintenance of the database has been ongoing at GRC to provide a temporally more accurate picture upon final analysis.

Analysis: Actual gap analysis is being completed for the assessment of statewide biodiversity, stewardship base assessment, and other final product creation. Analysis was performed using a variety of grids, such as for linkage to the Public Land Survey Net, 1/6 quad (as used in the Breeding Bird Atlas for Missouri), 7.5-minute quadrangle, county, and EMAP hexagon spatial units. A risk analysis is slated to be performed to assess biodiversity issues in light of population growth trends, road development, agricultural trends, and demographic factors.

Reporting and data distribution: A first draft of the final Missouri GAP report will be delivered January 1999. Data distribution will occur through the Missouri Spatial Data Information Service (MSDIS) (http://msdis.missouri.edu). Techniques, AMLs, models, and other ancillary data and source information will be provided through the Mid-America Vertebrate Resources Information Consortium (MAVRIC)(http://msdis.missouri.edu/mavric). This information will be mirrored at the Missouri GAP home page (http://msdis.missouri.edu/mogap). This will be completed within one month of final acceptance of the report by National GAP.

Other accomplishments and innovations: The first statewide effort to include crayfish, fish, mussels, and snails in gap analysis is now under way in Missouri. For more information see this.

<%Case "MT"%>

Montana

Land cover: complete (see map)

(Click for more detail)

Animal modeling: complete

Land stewardship mapping: complete (see map)

(Click for more detail)

Analysis: complete

Reporting and data distribution: complete (available for distribution by January 1999)

Other accomplishments and innovations:

1. We were one of the first states to develop a digital process for mapping existing vegetation from Landsat TM imagery. In contrast to traditional pixel classifications, which cannot be applied in a GIS over large geographic areas because of file size limits, our approach involves an unsupervised classification of pixels, followed by pixel aggregation to a user-defined map unit, assembly of a GIS database, and then a supervised classification of the raster polygons. Several manuscripts describing this process are in review:

Ma, Z., M.M. Hart, and R.L. Redmond. Using a color palette to drive unsupervised classifications of remotely sensed images: What you see is what you get.

Ma, Z., M.M. Hart, R.L. Redmond, J.C. Winne, and T.P. Tady. Mapping vegetation across large geographic areas: Integration of remote sensing and GIS to classify multisource data.

Winne, J.C., R.L. Redmond, M.E. Jensen, M.M. Hart, J.P. DiBenedetto, and I.A. Goodman. Integrating satellite imagery and potential vegetation to delineate pattern and map existing vegetation in a mixed-grass prairie ecosystem.

Steele, B.M., and R.L. Redmond. A method of exploiting spatial information for improving classification rules: Application to the construction of polygon-based land cover maps.

2. We developed methods to predict and then to map the magnitude and distribution of classification errors. This work was published in:

Steele, B.M., J.C. Winne, and R.L. Redmond. 1998. Estimation and mapping of misclassification probabilities for thematic land cover maps. Remote Sensing of Environment 66: 192-202.

3. Using the land cover database, we reanalyzed U.S. Department of Commerce data from the 1990 census for western Montana and northern Idaho. Actual counts of people from both census blocks and block groups were partitioned to new spatial units based on land cover, land ownership (public and corporate timberlands vs. other private holdings), and topography. The result is a GIS database containing human population density at a finer scale than is normally available. Other census variables can also be mapped using similar dasymetric techniques. The initial work was published in:

Holloway, S.R., J.V. Schumacher, and R.L. Redmond. 1998. People and place: Dasymetric mapping using ARC/INFO. Pages 283-291 in S. Morain, editor. GIS solutions in natural resources management: Balancing the technical-political equation. High Mountain Press, Santa Fe, New Mexico.

4. Statewide atlases of land cover and terrestrial vertebrates were prepared in digital form (Adobe Acrobat format) for ready electronic distribution.

<%Case "NE"%>

Nebraska

Land cover: We have completed a preliminary statewide classification and are in the process of refining the product by major watershed to reach the vegetation alliance level. We have initiated planning for accuracy assessment. Plans for next 12 months: complete initial alliance-level land cover mapping, continue field work for land cover mapping, and conduct pilot study for regional accuracy assessment protocol (see below).

Animal modeling: We have initiated work on assembling species lists and are conducting an extensive literature review including acquisition of range maps. We continue to work with the Nebraska State Museum to determine how we can best use their collection records. Plans for next 12 months: complete species lists, assemble and digitize range maps, automate bibliographies, and identify preliminary models for vertebrate models.

Land stewardship mapping: This product is nearing completion. We continue to work extensively with the Nebraska Game and Parks Commission in this endeavor. Plans for next 12 months: complete stewardship database.

Analysis: Plans for next 12 months: preliminary/pilot testing of methodology.

Reporting and data distribution: Metadata assemblage and documentation ongoing.

Other accomplishments and innovations:

1. Completed assessment of bighorn sheep habitat in northwest Nebraska under contract to the Nebraska Game and Parks Commission;
2. completed inventory of national GAP land cover mapping procedures for USGS/BRD/GAP (see www.calmit.unl.edu/gapmap);
3. continued to work with regional partners in Great Plains GAP consortium (see http://www.calmit.unl.edu/midam/ and http://msdis.missouri.edu/mavric/);
4. worked with regional partners to develop regional land cover accuracy assessment protocols (to be partly funded by EPA Region 7).

<%Case "NV"%>

Nevada

Data sets are complete. Analyses are under way. Update has been initiated as a regional project (see Southwest Regional Breakout Session on p. 64 of GAP Bulletin No. 6)

<%Case "NM"%>

New Mexico

NM-GAP data will be available on CD soon. Update has been initiated as a regional project (see Southwest Regional Breakout Session on p. 64 of GAP Bulletin No. 6)

<%Case "NH", "VT"%>

Vermont and New Hampshire

Land cover: The land cover map for New Hampshire is complete, and accuracy assessment is finished. Edge-matching along the border with Maine is acceptable. Vermont’s land cover map also is complete, and we are arranging to edge-match with New York.

Animal modeling: Models are complete pending occasional revisions.

Land stewardship mapping: Mapping is complete except for numerous updates on federal and state lands. Land stewardship categories have not been assigned.

Analysis: Predictive habitat models are being run in late 1998, with other aspects of analysis to follow in early 1999.

Reporting and data distribution: The goal is to have draft reports written by July 1999 and final reports submitted by the end of September.

Other accomplishments and innovations: Preliminary products from the Vermont and New Hampshire GAP Project have been used in ecological reserve selection projects for the two states. In New Hampshire, priority areas for ecological reserves were selected after a watershed-based analysis of landscape diversity, an identification of core forest sites from GAP land cover maps, and an analysis of data from the Natural Heritage database were made. In Vermont a similar project is under way, but the analyses are more sophisticated and rely heavily on landscape features that are linked to biological diversity.

<%Case "NY"%>

New York

Land cover: The statewide land cover map was completed in February 1998. The map is based on 240 spectral clusters derived from six spectral bands of single-date, Landsat-5 Thematic Mapper data acquired between 1991 and 1993. The clusters were labeled into one of 45 land cover types using field knowledge and observations, edaphic factor data, aerial photo interpretation, and expert consultations. The classification was based on the National Vegetation Classification System and correlated with ecological community classification systems of The Nature Conservancy and the New York Natural Heritage Program.

Accuracy assessment and refinement of cluster label assignments were conducted between March and December 1998. Accuracy assessment was conducted using a stratified random sample of 109 circular, 1600-ha plots centered on the coordinate position of the start of a Breeding Bird Survey route. Each plot contained between 40 and 200 unlabeled polygons representing predicted land cover types. These polygons were labeled by field crews, based on field observations or interpretation of NAPP 1:40,000 color infrared aerial photos. A total of 10,154 data points was used for traditional accuracy assessment. A subsample of 934 polygons from 37 plots was used to measure the "fuzzy" accuracy of the land cover map, using a linguistic scale to determine how well each land cover type was predicted from satellite-based spectral information. In addition, 12 on-site, structured interviews with more than 75 wildlife biologists, land managers, and administrators from the New York State Department of Environmental Conservation (NYSDEC, our principal state cooperator), Adirondack Park Agency, and U.S. Fish and Wildlife Service allowed us to identify and correct major cluster labeling errors throughout various regions of the state.

Habitat characterization for Aquatic GAP: Habitat has been characterized using the parameters stream size, habitat quality, water quality, gradient, and riparian forest cover. The first three parameters were combined to form a habitat characterization from which fish diversity was predicted (see Figure 1). The latter three parameters were used for invertebrate diversity predictions (Figure 2). Additional GIS layers of surficial geology, bedrock geology, depth to bedrock, and point-source pollution will be included. The first round of habitat characterization involved static, manually intensive classifications from topographic and Mylar land use overlay maps. In an effort to deviate from such limiting classification, the NY Aquatic Gap Analysis group is developing computer programs to automate classification from digital elevation models, land use, road, and railroad coverages. This will provide substantially more flexibility and enable us to calibrate the model, using previously collected data.

Figure 1. Fish prediction map.

Figure 2. Invertebrate prediction map.

Animal modeling: Spatially referenced coverages for breeding birds and mammals were completed and checked for accuracy during 1998. We continue to cooperate with the ongoing NY Amphibian and Reptile Atlas, sponsored by NYSDEC. We receive updated information on current distributions of amphibians and reptiles annually from the herpetological atlas project, which is scheduled to be completed by late fall 1999. Association matrices, relating vertebrate species occurrences to each of the 45 land cover types we have identified will be completed and assessed for accuracy during the first quarter of 1999. Predicted occurrences of terrestrial vertebrates will be assessed for accuracy using known occurrences of species from recent museum data, herpetological atlas field data, check lists of birds from state parks, breeding bird occurrence data from the NY Breeding Bird Atlas, and check lists of birds and other vertebrates from federal refuges.

Biological databases were developed for fish and invertebrates linking them to specific habitat type. These databases are posted on the web at: http://www.dnr.cornell.edu/hydro2/fishcode.htm and http://www.dnr.cornell.edu/hydro2/invcode.htm.

Land stewardship mapping: During 1998 we completed our land stewardship coverage which includes boundaries for all state wildlife management areas, state forests, state parks, New York City reservoir watersheds with restricted access, Adirondack Park Preserve, Catskill Park Preserve, Department of Defense lands, national parks and historic sites, and federal wildlife refuges. Assignments of management status categories to these areas is under way and will be completed during the first quarter of 1999. In addition, information was obtained and mapped for state-regulated wetlands, water quality management classification, regulated fishing areas, fish-stocked waters, and public fishing rights.

Analysis: Accuracy assessments for land cover and vertebrate distributions will be completed during the first quarter of 1999. Gap analysis for the entire database will be initiated and completed during the first half of 1999.
Field data was collected in the summer of 1998 on fish species diversity, invertebrate species diversity, stream width and depth, substrate, general habitat assessment, water chemistry, and gradient at 39 sites. This information will be used to test fish and invertebrate diversity predictions from the aquatic model and the accuracy of the factors predicting habitat characterization on which the diversity predictions were based. The analysis phase of the Aquatic GAP project is expected to be completed by May 1999.

Reporting and data distribution: A draft final report will be submitted to the National GAP office for review by late May 1999. A final report is expected to be completed by late September 1999. Data distribution is expected to be in printed text form, digitally on CD-ROM, and from web pages. Data distribution by NYSDEC is expected to be accomplished on a regional basis using a central server linked to GIS workstations in each of the nine regional offices of NYSDEC. The recent creation of the Cornell University Geospatial Information Repository (CUGIR, http://cugir.mannlib.cornell.edu), a major National Spatial Data Infrastructure node at Cornell’s Mann Library in the College of Agricultural and Life Sciences, makes additional modes of data display and distribution available.

The NY Aquatic Gap Analysis group maintains a web site for dissemination of up-to-date information on model methodology and results. This site can be found at http://www.dnr.cornell.edu/hydro2/aquagap.htm. In addition, a comprehensive report was compiled in May 1998 for the U.S. Geological Survey discussing the methods, results, products, and conclusions to date.

Other accomplishments and innovations: A collaboration with the Cornell Watershed Modeling Team was developed for the final leg of the application of Gap Analysis to New York waters. A product of the collaboration is an enhanced model used to predict water quality, adapted from one developed by Adamus and Bergman.

Three innovative accomplishments of NY-GAP during 1998 have been 1) the development of new image and geographic data analysis techniques to identify urban lands; 2) the implementation of new procedures for "fuzzy" accuracy assessment of our land cover map for comparison with traditional accuracy assessment methods; and 3) incorporating socioeconomic information into our GAP database to develop a habitat vulnerability assessment for parts of the Hudson River Valley. Each of these accomplishments is described briefly below.

To help identify urban lands, a map of housing density was developed from national census data to refine the cluster labeling process by using an inverse distance weighting algorithm to predict the number of households per unit area. Centroids of census block groups were assigned a household density value based on the number of households and area of the block group. The resulting continuous grid at 30-meter resolution was classified into five categories using equal intervals. Spectral cluster assignments within each category of urban land where evaluated using a threshold cluster value. If the cluster value was above the threshold, the land remained urban. If the cluster value was below the threshold, a land cover type was assigned based on previous experience in nonurban areas.

For our exploration of fuzzy accuracy assessment methods, field crews collected a subsample of accuracy assessment data from the 109 circular 1600-hectare plots during summer and fall 1998. A fuzzy accuracy assessment protocol described by Gopal and Woodcock (Photogrammetric Engineering and Remote Sensing 60:181-188) was implemented by the field crews for approximately 1000 polygons from over 10,000 polygons used in our accuracy assessment process. A numeric value from one to five representing a linguistic scale of map accuracy was assigned to each polygon for each land cover type. Map accuracy is being assessed using both methods at three levels of taxonomic detail. During fall 1998 we concluded map refinements and data analysis for both traditional and fuzzy accuracy assessment activities.

In addition to our research related to urban area delineation and map accuracy assessment, we have continued our participation in the Hudson River Estuary Biodiversity Conservation Initiative under contract with NYSDEC. This project began in 1997 for the purpose of furnishing data and analytical support to facilitate conservation planning in the Hudson River Valley (HRV). It provides funding for more detailed vegetation mapping of the HRV from Albany to New York City. In cooperation with the NY Natural Heritage Program, a gap analysis is under way for the HRV, and a separate interim report will be produced by April 1999, including more detailed land cover maps at a greater resolution than for the statewide vegetation map. Additional funding has been approved for several complementary projects anticipated to begin in April 1999. We will continue to provide GIS support and analytical services for landscape conservation efforts in the HRV, including maintaining the growing biodiversity database for the region, analyzing data as required for the implementation of biodiversity conservation strategies, and providing information to communities in the region to aid in local conservation efforts.

A more extensive habitat vulnerability assessment of the HRV will be performed during 1999/2000 by incorporating socioeconomic information into our GAP database in the form of population growth models that will predict areas of biodiversity most vulnerable to urban and suburban expansion using "build-out" scenarios (Sayeweh 1998). With designation of the Hudson River as a National Heritage River in 1998, our work takes on added significance to planning efforts along the river corridor.

Papers and posters presented at professional meetings:

DeGloria, S.D., M. Laba, S.K. Gregory, J. Fiore, E. Hill, J. Braden, J. Beecher, D. Ogurcak, R. Elliott, A. Stalter, and J. Weber. Accuracy assessment of the New York GAP land cover and ecological communities map. 8th Annual National Gap Analysis Meeting, Santa Barbara, California. 20-24 July 1998.

Fiore, J.J., M. Laba, S. Gregory, S.D. DeGloria, E. Hill, and J.T. Weber. Creating New York state’s land cover map – The Adirondack Park as an example. 8th Annual National Gap Analysis Meeting, Santa Barbara, California. 20-24 July 1998.

Laba, M., S.D. DeGloria, S.K. Gregory, J. Fiore, E. Hill, J. Braden, J. Beecher, D. Ogurcak, R. Elliott, A. Stalter, and J. Weber. Accuracy assessment of the New York GAP land cover and ecological communities map. 14th NYS Geographic Information Systems Conference, Rochester, New York. 23-24 September 1998.

Smith, C.R., J.T. Weber, and M.E. Richmond. Using Gap Analysis information to guide planning for conservation of birds in New York state: A comparison of science-based and expert-opinion approaches. 8th Annual National Gap Analysis Meeting, Santa Barbara, California. 20-24 July 1998.

Publications:

Sayeweh, P.W. 1998. Combining a growth allocation model and GIS results to assess suitable biodiversity management areas: A prototype for Hudson Valley Biodiversity Program in New York state. M.S. thesis. Cornell University, Ithaca, New York.

Smith, C.R., and S.K. Gregory. 1998. Bird habitats in New York state. Pages 29-41 in E. Levine, editor. Bull’s birds of New York state. Cornell University Press, Ithaca, New York.

<%Case "NC"%>

North Carolina

Land cover: In 1998, we developed detailed land cover classifications for the majority of the coastal plain of North Carolina. Using decision rule methodology we have delineated 54 different map units in the southern flatwoods study area (see page 28). This total includes 10 nonvegetated and agricultural classes from the regional land cover. By querying the classified image with our set of interpreted points we have been able to identify which classes contain confusion. Until higher resolution imagery or additional ancillary data sources are added to the analysis, these classes will be combined and attributed appropriately as mixtures. With this information in hand, we have started discussions with the other GAP projects of the southeastern region to identify a consistent regional set of map units that could be used to produce detailed land cover map products across state lines (see page 28). In 1999 we will mosaic the coastal plain scenes together, incorporating comments from reviewers and assessing the region as a whole. We will be using the same approach with a few additional ancillary data sets in the piedmont and mountains.

Animal modeling: At the Southeast Regional GAP Workshop, it was decided that the habitat relationships would be built to alliances. Since then, each of the classification systems used in the primary references have been crossed-walked to the National Vegetation Classification. A database for storage and retrieval of the vertebrate species information has been developed and the core data added. A list of reviewers has been identified for both range and habitat associations. Methods for review by hard copy and via the Internet are being developed. By February 1999 we will have the preliminary range maps posted for review. Throughout the year we will be working on refining the habitat associations based on the recent literature and reviewers’ comments.

Land stewardship mapping: At the request of the North Carolina Center for Geographic Information and Analysis (NCCGIA), we participated in the Framework Data Survey in which the current status and plans for developing framework data were assessed for North Carolina. In follow-up discussions with NCCGIA it became apparent that the development of a new land ownership layer would not be required for North Carolina. In 1999 we will start with the land ownership data layer that has been developed and spend our efforts on assigning the land management status codes.

Analysis: The North Carolina Natural Heritage Program in cooperation with the North Carolina State Museum of Natural Sciences is currently summarizing the results of the analyses done in the study "A Model Biodiversity Analysis for Southeastern North Carolina." In this analysis, species guilds were identified and conservation status of each assessed. The draft vegetation map for the southeast coastal plain was used to identify areas of contiguous suitable habitat for several of the species guilds. Specimen records from the museum and location data from the Heritage database were used to define the range of each of the species in the guilds.

Reporting and data distribution: Distribution of the vegetation classification is currently limited to cooperating agencies willing to provide review on the draft maps.

Other accomplishments and innovations: In April 1998 we hosted the Southeast Regional GAP Workshop. For most of the two days there were two sessions run in parallel-vegetation mapping and vertebrate modeling. During the workshop some important decisions were made to help facilitate consistency across the region. We started the process of developing regional map unit labels from groups of alliances that are consistently mapped together, either because of spatial or spectral proximity. Given the experiences in the region, the need to work on defining compositional groupings and ecological complexes for the region was recognized. This meeting was the start of a flurry of activity around transferring technical expertise in the Southeast.

<%Case "ND"%>

North Dakota

ND-GAP began organizing in 1998. Northern Prairie Wildlife Research Center hosted the first organizational meeting for ND-GAP in November 1998. Representatives from ten federal government agencies, five state government agencies, three universities, and one nonprofit organization attended the meeting.

Land Cover: The first draft of a hierarchical vegetation and land cover classification scheme for ND-GAP was developed. The vegetation and land cover classification scheme will be revised to include natural and seminatural vegetation communities at a scale appropriate to classification using TM imagery and environmental data in a Geographic Information System. Existing vegetation maps from U.S. Fish and Wildlife Service, Environmental Protection Agency, Ducks Unlimited, Forest Service, and the National Agricultural Statistical Service have been acquired and imported into the ND-GAP GIS. These maps do not describe vegetation to the detail that is desired for GAP but are useful for stratification for a more detailed analysis and classification. Vegetation inventory data were identified and plans developed to convert the data into a spatial format compatible with the ND-GAP GIS. These data sets include the ND Natural Heritage Program’s Conservation Data System, ND State School Lands Range Inventory Database, and data collected by biologists from Northern Prairie Wildlife Research Center, U.S. Forest Service, and Natural Resources Conservation Service, and graduate students at ND universities. Three dates of TM imagery (spring, summer, and fall) and environmental data including 3-arc second digital elevation models, climate surfaces, STATSGO soils data, a surface geology map, and National Wetlands Inventory data will be used to map natural and seminatural vegetation at a level between the formation and alliance levels of the FGDC Vegetation Classification Standard.

Animal modeling: A search and review of species check lists for ND identified 422 species of terrestrial vertebrates for probable inclusion in ND-GAP. A breakdown of this list includes 223 species of breeding birds, 95 species of avian migrants, 78 mammals, 11 amphibians, and 15 reptiles. Location records for most species are recorded by township, range, and section of the Public Land Survey System (PLSS). Because this system records the location of species observations at much greater spatial detail than the 650-km2 hexagons used in GAP projects, we will develop the species observation database using the PLSS. We will begin building the database with breeding birds, as observation records for birds are more abundant and better organized than those for mammals, reptiles, and amphibians. Tom Skelbar joined the ND-GAP staff in October 1998 and will take the lead for the terrestrial vertebrate distribution mapping.

Land stewardship mapping: Organizations with stewardship responsibilities for public lands in North Dakota have been contacted and queried about the status of maps, digital boundary and management unit data, and realty records for their lands. Few organizations had digital boundary and management status data for all their lands. Boundary and management unit data was requested from those organizations which had digital data for all their lands. The scale of the source data is variable among the agencies ranging from larger than 1:24,000 to 1:100,000. The ND Department of Transportation (NDDOT) is converting their databases and county mapping process to a GIS. They have created a state layer to complement county databases. The state layer contains the boundaries for all public lands in the state, but the parcels are not attributed. NDDOT anticipates completing the vector in January 1999 and will make it available to ND-GAP. In 1999 we will continue to acquire existing digital data from the appropriate agencies and the state vector prepared by the NDDOT.

<%Case "OH"%>

Ohio

Land cover: Mapping will begin in January 1999. The land cover mapping effort for Ohio will be led by Dr. Joan Nichols of the Ohio State University’s School of Natural Resources. Goals for the next 12 months include team-building and training of all individuals involved in mapping. Other efforts within the state of Ohio have produced an Anderson level 1 land use map-these data will be examined for their potential use in the GAP program. It is anticipated that at least a portion of the land cover map will be complete by the end of 1999. An additional component of this work will be to evaluate different means of mapping riparian areas using remote sensing or other techniques.

Animal modeling: Compilation of animal data sets will begin in January 1999. Preliminary efforts have focused on examination of methodology of other states. Goals for the next 12 months include compilation of data regarding habitat affinities and documented occurrence of species within Ohio. Additionally, the Ohio GAP effort includes work on selected aquatic species. Several aquatic data sets have already been identified, and we are working towards the exchange and reformatting of these data for use in aquatic GAP.

Land stewardship mapping: Land stewardship mapping is to begin in January 1999. Plans for the next 12 months include examination of adjoining states’ methodologies and determination of data sources.

Analysis: Analysis will only begin as we compile the animal data. Efforts are focused on examining methods of other states.

Reporting and data distribution: A fact sheet for OH-GAP is planned for late 1999. We hope to use this fact sheet to develop a cooperator base and for general project information.

Other accomplishments and innovations: OH-GAP is unique in that it represents the first integrated terrestrial and aquatic analysis in the United States. As such, it will bring together researchers from the Water Resources Division and Biological Resources Division of the USGS, the Ohio State University, and several other state and local government agencies. We anticipate that this effort will encourage research to be focused on large-scale ecosystems whose terrestrial and aquatic components are related and codependent.

<%Case "OK"%>

Oklahoma

Land cover: A draft land cover map has been completed. We are currently gathering ground-truth data for conducting an accuracy assessment of the draft map. Our goal is to complete the accuracy assessment by late winter 1999.

Animal modeling: Range maps and habitat association models for 427 vertebrates have been developed. A pilot study of the vertebrate modeling process has been completed. Modeling with the draft land cover map is ongoing. Our goal is to complete all animal modeling by late spring 1999.

Land stewardship mapping: The land stewardship map is complete, and final review of stewardship status is ongoing. Our goal is to finalize the map by late winter 1999.

Analysis: Our goal is to conduct the final analysis of GAP data layers by late spring 1999.

Reporting and data distribution: The final report and data distribution will be completed by June 1999.

Other accomplishments and innovations: The Gap Analysis methodology was used to develop a GIS-based inventory of the habitat and biological resources of Tishomingo National Wildlife Refuge in Oklahoma.

<%Case "OR"%>

Oregon

Land cover: An original land cover map was completed for Oregon in 1992-the second GAP prototype after the original Idaho map. A second-generation land cover map was completed in August 1998 and is currently available. Metadata is available for this second-generation cover, but the accuracy assessment is incomplete, and there is no vegetation manual associated with the new map. Over the next seven months, we plan on completing the chapter on land cover mapping for the Oregon Final Report and producing a vegetation manual for the updated map as well. See Figure 1 for the second-generation map.

Figure 1.

Animal modeling: Animal modeling was completed for Oregon using the first generation GAP vegetation map. The result of this project is the book Atlas of Oregon Wildlife (Csuti et al. 1997). The initial models relied on hexagon distributions prepared for all native wildlife species. Since these were completed over eight years ago, we have recently (October 1998) updated the hexagon distribution covers for all wildlife species. Our goal is to update the model distributions for all vertebrate species using the updated hexagon distributions and the second-generation land cover, with all of the draft models to be completed by December 1998. We have updated the wildlife habitat relationship matrix to reflect the differences between the first- and second-generation land coverages and have created ARC/INFO programs (AMLs) to generate the coverages. While the existing coverages have all been peer-reviewed, it is hoped the updated modeled distributions can also receive some peer reviews, although given the limited time frame these will probably be limited to local experts.

Land stewardship mapping: The land stewardship cover is complete after extensive work over the last six months. A graphic showing the cover, with GAP status 1 and 2 lands in red, is shown in Figure 2. We intend to update the cover with newly designated or newly acquired public lands until sometime in December 1998, when all of the species distributions have been completed. The cover will then be archived and used to generate the tables and maps in the analysis. After that, the cover will be continually updated and maintained on the OR-GAP web page.

Figure 2.

Analysis: Analysis will begin as soon as the species distribution covers have been generated from the updated habitat models, the updated vegetation cover, and the updated hexagon distribution database. Analysis to complete the basic gap analysis chapters will be complete by February 1999. All analyses will be completed on a statewide and an ecoregional basis. Additional analysis may include an analysis of habitat losses based on a presettlement vegetation coverage for Oregon and a comparison of different methods of generating land coverages.

Reporting and data distribution: A final state report will be prepared during the next six months. The final report and the basic coverages will be posted at the OR-GAP home page. The basic coverages that will be posted include stewardship, species distribution, first- and second-generation GAP vegetation coverages, and a presettlement vegetation coverage. All of the final GAP products and coverages will also be available on the standard GAP CD-ROM.

During the next 10 months of the project, using GAP funds as well as funds from the U.S. Fish and Wildlife Service, we also will distribute GAP data to watershed councils, local governments, and other decision makers in Oregon. We intend to provide GAP data, Oregon Natural Heritage Program threatened and endangered species data, and analyses to assist in watershed, basin, or ecoregional planning efforts. We are working with the Northwest Office of the Defenders of Wildlife to distribute GAP data and also to assure that updated GAP coverages are included on their CD-ROM product, Oregon’s Living Landscape, an Interactive Introduction to Oregon’s Biodiversity. This CD-ROM was produced as part of the Oregon Biodiversity Project in which the Oregon Natural Heritage Program and OR-GAP were partners.

Other accomplishments and innovations: As part of our association with the Oregon Biodiversity Project, an application was developed which allows both GAP and Natural Heritage data to be summarized on a watershed basis. This application was developed by the Defenders of Wildlife staff and is included on their CD-ROM. We are working on developing a similar web-based application.

Literature Cited:

Csuti, B., A.J. Kimerling, T.A. O’Neil, M.M. Shaughnessy, E.P. Gaines, and M.M.P. Huso. 1997. Atlas of Oregon wildlife: Distribution, habitat, and natural history. Oregon State University Press, Corvallis, Oregon. 492 pp.

<%Case "PA"%>

Pennsylvania

Land cover: A multi-scale land cover product has been completed. The forest landscape matrix was mapped at a 100-ha resolution by on-screen photointerpretive delineation from TM. High-intensity and low-intensity development was mapped to the perceptual limit by photointerpretive delineation from TM display with PennDOT road coverage superimposed. General land cover was mapped at 30-meter TM pixel level using eight classes by semisupervised labeling of 255 hyperclusters per scene, with a different approach to spectral clustering than the Spectrum method of EROS Data Center. The clustered image product is directly usable in image mode with ArcView and/or ArcExplorer, and the land cover rendition just requires a different color or legend file. Checking the confirmatory cluster assignments remains to be done. On-line DOQs will be used as reference material. Pixels are being aggregated to a 2-ha level with some combination of classes for vectorization. A mix of leaf-on and leaf-off video will also be used for accuracy assessment. Some preliminary superalliance classification may be conducted but will not be used in habitat models or formally validated in these first-generation products.

Animal modeling: Avian models are completed and ready for review. Amphibian and herp models are in progress. Models use generalized land cover and terrain feature criteria. Modeling of fishes is under consideration and will be of a preliminary nature.

Land stewardship mapping: A first version is complete and was presented at a November 1998 conference in Harrisburg, PA. It was well received, but some suggestions for additions were forthcoming.

Analysis: Analysis will be conducted in terms of species diversity by groupings of habitat affiliates (generalized guilds) to locate major habitat complexes. Echelon analysis is a complementary methodology (see PA-GAP web page at http://www.erri.psu.edu/web/projects/gappage.htm).

Reporting and data distribution: PA-GAP geographic data will be distributed through Pennsylvania Spatial Data Access (PASDA) as introduced in the PA-GAP web page. There will, of course, be a conventional GAP report as well. Project completion was slated for December 1998 but rescheduled to September 1999 because of W. Myers’ fall sabbatical to lead the Pennsylvania ECOMAP effort at landscape levels. GAP in Pennsylvania is seen as a progression rather than a singular end in itself.

Other accomplishments and innovations: PHASE analysis of remotely sensed data and echelon analysis of (virtual) surface information are outgrowths of Gap Analysis work through a companion project sponsored by an NSF/EPA watershed research partnership. Please consult the PA-GAP web page for an introduction and links to software.

<%Case "SC"%>

South Carolina

Land cover: The South Carolina Gap Analysis Project (SC-GAP) has completed an initial land cover map with 28 map units comprised of general vegetation, urban, agriculture, and water classes. The purpose of this map is two-fold: to provide a background for incorporation of ancillary data and modeling of vegetation classes at the alliance level, and to provide land cover classes to be used in the animal habitat modeling. During the next year, SC-GAP will be incorporating ancillary data, developing decision trees, and conducting a pilot project for alliance-level mapping along the coast with the U.S. Fish and Wildlife Service and the South Carolina Department of Natural Resources. We hope to have much of the coastal plain mapped in the next twelve months. As part of a project to develop an ant layer in cooperation with Clemson University, South Carolina Cooperative Extension Service, and the South Carolina Cooperative Research Unit, a technician and graduate student will be collecting information for accuracy assessment of the completed general vegetation map in a stratified random sample of public lands.

Animal modeling: SC-GAP has compiled a list of 488 vertebrate species for inclusion in the animal layer. Data on distribution by county have been gathered and compiled from museum records and state and regional accounts. We have completed the database for all vertebrate species (this includes information on distribution and habitat affinities) and created maps of species distribution by county. These are now out for review to experts within the state. We have also formed a cooperative arrangement with Clemson University, the Cooperative Extension Service, and the Cooperative Research Unit at Clemson to collect ants throughout the state, describe their habitat affinities and develop a distribution layer of ant species in South Carolina (see Gap Analysis for Ant Species for more on the SC-GAP ant project). Other invertebrates to be included in the animal layer include butterflies and tiger beetles.

Land stewardship mapping: The database of public lands is approximately 80% complete. We will continue compiling data on those lands not currently inventoried and will finish digitizing those maps not currently completed. Additionally, we will be refining our classification scheme for protected lands within the state.

Analysis: The analysis portion of SC-GAP is just beginning. Development of ancillary data to be used in the modeling of vegetation is currently under way. We are developing decision trees that will allow us to use ancillary data such as soil moisture content, elevation, and physiographic province in refining our map units from 28 general classes to the TNC alliance level of vegetation classification. Upon return of the animal distribution maps and habitat matrices from our experts, animal modeling will also commence.

Reporting and data distribution: We have reported the status of SC-GAP at the National Gap Analysis Meeting in Santa Barbara in July 1998. There was also a meeting of state cooperators in September to discuss progress and future directions. No data or maps have been distributed yet.

<%Case "SD"%>

South Dakota

Land cover: We are completing general land cover maps for eastern South Dakota and the Black Hills region. Six of the eight satellite scenes east of the Missouri River have been processed using an Imagine model to isolate spectral signatures of perennial vegetation from agricultural and wetland habitats (previously classified). Human population data (CIESIN) for all of South Dakota has been obtained and processed to identify developed areas. Two satellite scenes with just the perennial vegetation have been run through a clustering algorithm in Land Analysis System (LAS) software to allow more detailed discrimination between deciduous and coniferous trees and grasslands. Satellite images covering the Black Hills region have been classified; however, many of the vegetation classes have the same spectral characteristics. Currently we are incorporating ancillary data into the Black Hills vegetation classification process and will be running and interpreting regression tree analyses. Goals for the next twelve months include processing satellite images covering the remainder of the western half of the state and creating vegetation models to distinguish vegetation alliances throughout the state to complete the land cover map.

Animal modeling: Distributions for South Dakota’s 88 mammalian species were constructed based on EPA’s hexagonal grid using museum records, agency records, and trapping data obtained during the summer of 1998. Completed distributions are being reviewed by a panel of experts throughout the state. Wildlife habitat relationships are being researched for construction of models to predict mammalian distributions. Habitat relation models of two species of carnivores occurring in the Black Hills are currently being constructed. Additional mammalian models will be formulated within the next year. A literature search has begun to collect habitat information for 350 avian, 37 reptilian, and 15 amphibian species.

Land stewardship mapping: A draft of the SD-GAP stewardship layer is published on our web site. We are in the process of digitizing TNC preserves, 27 state parks and recreation areas, state wildlife refuges, and Indian reservations. U.S. Army Corps of Engineers lands are in digital format but need to be combined with the web-published stewardship map. To date, approval to use the Indian reservation digital coverages has not been received from tribal counsels. Our stewardship map will be completed this year.

Analyses: Accuracy assessment of eastern South Dakota and Black Hills general land cover maps will be completed this year.

<%Case "TN"%>

Tennessee

Land cover: The detailed vegetation map is completed. The vegetation map was produced using classification techniques applied to Landsat TM imagery and aerial videography. Accuracy assessment was performed using a subset of points set aside from the aerial videography interpretation. Goals to accomplish in spring 1999 include finalizing the results of the accuracy assessment, implementing the aggregation algorithm, and tiling the vegetation map.

Animal modeling: Predicted species distributions and species richness data have been produced for Tennessee’s 364 terrestrial vertebrates. Expert review and accuracy assessment of the species distributions need to be completed. Goals include formatting the animal distribution data into National GAP’s standards for CD-ROM production.

Land stewardship mapping: The land stewardship layer is completed. Lands mapped are current through December 1997. The public lands coverage has been updated, while land management status needs to be assigned prior to any further analysis.

Analysis: A preliminary predicted species distribution analysis has been completed. Species richness for each animal group was intersected with Tennessee’s land stewardship coverage. Species diversity by land status was based on the species richness totals. Individual species have not been separated by land status categories. Minimum and maximum species numbers per land status category were calculated for each animal group. TN-GAP will divide each species by land status type and reevaluate the predicted species and land stewardship analysis in the next 12 months.

Reporting and data distribution: The first draft of the final report will be completed December 1998. Goals to accomplish by June 1999 include refining the data to meet National GAP standards. Plans are to present TN-GAP data as part of the TWRA web page.

Other accomplishments and innovations: The Tennessee Biodiversity Program (established by the Tennessee Conservation League) and TWRA’s GIS division have been working together to provide planners and community leaders, landowners, natural resource professionals, and educators with information on Tennessee’s natural resources. TWRA provides TN-GAP data and related GIS data layers as ArcView files to county planners and community leaders.

<%Case "TX"%>

Texas

Land cover: The land cover map for West Texas has been completed in draft form, and scenes have been stitched together. Field work has been completed for 229 of the 254 counties in Texas. Classification is under way and progressing at the rate of about four scenes per month. We expect to complete the field work in the remaining 25 counties before January 1999 and to classify the remaining scenes. However, massive flooding in the San Antonio and Houston areas in October 1998 has made this area inaccessible.

Vertebrate modeling: We have identified 637 terrestrial vertebrate species as being native to and breeding in Texas. GIS layers representing the range extents for each of these species have been developed from existing range maps. In addition, a database consisting of 34,441 location records has been developed. Habitat profiles have also been prepared for 487 of the 637 species being modeled, and state-wide GIS layers have been created from the following profile variables: precipitation, temperature, soils, hydrology, ecoregions, and elevation.

Analysis: Data prepared for West Texas is being analyzed for accuracy and to prepare selected species-specific maps (e.g., prairie dog towns and scaled quail distribution).

Reporting and data distribution: Draft maps were provided to 89 landowners in West Texas to solicit their evaluations for use in accuracy assessment. Draft maps have also been prepared for Texas State Parks, the National Park Service, the U.S. Border Patrol, USDA, and cooperators in Texas and Mexico.

Other accomplishments and innovations: We have established a bioinformatics program at Texas Tech University that involves the museum and draws from multidisciplinary fields and programs such as the Texas GAP Project, the Natural Science Research Laboratory, and the Department of Biology. One goal of this program is to build a relational database, as a linkage of distributed databases, which can be accessed through the Internet. We are using Oracle as the primary relational database with specific existing applications running in FoxPro, MS Access, etc. Examples of the data now available include vertebrate collections from the museum; field notes, photos, and records of the 1895-1906 biological survey of Texas; routes of the Breeding Bird Survey of Texas conducted annually by the U.S. Fish and Wildlife Service; daily weather data from 3860 sites in Texas with some records extending back over 100 years; current landscape photographs of habitats in Texas; soil maps for Texas; vegetation in Texas; aerial videography; digital elevation maps; and Landsat TM scenes for all of Texas and the North American Trade Zone established by the North American Free Trade Agreement along the border with Mexico.

Several products have been developed and placed on the Internet. Applications include soil maps by agricultural crop type, design and placement of constructed wetlands for extraction of nutrients from the effluent of cattle feedlots, integration of aquaculture with traditional agriculture, distribution models for vertebrate species in Texas, and identification of areas with high levels of biodiversity. These and other projects will allow researchers, resource managers, landowners, school children, and the general public to use the best available data in their projects, research, and decisions.

<%Case "UT"%>

Utah

Update has been initiated as a regional project (see Southwest Regional Breakout Session on p. 64 of GAP Bulletin No. 6)

<%Case "VA"%>

Virginia

Land cover: The primary land cover map for Virginia was completed in spring 1998 (Morton 1998). This map is currently available through anonymous ftp from the Fish and Wildlife Information Exchange (FWIE) server at ftp.fwie.fw.vt.edu:80. Presently, VA-GAP is completing a second iteration to a super-type classification of the entire Commonwealth. The Coastal Plain and Piedmont physiographic provinces are expected to be completed by the end of 1998. The Ridge-and-Valley and Mountain regions of the state are expected to be completed by spring 1999. Accuracy assessment for the detailed land cover map will be conducted throughout the winter and spring of 1999. A substantial database comprised of ground locations, aerial videography, and maplets from cooperators will be used to refine and assess the level II land cover map. A phase II sampling effort will be undertaken for cover types with low representation in the existing database and for types which require more data to obtain acceptable accuracy. For more information on VA-GAP’s land cover mapping efforts, please contact Scott Klopfer (sklopfer@vt.edu).

Animal modeling: The vertebrate model database has been researched and updated by the Virginia Department of Game and Inland Fisheries. This database (VAFWIS) will provide the basic information needed to estimate species distributions in Virginia. Animal species distributions will be examined using data from ongoing cooperative projects at Ft. A.P. Hill and Ft. Pickett military installations and the University of Virginia’s Mountain Lake Biological Research Station. Testing on small, selected areas of the state has already begun. The first species distribution models will be produced for the Coastal Plain and Piedmont regions in spring 1999.
Land stewardship mapping: VA-GAP has collected property boundaries from federal and state government agencies managing land in Virginia. Other stewardship properties are being sought and obtained. These data will be combined to create the land stewardship information layer by the end of 1998. This simple map will be improved by compiling information on the management status of these properties and their potential for future management for biodiversity. For more information regarding these efforts please contact Scott Klopfer (sklopfer@vt.edu).

Analysis: VA-GAP is planning on beginning the analysis phase in early spring 1999. The products of this effort will include both standard GAP products and a series of maps, data, and statistics requested by our many state cooperators.

Reporting of data and distribution: Many of the basic data sets used by VA-GAP are available to the public via ftp. Simple land cover as well as roads, rivers, and wetland vector layers are currently served. VA-GAP is pursuing other, more user-friendly distribution options. We hope to develop a web-based site where those interested in GAP products may choose a specific area, data set, or format to download information. This site would employ scaleable maps, metadata, and other information to make GAP data more useful to a wider audience.

Other accomplishments and innovations: VA-GAP has been successful in forging strong relationships with both federal and state government agencies and nongovernmental organizations in Virginia. We continue to develop and improve a hybrid classification technique incorporating remote sensing and ecological models which we hope will improve land cover classification for areas in complex terrain. In addition, we have established ourselves as a front-runner in landscape-scale data analysis and research in Virginia.

<%Case "WA"%>

Washington

CD products are in progress. Data can be accessed on the Internet at http://salmo.cqs.washington.edu/~wagap/.

<%Case "WV"%>

West Virginia

Land cover: Preliminary land cover and alliance/ecological complex mapping are complete. We are refining the classifications with specialized ancillary data (e.g., a methodology for improved mapping of cove hardwoods) and additional videography classification.

Animal modeling: Models have been developed and will be implemented once final land cover tiles are available. We have developed specialized techniques for herpetile modeling that we hope to publish in 1999.

Land stewardship: Completed.

Analysis: Most analyses have not yet been undertaken. We have developed land cover data sets for the years 1925, 1932, 1950, and 1976 and are involved in change detection experiments with the land cover data from those periods and the GAP data sets.

Reporting and data distribution: We have distributed preliminary products to over a dozen federal and state agencies. We have prepared an animal modeling background document (that details the habitat abstracts and models) and have been circulating that for over six months.

Other accomplishments: GAP is being institutionalized in a West Virginia Agricultural and Forest Experiment Station project titled "The West Virginia Land Status and Trends Project." In this project we will be obtaining new imagery every three years and completing forest cover, open land, and land use/land cover status and trends assessments on a three-year cycle. GAP data will be permanently housed in the WVDNR nongame program and in the GIS research program of the West Virginia Cooperative Fish and Wildlife Research Unit.

<%Case "WY"%>

Wyoming

Other accomplishments and innovations: Though WY-GAP was officially completed in January 1997 with the publication of the final report, there is still plenty of GAP-related activity going on in the state of Wyoming.

Wyoming Bioinformation Node: A partnership with National GAP and the University of Wyoming’s Spatial Data and Visualization Center has resulted in the development and continued expansion of the Wyoming Bioinformation Node, providing Wyoming with data and services as part of the National Biological Information Infrastructure (NBII). Please see our web site, www.sdvc.uwyo.edu/wbn, for information about the following projects:

• Wyoming Species Atlas: Provides on-line access to information and GAP data for Wyoming’s vertebrate species.
• Biodiversity Expert Systems Tool: incorporates GAP data and expert rules within an ArcView interface to assist county planners in reviewing development propositions.
• NBII data and metadata for many biological data sets available from various Wyoming sources.

Refuge GAP: This is an ArcView Spatial Analyst-based decision support system, utilizing GAP data, to assist the U.S. Fish and Wildlife Service in project reviews and land acquisition prioritization in Wyoming.

Wyoming Internet Map Server: The WY-GAP data has been incorporated into this interactive map and query system, which provides access to hundreds of Wyoming data sets and simplified GIS functions over the web. Web site: wims.sdvc.uwyo.edu

Selecting Plant Species for Gap Analysis in Wyoming: GAP projects have made an important contribution towards modeling distributions of terrestrial vertebrates but rarely have addressed other organisms. Vascular plants in particular have been treated primarily as components of vegetation rather than as individual species. Like vertebrates, plant species also show "gaps" between their known ranges and protected areas. This project is addressing the unique issues involved in selecting and modeling plant species for gap analysis.

<%Case Else %>

Go to the State Projects section and choose a state first.

<%End Select %>