This file was created by scanning the printed publication. Errors identified by the software have been corrected; however, some errors may remain. USGS-NPS Vegetation Mapping Program 1 Thomas Owens 2 Abstract-The USGS Center for Biological Informatics is cooperatingwith the National Park Service to classify and map vegetation communities on 235 Park units. The USGS-NPS Vegetation Mapping Program's approach is to develop scientifically valid basic vegetation classification and spatial vegetation which can be used as building blocks for management planning, inventory, monitoring and research by local, regional, and national managers and scientists. The Program is noteworthy for its approach to vegetation mapping which includes developing: 1) a national vegetation classification system, 2) standardized inventory and mapping protocols, 3) accuracy assessment, and 4) scientific peer review. The Program uses extensive field work and aerial photography to perform classification and mapping work. Products include digital databases, comprised of classifications and descriptions of vegetation communities, spatial vegetation data, and metadata, which are important for detailed Park management as well as regional and national assessments of biodiversity. The Program has attracted the interest of other federal and state agencies, including the Fish and Wildlife Service, the Bureau of Reclamation, the Forest Service, and the EPA. These agencies are interested in incorporating the Program's protocols and standards into their own vegetation inventory and monitoring projects. For more information visit the website at http://www.biology.usgs.gov/npsveg. Basic data are fundamental to the management ofnatural resources. Natural resource inventories allow managers and scientists to account for resources under their management and study, including the presence, class, distribution, and normal variation of plants and animals, and such important abiotic components as water, soils, landforms, and climate. It is necessary to compile the basic building blocks of information on these resources in order for the information to be combined in different ways to answer different questions. The USGS-NPS (U.S. Geological Survey, N ational Park Service) Vegetation Mapping Program is a vegetation inventory program that provides basic vegetation community information to National Park managers and scientists. The USGS Biological Resources Division's Center for Biological Informatics administers the Program and provides scientific and technical oversight for the National Park Service. Program Goals The Program's goals are to create highly accurate, detailed vegetation spatial databases in an efficient manner lpaper presented at the North American Science Symposium: Toward a Unified Framework for Inventorying and Monitoring Forest Ecosystem Resources, Guadalajara, Mexico, November 1-6,1998. Ztrhomas Owens is Coordinator, USGS-NPS Vegetation Mapping Program, at the USGS Center for Biological Informatics, Building 810, Denver Federal Center MS 302, P.O. Box 25406, Denver, CO 80225-0046 U.S.A Phone: (303) 202-4259; Fax: (303) 202-4219; e-mail: tom_owenS@Usgs.gov USDA Forest Service Proceedings RMRS-P-12. 1999 for the park units and vicinities, and expand the National Vegetation Classification Standard (NVCS) to include all detailed classes at these units. The data must be spatial1.1 and thematically accurate, meet the professional standards of the scientific community and applicable standards of the Federal Geographic Data Committee FGDC), have a nationally consistent, hierarchical, useful, classification scheme, have a level of detail (spatial resolution and classification level) useful to park management. The Program serves local management needs of each park while providing comprehensive vegetation information at national and regionallevels. The Program is founded on good science, which means all data are based on stati3tically sound field work and proven, up-to-date technology is used. Program Standards and Protocols The maj or standards for the Program are 1) the vegeta tion standard, 2) the accuracy standard, 3) the metadata standard, and 4) the spatial data transfer standard. The classification system must be based on sound science, be repeatable across the landscape, be based on standard field and data analysis methods, be broadly accepted, be ecologically meaningful and classify existing biological associations, and be mappable from imagery. In addition, it must be hierarchically organized, appropriately scaled, flexible and open ended, well documented, and can be cross-walked with vther frequently used systems. The system that meets all these requirements is the NVCS. It is organized in the following manner: classificat~on Physiognomy Division/Order-dominant life form (example: Tree Dominant) Class-spacing & height of dominant form (example: Woodland) Subclass-morphological & phenological similarity (example: Evergreen Woodland) Group-climate, latitude, growth form, leafform (example: Temperate Evergreen Needle-leaved) Formation-mappable units (example: Evergreen Needleleaved Woodland with Rounded Crowns) Floristics Alliance (Cover Type)-dominant species (example: Douglas-fir Woodland) Community (Association)-subdominant or associated species (example: Douglas-fir / Snowberry Woodland) Uniformity in classification methodology over the entire park system is critical to achieve the overall inventory and 199 ~. : monitoring goals of this initiative. All parks must be mapped at the same level of classification detail (i.e. plant association/cover type) and must be able to be re-aggregated to the FGDC vegetation classification scheme. The accuracy standard requires that the minimum mapping unit is 0.5 hectares, that locational accuracy meets the National Map Accuracy Standards at 1:12,000 (a welldefined object is located within 10 meters of its actual location on the ground), and classification accuracy will greater than 80% for each vegetation class. To ensure that the classification accuracy requirement is met an accuracy assessment is performed at all park projects. Stratified random sampling with replacement is used for.the accuracy assessment. Ground visits are used to independently verify the mapping classes; the field crews have no knowledge of the mapped classes to ensure no bias exists in the accuracy assessment. The metadata standard requires that all datasets are documented with metadata meeting the FGDC metadata standard. The meta data compilation tool MetaMaker is used to achieve this standard. All spatial data produced must be in the Spatial Data Transfer Standard (SDTS) format. Basic protocols were generated early on the program. There are four documents which outline the protocols: 1) Developingand documenting a National Vegetation Classification Standard, 2) Establishing standards for field methods and mapping procedures, 3) Producing rigorous and consistent accuracy assessment procedures, and 4) Establishing standards for using existing vegetation data. The protocols provide general guidance to each park project. The protocols are being tested in prototype studies, protocols will vary based on a matrix of variables related to size of area to be mapped (park & environs), existing data and knowledge, complexity of vegetation and access/logistics. All field samples, accuracy assessment samples and potential mapping approaches (i.e. softcopy) will be GPSbased. The GPS based approach is important because all field data becomes spatial data. and relatable to other layers in a spatial analysis. The development of protocols is near its final stage, although there will always be new challenges and issues as the program maps new units in different parts of the country. Specifically, the protocols for classification field methods include data and information discovery meetings at the start of each project and the "gradsect" (short for gradient tran~ approach. The gradsect approach stratifies the park into bio-physical units using geological, soils, topographic, wetland, disturbance and other data to find ''hot spots" of environmental diversity where sampling will be the most efficient. This enables sampling the greatest amount of vegetational diversity within the smallest area. It is an efficient way to gather data for testing classification, describing classes and developing photointerpretation signatures/keys. The specific protocols for mapping are that photointerpretation will be used. New aerial photography of the appropriate type(s), scale(s), and season(s) will be acquired for all areas (if useful, recent aerial photography does not exist). Small parks will be mapped at scales larger than 1:24,000. Softcopy approaches to mapping will be used, if technically feasible and cost effective. 200 Program Methods The major steps that are taken to complete each park are 1) data review, 2) data acquisition, 3) gradsect, 4) initial field reconnaissance, 5) field sampling, 6) classification characterization, 7) photointerpretation, mapping and automation, 8) accuracy assessment, and 9) final product review. There are many other intermediate steps required, and these steps are often iterative, but this is the basic process flow for each park project. The Program contracts with experts in vegetation mapping and vegetation ecology around the country to provide the best expertise for each project. The Nature Conservancy is a major national partner, and the individual state Natural Heritage Program ecologists are heavily involved in specific park projects. National Parks tend to be centers of scientific research and monitoring and have long histories of data collection. Data review and analysis can be a major task because of the wealth of available data and information. It is important to carefully analyze existing datasets because they may provide efficiencies in the project. Datasets that usually exist and may be useful are aerial photography, field plots, historical vegetation maps and spatial data, reports and monographs. The next step is to acquire baseline data. This includes taking aerial photography, acquiring ortho images (usually USGS digital ortho quarter quads), acquiring other spatial datasets such as elevation, soils, geology, and disturbance history to be used in the gradsect. The gradsect is run (perhaps several times with different combinations ofvariabIes) to model the major environment drivers in a park. The resulting spatial database is used to help allocate sample points for field sampling, and to inform the photointerpretation process. The mappers and ecologists who perform the classification participate in the initial field reconnaissance. The purpose of this trip is to familiarize the mappers with the photo signatures of the major vegetation classes, provide more information to the ecologists on the local environment, and to allow a discussion between the mappers and ecologists on how the park will be classified and mapped. Once the initial reconnaissance is complete, the mappers will initiate photointerpretation and provide this information to the ecologists to help them plan their field sampling strategy. The field sampling may take one or more field seasons, depending on the size of the park. Three to five plots are taken in each vegetation association to measure the variability for each class. If the association is well defined in an area, only one plot may be taken. However, it is more common to develop many new associations for an area based on the work in a park. The field sampling strategy will evolve as plots are taken and photointerpretation proceeds, beca use new associations may be discovered. There are at least two versions of the classification system. A preliminary version is developed early in the project by analysis of regional TNC databases and advice from local experts. This preliminary version is used to start the photointerpretation process, and is modified during the course of the project. The field data are automated into a database and analyzed quantitatively using computer programs or qualitatively using methods described in the NVCS document to develop the final classification. The USDA Forest Service Proceedings RMRS-P-12. 1999 photointerpretation and classification process are heavily interdependent and new information gleaned from one process may modify the other process. Once a preliminary classification system is in place photointerpretation may begin. It is done under a stereoscope to allow topographic information to assist the process of identifying vegetation associations. The photointerpreters may take several field trips to the park for ground truth data which mayor may not coincide with the field work for the classification. When the photointerpretation is completed, the data are rectified and automated. This may be a heads-up digitizing process using the ortho image as a background, or it may be a manual transfer to a hard copy ortho image with subsequent scan digitizing. Whichever process used, the products must meet the Program accuracy requirements for classification and location accuracy. When the spatial database of the vegetation communities is complete a stratified random sample strategy for accuracy assessment is designed. Each vegetation association, that is extensively represented in the park, has at least 30 plots located in it. Fewer plots may be taken if the association is rare. The plots are navigated to with the aid of GPS receivers. The field crews may take a plot of the orthoimage with polygons delineated on them for locational purposes, but do not know how the polygons are classified. The field crews take extensive information in each sample polygon. The point data from the accuracy assessment are then compared to the mapped data to ensure that the spatial data meets the classification accuracy requirement of 80%. The final data products, including the spatial data, the classification data, the accuracy assessment data, and the reports and metadata are reviewed by Program and park staffto ensure that all standards are met. Then the data are delivered to the park and placed on the website. Peer Review -----------------------------------------------------------Peer review is an import.ant component of the program. Reviews by independent scientists and managers ensure that the program is effective at meeting the users' needs, while maintaining strong standards. A review of the program occurred in September 1996 by USGS and Park Service scientists and managers. A second review has been recently completed by the National Park Service and another review ofthe program is currently being conducted the USGS. The reviews findings and recommendations have placed on the Program's website at http://biology.usgs.gov/ npsveg. Recommendations made by the 1996 review have been implemented. Recommendations by the Park Service review and the USGS will be implemented as appropriate. Program Status _ _ _ _ _ _ __ The Program has developed and published protocol documents that establish the basic methods (discussed above). The Program has initiated mapping and classification projects in twenty parks and has completed five of these USDA Forest Service Proceedings RMRS-P-12. 1999 (Tuzigoot National Monument, Devils Tower National Monument, Mount Rushmore National Monument, Jewel Cave National Monument, and Scotts Bluff National Monument). The program plans to initiate projects in five new parks this fiscal year. Applications _ _ _ _ _ _ _ __ The Program's data and information can be applied in numerous ways. Some examples include 1) invasive exotics inventory, such as leafy spurge at Theodore Roosevelt National Park, by combining the spatial data with the classification descriptions, 2) Threatened and Endangered plant species inventory by applying the spatial data with the community description data to model the possible distribution of these species, 3) Using the vegetation associations' distribution, along with the structure information the Program collects, and topographic data to model fire behavior, 4) Scotts Bluff National Monument used the species listings in the community description data to develop seed mixes to restore natural prairies in disturbed areas, 5) use the spatial and description data to model wildlife habitat, for example grizzly bears in Glacier National Park, and 6) Use the vegetation classification data and descriptions for regional, national, and global assessments of endangered vegetation communities. Conclusions --------------------------------------------------------The program has been in existence since 1994. The lessons learned from the experiences to date are: 1) providing all products is critical-not just mapping products. The classification data and information at as important as the mapping products to resource managers and scientists. The results from the Park Service review indicate that parks where work has been completed are using all the data for resource and management applications. The parks do not wan t a cut-back in the products produced for each park or the resolution of the products, because they feel that each product is valuable, the resolution is adequate for sitespecific management purposes, and the process is costeffective, 2) these products are the basic building blocks of vegetation information, which is in turn a basic building block for a management and information support database, along with other information building blocks as topography, soils, wildlife habitat, transportation, and humans factors. Once these data have been developed, it is relatively simple to combine these data with other datasets and expert systems to answer management and scientific questions. It is not easy to dis aggregate information into its basic building blocks, and 3) providing sound, scientific, well documented baseline information is critical to developing a monitoring program. The experience has been that there are many interesting historical datasets that have been developed for national parks, but if good meta data are not available that document the data development process, the data are worthless. 201