The Inventory and Monitoring (I&M) Program of the National Park Service (NPS) sought to provide reliable and consistent land cover
information for Katmai National Park and Preserve (KATM) that may be used to assess the status and trends of national park ecosystems. GRS was selected based on our unique land cover mapping techniques, which enabled the estimation and mapping of detailed land cover characteristics.
GRS performed landcover mapping on the 4-million acre project area with the goal of providing detailed vegetation information that is useful for making resource management decisions on a park-wide basis. The project was undertaken by GRS, in association with the Alaska Natural Heritage Program (AKNHP), and with support from the Alaska Regional Office (ARO). GRS directed training site selection, navigation, and verification components of field data collection efforts, as well as all image data acquisition, preparation, training, evaluation, and image classification. The AKNHP provided expertise regarding vegetation systems, description, and species identification during field data collection efforts. Simultaneously, NPS personnel carried out ground vegetation transects concurrent with aerial surveys and managed field data collection logistics.
GRS initiated land cover classification and mapping efforts of the 4-million acre project area in mid-2000 for the Alaska Support Office (AKSO) of the National Park Service. Data preparation entailed the development of eco-type masks, field maps, field data collection procedures, and illumination correction of three scenes of Landsat TM imagery. GRS used extensive preprocessing of the imagery to define field data sampling plans that maximized the collection of useful information and minimized the collection of redundant and erroneous field data. Imagery was corrected using GRS’s illumination correction software. Sample sites were identified using unsupervised ISODATA classification processes to stratify the project area and map large homogeneous areas suitable for field sampling. Aerial and ground field data collection efforts were completed during a three-week period in late July-August, 2000. Detailed field data were collected over the entire project area to provide the land cover descriptions that both guided the classification efforts and described the resulting land cover features in the final classification map. These training data sets encompassed the complete range of significant land cover characteristics present in the project area. The image training sets were developed and tested to resolve class confusion using GRS Confusion Reports and Fidelity Tests.
GRS used both supervised and unsupervised classification combined in a unique Discrete Classification process that preserves the integrity of the field data and allows for the detailed landcover characteristics to be maintained through the entire classification and mapping process. Detailed quantitative land cover data descriptions were developed using an AKSO-modified Viereck Alaska Vegetation Classification System. The map data set inalso included species-specific estimates of major species, as well as for significant landscape features. The resulting landcover map data set consists of more than 1,000 individual pixel classes that are each linked to the detailed field data. Since these maps retain the complete range of characteristics measured in the field, the data can be easily modeled or generalized into specific categorical maps such as bear habitat or wildfire fuels. Estimates of map accuracy were reviewed by NPS staff and the final map was accepted in December, 2004. GRS classification efforts were completed and the final raster map data set, along with descriptive reports, were delivered in December, 2002.
The complete Katmai report may be downloaded at the following NPS link: https://irma.nps.gov/Reference.mvc/DownloadDigitalFile?code=152196&file=LC_KATM_2003_2.pdf
