|
|
2008 Alaska Park Science Symposium
October 14, 2008
An Ecological Land Survey for the Wrangells-St. Elias National Park and Preserve
Torre Jorgenson1, Ken Stumpf2, Joanna Roth3, Trish Loomis4, Tim Cater5, Erik Pullman6, Michael Duffy7, Wendy Davis8
1ABR, Inc., PO Box 80410, Fairbanks, AK, 99709, USA, Phone 907-455-6374, tjorgenson@abrinc.com
2Geographic Resource Solutions, Arcata, CA, 95521, USA
3ABR, Inc., USA
4ABR, Inc., Fairbanks, AK, 99709, USA
5ABR, Inc., USA
6ABR, Inc., USA
7ABR, Inc., USA
8ABR, Inc., USA
We performed an ecological land survey for the Wrangell-St.Elias National Park and Preserve (53,352 km2) during 20032008 that included integrated field surveys, ecological classification, and landcover and ecological mapping. Field surveys at 569 intensive plots collected information on the topographic, geomorphic, hydrologic, pedologic, and vegetative characteristics of boreal and maritime ecosystems across the entire range of environmental gradients. Individual ecological components (e.g., geomorphic unit, Alaska vegetation classification) were determined using standard classification schemes for Alaska. We also developed 67 plant associations through multivariate classification techniques. We used the hierarchical relationships among ecological components to develop 68 ecotypes (local-scale ecosystems) that best partition the variation in ecological characteristics across the entire range of aquatic and terrestrial environments. Soils described at 423 plots were classified into 53 soil subgroups. Two types of maps products were developed: landcover maps that use vegetation classes similar to the AVC classification, and ecosystem maps derived from the landcover maps through rule-based modeling. GRS developed the primary landcover map by preprocessing of 11 Landsat ETM scenes; developing unsupervised classifications to guide field surveys; developing spectral training areas by sampling spectrally homogenous patches by helicopter; developing a database linking spectral and vegetation characteristics; evaluating spectral signatures; classifying the vegetation type of each spectral signature using cut-point rules for the spectral database; performing a supervised classification of the scenes using the classified signatures; and reducing errors in the resulting scenes through rule-based modeling with ancillary data. We then develop an ecotype map, which differentiates closely related geomorphic, soil, and vegetation characteristics into 67 classes, through rule-based modeling involving maps of landcover, climatic regions, physiography, elevation, and slope. We then aggregated the 67 ecotypes into 25 soil landscapes based on the landscape-relationships analysis. This linkage of landcover maps with climatic, physiographic, and topographic variables to develop ecosystem maps will improve the ability of scientists and land managers to predict the response of ecosystems to human impacts, natural disturbance, and climatic change.
Previous Abstract | Next Abstract
|
