Abstract:
We have used two seasons of L-band SAR imagery to produce a thematic map of wetlands throughout Alaska. The classification was developed using the Random Forests statisti...Show MoreMetadata
Abstract:
We have used two seasons of L-band SAR imagery to produce a thematic map of wetlands throughout Alaska. The classification was developed using the Random Forests statistical decision tree algorithm. Input data included mosaics of summer and winter JERS-1 SAR imagery with associated image collection dates, summer and winter SAR backscatter texture, elevation, slope, proximity to water, and geographic latitude. The accuracy of the resulting thematic map was quantified using extensive ground reference data. The overall aggregate accuracy calculated based on all classified pixels was 89.5%, with individual per-tile aggregate accuracies ranging from 80% to 97%. As the first high-resolution large-scale synoptic wetlands map of Alaska, this product provides the basis for improved characterization of land-atmosphere CH4 and CO2 fluxes and climate change impacts associated with thawing soils and changes in extent and drying of wetland ecosystems.
Date of Conference: 23-28 July 2007
Date Added to IEEE Xplore: 07 January 2008
ISBN Information: