Alaska Science Center
ABOUT THE ALASKA
Alternative Approaches for Remote Sensing of Forest and Shrub Cover in Arctic Tundra and Boreal Forest Environments
The recent completion of the National Land Cover Database product in Alaska allows for a valuable first look at the distribution of broad vegetation types across the state. A comprehensive characterization of Alaska’s vegetation, however, requires detailed information about vegetation structural characteristics such as canopy density and height that are not included in the Alaska NLCD database. This project explores and compares the utility of several remote sensors for mapping height and density characteristics of forests and shrubs in Alaska’s arctic tundra and boreal forest at a moderate spatial resolution (250 m).
This project aims to explore two key questions: (1) at a moderate spatial resolution (250 m), which sensor is most effective for mapping vegetation density and height in arctic tundra and boreal forest environments and (2) can a multi-sensor data fusion approach provide more accurate vegetation density and height estimates than would be possible with data from any single sensor. In answering these questions, it will be necessary to consider not only how accurately a sensor can map vegetation density and height when the ideal dataset from that sensor is available (e.g. cloud-free, mid-summer imagery for multi-spectral sensors), but also the probability of actually acquiring a dataset with these ideal characteristics as well as how accurately a sensor can map vegetation density and height when less than ideal datasets are the only ones available.
Early work on this project has focused on comparing the ability of the Landsat, Multi-angle Imaging SpectroRadiometer (MISR), and Moderate Resolution Spectroradiometer (MODIS) sensors for mapping shrub cover in the arctic. Work has focused on a study area paralleling the Dalton Highway on Alaska’s North Slope. Shrub cover in this area is typically confined to stream drainages (figure 1), floodplains of large rivers (figure 2), and isolated hillslopes (figure 3). During the summer of 2008, field crews collected shrub cover density and vegetation height measurements along 61 transects covering the full range of spectral variability contained in fine resolution IKONOS imagery covering the study area (figure 4). These measurements were then used along with the IKONOS imagery to produce fine resolution maps of shrub canopy density and vegetation height for the entire study area. Spatially degraded versions of these maps will be used to train regression tree models using multi-spectral data from Landsat, multi-spectral/multi-angular data from MISR, and multi-temporal data from MODIS. The models will be used to produce shrub canopy density maps that can be compared to the original IKONOS-derived maps and assessed for accuracy.
Principal Investigator: David Selkowitz (firstname.lastname@example.org, 907 786 7146)