Electrical Engineering and Computer Science

Satellite Imagery of Alaska Wetlands Earns Best Paper Award   Bookmark and Share

Prof. Mahta Moghaddam

Jane Whitcomb

Mapping of wetlands in the entire state of Alaska was accomplished for the first time under the direction of Prof. Mahta Moghaddam using radar remote sensing. The process for generating the wetlands map was described in the article, Mapping Vegetated Wetlands of Alaska Using L-Band Radar Satellite Imagery, by graduate student Jane Whitcomb, Prof. Mahta Moghaddam, researchers from JPL, Kyle McDonald and Erika Podest, and Josef Kellndorfer from The Woods Hole Research Center. This article, published in the Canadian Journal of Remote Sensing, vol. 35, no. 1, pp. 54-72, February 2009, was selected as the Best Published Paper in the journal for the year 2009. [Additional Information]

It is believed that the direct impacts of global warming will be most strongly felt in the Earth’s high latitudes in various forms. Wetlands, in particular, are among the most vulnerable ecosystems, which can act as both major sinks and sources of atmospheric greenhouse gases depending on the warming trends and their feedbacks. However, the locations, types, and sizes of northern wetlands are not accurately known; it is difficult to identify and classify wetlands because they are often located in remote and hard-to-access places, they span very large regions (25% or more in Alaska and Canada), and appropriate remote sensing data and analysis techniques have not been available until now.

This paper describes a method for continental-scale mapping of northern wetlands using L-Band synthetic aperture radar (SAR) from space. Using the “Random Forests” statistically based decision tree algorithm along with training and testing data compiled from the National Wetlands Inventory (NWI) and the Alaska Geospatial Data Clearinghouse (AGDC), the first high-resolution synoptic wetlands map of Alaska was generated at the University of Michigan. This wetlands map provides an initial basis for improved characterization of land-atmosphere CH4 and CO2 fluxes and climate change impacts associated with thawing and inundated soils and changes in the size of wetland ecosystems.

Using this method, large-scale boreal wetlands can be mapped and tracked over time, offering important insights into climate change and a quantitative input into global climate models.

Posted: May 25, 2010 by
Catharine June
EECS/ECE Communications Coordinator
cmsj@umich.edu or 734-936-2965

Related Topics:  Electromagnetics   Remote Sensing