Jul 3, 2009
Mapping wet landscapes in more detail
Researchers at Columbia University in New York, US, and the University of Western Ontario, Canada, can now map wetlands at higher resolutions using remotely sensed data from satellites. Their new technique could be important for improving the management of landscapes such as wet forests.
Yasir Kaheil of Columbia and Irena Creed of Western Ontario aimed to map wet areas at high resolution so that forest managers can better assess land in the future without disturbing ecosystems, for example by having to build a road that cuts through a wetland. The researchers used data from the European Remote Sensing 2 (ERS-2) synthetic aperture radar (SAR) satellite, the Landsat Thematic Mapper (TM) and Light Detection and Ranging (LiDAR). They downscaled coarse resolution images to finer resolution ones using a special algorithm that simply models the details at each image scale to reconstruct new images at ever finer scales.
The team applied its algorithm to a remote boreal landscape in northern Alberta, Canada. The landscape has a relatively flat terrain consisting of low hills with small isolated wet areas as well as large complex wetlands. Its climate is continental with cold long winters and short cool summers with average annual temperatures reaching 1.7 °C, and an average of 503 mm of precipitation per year.
The technique was able to reproduce a provincial government map of wet areas as large as 25 m across and downscale it to a spatial resolution of around 6 m. Misclassification errors were less than 2%.
Although in the past hydrological models have been used to detect how wet areas form, they have not been able to determine how these areas are distributed throughout the boreal watershed. These earlier models could accurately predict run-off but not the wet areas contributing to the run-off. This was especially true for the area studied by Kaheil and Creed, where data was lacking.
"What is new about our approach is that it makes no assumptions about what the final results should look like," Kaheil told environmentalresearchweb. "It just lets the input data shape the outcome image."
The algorithm can also easily be transferred to landscapes other than wetlands. "This is in contrast to other approaches that are case-study dependent," says Kaheil. Indeed, future work on this project will look at how robust the model is for regions characterized by more complex topographies.
Kaheil is now working on climate risk management through optimal water allocation and storage and/or financial hedging strategies, such as index insurance, at the International Research Institute for Climate and Society at Columbia University.
The boreal wetlands work was reported in IEEE Geosciences and Remote Sensing Letters.
About the author
Belle Dumé is a contributing editor to environmentalresearchweb.