Doctoral Student Takes Her Archeological Imaging Research On the Road
by Amy Mednick
Just back from a conference in Rio de Janeiro where she presented her research and received an international award, CIS doctoral student Kelly Canham is gearing up to pack her toothbrush, phrase book, and a spectroradiometer for a trip to Oaxaca, Mexico in December. There Canham will collaborate with William Middleton, associate professor of sociology and anthropology, to take ground-based spectral measurements to fill in some of the gaps left after analyzing the satellite data of the Nochixtlan Valley.
“The project in Nochixtlan will be for ground truthing various landscape taxa that Kelly has identified, and taking on-the-ground spectral measurements to better identify and interpret the satellite data,” Middleton says.
Canham found herself on her way to the Latin American GeoSpatial Forum in Brazil this August after winning Phase 2 of the Digital Globe 8-Band Research Challenge—one of five winners out of roughly 300 applicants worldwide. Canham proposed to apply an algorithm that she developed for examining hyperspectral data to Digital Globe’s multispectral WordView-2 satellite data. The Digital Globe WorldView-2 satellite data includes only eight spectral imaging bands compared to the 100s typical of hyperspectral data. The results Canham presented at the conference surpassed what was expected for a multispectral dataset.
Canham and David Messinger, Canham’s advisor and director of the Digital Imaging and Remote Sensing Laboratory, are developing image-processing tools to analyze hyperspectral satellite images so that Middleton can better understand the area of Oaxaca the Zapotec civilization once populated. Previously, archeologists had taken advantage of remote sensing imagery to identify individual sites by eye or to use a few spectral bands to find archaeological markers in the vegetation and terrain. The hyperspectral satellite data, obtained by the Hyperion sensor aboard NASA’s Earth Observing 1 satellite, includes many more spectral bands that form a data cube and allow researchers to map the area in more detail.
Canham’s algorithm allows her to automatically analyze and identify the spectral signature of the materials located within each pixel in hyperspectral, and now multispectral, satellite images. The algorithm allows Canham to hypothesize, for example, that a given pixel might contain a small fraction of a material with a prominent spectral signature, which might be indicative of limestone rock. A larger fraction of that pixel might have a weak spectral signature, which could indicate an asphalt road. However, Canham and Middleton cannot verify this hypothesis until they actually survey the area by land to determine if the spectral signatures obtained through the analysis indicate the presence of limestone and asphalt.
In order to take the measurements necessary to create a spectral library of the area, Canham will use the FieldSpec Pro spectroradiometer. She won the use of this instrument from the Alexander Goetz Equipment Program last spring. A CIS microgrant will fund Canham’s travel expenses to Mexico. ( See http://www.rit.edu/news/story.php?id=48248)
The eventual goal is to create a viable land-use map to add to Middleton’s research on the Zapotec civilization. “Overall, what we can do with hyperspectral data will never replace archaeologists, but we can help them look for "candidate" sites of interest,” Canham says. “Our goal is just to make their life a little easier and possibly, eventually, a bit cheaper than the old ground walking surveys.”
Last Modified: 12:04pm 07 Oct 11