Successful protocols for the use of sophisticated synthetic aperture radar (SAR) technologies for such inventories in certain environments were formulated recently (Comer and Blom, 2007a, 2207b, 2007c).
The test area was San Clemente Island, California. Data was collected with the experimental AIRSAR platform and processed to images by JPL/NASA in a special way. Therefore, this technology can not be employed in many places. The current research was intended to expand the range of data sets, technologies, and environments to which key protocols can be applied.
Using only images from commercial vendors of aerial and satellite data, protocols were developed which, based on limited ground truthing, produced signatures for archaeological sites that correlate closely with site locations. This is especially encouraging because the test area for the NCPTT sponsored research was Santa Catalina Island, which has proven too challenging for traditional survey approaches.
Santa Catalina Island is only about 20 miles north of San Clemente Island (Fig. 1), but receives roughly twice the precipitation, and has thick vegetation and terrain extremely difficult to traverse (Fig. 2).
Signatures for both lithic and habitation sites were developed from only the SAR digital elevation model, which is very precise, but could be produced by other more widely available technologies (e.g., lidar), and an image generated from an algorithm run on IKONOS satellite imagery, which can be obtained for almost the entire world (see Figure 3).
Also, protocol enhancements were made that generated what limited ground truthing indicates to be even more precise signatures by combining a third image, a rescaled X-band image (see Figure 4 for a close-up image displaying these results). Arrangements were made with the UCLA Cotsen Institute for a series of field schools that will provide more ground truthing of the signatures.
Finally, prototypical software was developed through supplementary funding by CSRM that reduces the time required to run the protocols from five hours to 15 minutes, paving the way for future fine-tuning of analytical parameters and incorporation of other kinds of imagery (e.g., hyperspectral).