Finding Jim Gray: Multispectral should be in the mix

After a long and extensive back and forth communication with the folks who handle the EO-1 systems (the Hyperion and the ALI), we figured that there were no more option on the hyperspectral side of things. While multispectral sounds insteresting, we would be fighting with the resolution issue.

Initially we thought that with EO-1, even if the resolution was of the order of 30 meters (GSD), we would be expecting to pick up either the green dye or the boat because of the additional information provided by the 220 channels of Hyperion. In other words, use the spectral information to provide sub-pixel resolution. The thinking was then to take images over the bay area in order to calibrate either the green dye or the boat as a target (by having people on the ground reproduce these targets). One would then use these signatures and retask EO-1 to run an autonomous detection program (comparable to the one NASA devised for detecting volcanoes eruptions.) Since we figured it was not feasible within the time constraints we had, we took another look at the multispectral data. We found that Landsat 7, Landsat 4 and 5 and Envisat/Meris had data on the region of interest for the past two weeks. The ground resolution was still an issue.

Then Lawrence Ong (NASA GSFC) came back with this amazing insight/example: Chesapeake Bay on a Sunday when people are using their boats as seen by the ETM+ of Landsat 7.
You can easily see the boats and their wakes. Multispectral imagery should be added to the mix already in place (Radarsat, Ikonos, ER-2, drift models,...).


Any of these ideas deserved to be remembered for future search and rescue efforts.