VOLCANIC HAZARDS ASSESSMENT USING REMOTE SENSING DATA FROM THE TRES VIRGENES VOLCANIC COMPLEX, BAJA CALIFORNIA SUR, MEXICO

The Tres Virgenes Volcanic Complex (TVVC) is located on the east coast of central Baja California. The TVVC is composed of three volcanoes bearing a NE – SW trend. El Viejo, the oldest edifice lies at the northeasternmost point of the complex. In the center lies El Azufre, followed by the youngest of the three, La Virgen. This geographical setup indicates a SW migration of the volcanic activity. Volcan La Virgen is a Quaternary calcalkaline stratovolcano. It has an elevation of approximately 2000 m and a volume of nearly 15 km³ (Capra, 1989). The area surrounding La Virgen has active fumaroles, hot springs and recently recorded shallow seismicity. Potentially significant hazards exist for the towns of Santa Rosalia and San Ignacio (35 km from La Virgen), as well as for air and ground transport; the major peninsular highway lies only 3 km from the south flank of La Virgen, atop a recent lava flow.

The MASTER (Modis-Aster) airborne simulator collected overflight data for the TVVC. The data set includes 50 spectral bands in the visible through infrared wavelengths and high spatial resolution (~5 meter pixel size). By assigning wavelengths to different colors in a spectral image, rocks that look the same to the naked eye show up in different colors according to their composition. The data was processed using ENVI software, which allowed to overlap flight lines to form mosaic images. Band ratios are typically used to enhance the spectral differences between bands. By dividing a spectral band by another it produces an image that provides relative band intensities. The band ratios commonly used in this project were 2/13, 5/13, 8/21 RGB. Supervised classification was used to cluster pixels in a data set into classes corresponding to user defined training classes. These training sites were then defined as multiple irregular polygons. Michael Sawlan’s geologic map of the TVVC was used to determine what kind of rocks would be classified and where they could be found. By placing polygons over a designated area one would then match the region to Sawlan’s map, and give the region the indicated lithologic name. ENVI would then assign every pixel to its corresponding class. After a couple different regions of interest were set up, a classified image was created and a new geological map based on the rock’s spectral signature was created.