MAPPING OIL SPILL THICKNESS WITH A PORTABLE MULTISPECTRAL AERIAL IMAGER

Abstract

ABSTRACT Rapid determination of oil thickness patterns within a spill at sea is vital for efficient planning and management of spill response activities. Presently such determinations are made almost solely by airborne visual surveys which require specially trained observers and are prone to errors due to variations in illumination, water color, and other environmental conditions. Our goal is to eliminate the subjectivity of visual assessment techniques by developing a computerized portable imaging system that could provide detailed maps of oil-on-water thickness distributions in near-realtime. We have developed oil thickness determination algorithms that utilize multispectral images from a 4-channel sensor providing oil reflectance data in the UV and three channels between 500 and 700nm. A neural network-based algorithm first isolates all oil-on-water areas from oil-free water, sunglint and other potential artifacts. A fuzzy ratio classification algorithm then maps the oil-contaminated areas for thickness classes. The reflectance ratio-based algorithms were tested with several types of crude and fuel oils at MMS’ Ohmsett facility as well as over natural oil seepage areas in the Santa Barbara Channel, California. The methodology yielded accurate thickness estimates over oil films ranging from sheens to 0.2–0.3mm thick. Although the location of thicker films can be accurately mapped, their absolute thickness cannot be established using the UV-visible wavelength range. The addition of an infrared channel may expand the system'S thickness measurement range and is presently being investigated. The imager also allows the identification of emulsified vs. unemulsified oil, thus providing additional information to help guide efficient spill response. Real-time image processing capabilities are presently being developed to allow the system to disseminate a GIS-compatible map immediately after image data acquisition.