MULTISPECTRAL INFORMATION IN OPERATIONAL MARINE POLLUTION MONITORING: A DATA FUSION APPROACH *

Operational marine pollution monitoring requires a complex network and communication structure to be operated by a single operator in real time. Applying state-of-the-art remote sensing equipment like side looking airborne radar, IR/UV line scanner, microwave radiometer, imaging airborne laser fluorosensor and forward looking infrared, and combining them effectively in a sensor network, a variety of different information is av ailable at the operators console for the detection and analysis of spills aboard the aircraft. Recently the combination of the different (multispectral and multidimensional) sensor results towards condensed, application oriented information has met with increasing interest. Thus new application fields and parameter sets could be derived, like improved classification and distribution calculations. Results from routine operation of the German maritime surveillance aircrafts operated over North and Baltic Sea by the 3 rd Naval Air Wing on behalf of the German Ministry of Transport are shown to illustrate the successful application of the MEDUSA sensor network in maritime patrol of oil slicks and polluters. Data fusion methods will be considered as a tool for value adding and decision support during operation. Different approaches will be examined, including multispectral, and multidimensional fusion by overlay techniques and nonlinear dynamical attractors. Finally conclusions will be drawn in context to operational aspects and further steps will be outlined towards improved classification with neural networks and the usage of external databases or satellite information.