Abstract—Petroleum is a major pollutant that leads to climate change primarily through its combustion and the release of greenhouse gases, especially carbon dioxide. Although there is considerable interest in reducing our reliance on petroleum, to date there are limited, cost-effective alternatives to petroleum-based technologies. Petroleum leads to pollution in the oceans directly through oil spills and indirectly through the release of petroleum-based products such as plastics and tyre particles. More generally, oil spills jeopardise public health, pollute drinking water, destroy natural resources, and disrupt the economy. Oil spill response in the oceans must account for near real-time tracking of the movements of the oil to minimise damage and reduce the time for environmental recovery. However, oil patches appear on the water surface usually after the oil has been horizontally transported and vertically dispersed from an underwater source of a spill. Therefore, obtaining the 3-dimensional spatial distribution of the oil plume is critical for rapid and efficient oil spill response. Also, a non-contact method of investigation in the initial stages prior to response activities is desired in order to have as little influence on the plume as possible. In this paper, we present a set of scanning sonar tests, using a Ping360 sonar, which were conducted in Lake Barrington, Tasmania, Australia. We tested two types of relatively environmentally-friendly proxies to model patchy plumes of oil droplets. Based on the field experimental results, a method of real-time analysis of the in-situ sonar records is presented and it is shown, in simulation, how this approach can be used by an AUV to delineate an oil plume.