Algorithm to Estimate Oil Spill Area Using Digital Properties of Image

Abstract

Ocean accidents typically result in environmental pollution. Among the various causes of environmental pollution, such as ship wastewater or unauthorized oil discharge, oil spill accidents cause damage on the largest scale. Because the oil begins to diffuse instantly when it is spill, the contaminated area increases with time. Therefore, a quick response is necessary. Typically, when an incident is reported, aerial surveying is conducted over the incident area to prepare an initial report. The oil spill area can be calculated using a global positioning system (GPS), side-looking airborne radar, and infrared or ultraviolet scanners (IPIECA, 2016). According to the International Tanker Owners Pollution Federation (ITOPF), oil-contaminated areas are currently estimated using GPS measurements. The US national weather service uses satellites to not only estimate the range of an oil spill, but also to predict its spread. It is known that remote sensing technology can be used to collect information regarding the wind, ocean currents and tides, sea level heights, and other estimates, including the damage from oil spills (NOAA, 2016). Synthetic aperture radar (SAR) technology has been continuously used as a powerful tool for providing various types of information regarding sea conditions. An example of an oil spill region shown in a SAR image is depicted in Fig. 1 (Fiscella et al., 2000). A study was performed to distinguish between pixels with and without oil using the histogram shape of SAR images (Kim et al., 2013). A special model, called the thresholding-guided stochastic fully connected conditional random field model, was introduced to infer binary labels from SAR images (Xu et al., 2015). Machine learning techniques have been applied to find oil spill areas. Topouzelis (2008) Fig. 1 SAR image of oil spill area