Detection of Oil Spill Hotspots Using Time-Series MODIS Data (Case Study: Persian Gulf)

Nowadays, oil spill to the seas and oceans is one of the most critical factors in the sea-based pollution in the world. On-time recognition and prevention of oil spills to the marine environment and decreasing of its harmful effects on sea ecosystems are necessary. Remote sensing technology is suitable for early recognition of oil pollution from the air because of the data-collecting in various ranges of Electromagnetism and a broad view of the region. In this study, after evaluation of the scientific sources and the comparison of various sensors and about the time and place of the oil pollution, the images of the second level of MODIS sensor, from 2005 to 2015, have been used. So that a new method for monitoring of the oil spill has been developed. Moreover, the suggested method has focused on enhancment of sensor data. In this study, in order to enhance the bands 1 or 2 oil spill of MODIS sensor, the mean long-term statistical parameters and standard deviation of time-series data have been applied. Therefore, the undesirable data sets for each pixel have been deleted through the initiative process. Then, the mean and the long-term standard deviation of the data have been acquired from the long-term data. Moreover, finally, the value of each pixel has been standardized by utilizing those parameters. The oil spill event has been simulated by this method. Afterward, the normal frequency distribution method was applied to screen the oil patches, and the polluted pixels have been screened by applying  the appropriate threshold. In this way, the oil pollution happened in 2007, and 2010 have been screened by the designed model, and their accuracy has been compared to the real land map. The results of the map for each oil spill event have been acquired as follows: 96% total accuracy and 0.95 Kapa co-efficient for August 2007, 95% total accuracy and 0.92 Kapa co-efficient for July 2010. Finally, the overall results of the study have shown that the designed model has enough accuracy in recognition and screening of oil spills events.