Optical and Radar Remote Sensing and Contamination Probability Modelling for the Advanced Quantitative Risk Assessment of Marine Petroleum and Gas Industry

Abstract

Abstract This research focused on the following objectives: (1) using satellite data to characterize the spatiotemporal distribution of anthropogenic oil spills from Oil Rocks Settlement, Chilov and Pirallahi Islands (2) stochastic modelling of the oil spill risk pose to water quality and shoreline ecosystems, and (3) validating model predictions using satellite images. 165 satellite images acquired between 1996 and 2015 were used for the detection of oil spills using object-based classification and visual interpretation. Anthropogenic hotspots were observed at Oil Rocks Settlement, Chilov and Pirallhi Islands, three oldest oil production sites with estimated oil spilling up to 1264 m3 per day. They had different degrees of temporal repetition of oil spills. The largest area (5639 km2) experienced 1-10 detected oil spills, while 993 km2 experienced 11-20 oil spills, 775 km2 experienced 21-50 oil spills, 208 km2 experienced 51-100 oil spills, and 36 km2 experienced 101-150 oil spills. The majority (83% or 6157 km2) of sea surface area within the combined boundary of detected oil spills (7422 km2) had a 50% or greater chance of oil spill contamination, indicating good agreement between predictions and data. Additionally, exponential regression analysis revealed positive correlation between pixel values for contamination probability and calculated oil spill frequency. Approximately 6% (44 km of 751 km) of Azerbaijan’s shoreline had a 50% or greater probability of contamination from oil spills over the period of 2006 -2009 with land use classes sensitive to pollution. This research demonstrates how remote sensing data can be used to identify oil pollution hotspots quantitatively assess the risk to shoreline areas with high environmental value.