Assessment of offshore oil/gas platform status in the northern Gulf of Mexico using multi-source satellite time-series images.

Accurate knowledge of the detailed status (location and lifetime) of each oil/gas platform is crucial for ocean management and for understanding its potential environmental impact. However, this requirement is often compromised due to incomplete data records, especially when platform abandonment is not properly planned, designed, and/or executed. The availability of multi-source optical and SAR images provides the possibility of a systematic assessment, but the ubiquitous false positives and frequent geocoding errors make it technically challenging. Here, we develop a time-series remote sensing approach (TSRS) for detecting offshore platforms and determining their status, where the technical challenges are overcome using two strategies, as follows: ( i ) A stepwise optimization strategy applied to individual images and time-series images to mitigate numerous noises, to select platforms from candidates, and to assess platform status; and ( ii ) A cross geo-correction strategy using platforms derived from high-geometric accuracy images as ground control points to rectify poorly geo-referenced images. We applied the TSRS approach to the northern Gulf of Mexico using about 26,000 moderate-resolution optical/SAR images from eight satellites spanning the period from 1982 to 2017. TSRS detected a total of 9260 (5556) platforms in both Federal and State waters, including 3140 (1803) existing platforms, 5044 (3426) removed platforms, and 1076 (327) status-to-be-determined platforms where the numbers in parentheses represent those from Federal waters. Of these TSRS platforms, 4751 coincide spatially with 5476 (77.68%) of platform records of the Bureau of Safety and Environmental Enforcement (BSEE) platform database; 2846 (387 in Federal waters) are not found in the BSEE platform database but are convincingly supported by data from other sources (including BSEE-pipeline-database, time-series VIIRS nighttime light products, and littoral ground-truth images), meaning that 82.04% of the TSRS-derived platforms can be directly/indirectly confirmed. A site-specific validation in East Bay using multi-temporal ground-truth aerial images indicates that, for this site, the false negative rate (FNR) is ~9.63% and the false positive rate (FPR) is only 4.44%. The detailed beginning/end dates of the TSRS platforms are mostly confirmed by the corresponding Installation/Site-clear dates recorded in the BSEE platform database. Our findings suggest that the TSRS approach is a robust and cost-effective way to assess oil/gas platform status in large regions, and our results here may complement the official database, narrow down locations for future field investigations, and provide critical information for future assessment of the potential impacts of active and decommissioned platforms on the marine environments. [ABSTRACT FROM AUTHOR]