SkyTruth’s project Cerulean helped intern Breanna Xiong document Trinidad and Tobago’s chronic oil pollution problem.
Trinidad and Tobago – the small dual-island Caribbean nation – has been a top exporter of liquefied natural gas for years. Because of its significant oil and gas exploration, and what some Trinidadians would call poor corporate and government responsibility, oil leakages from offshore infrastructure are very common events in Trinidad’s waters. With the help of satellite imagery, SkyTruth has been able to directly observe and continuously monitor Trinidad’s chronic oil pollution.
Between June 2020 and November 2021 alone, we’ve estimated the cumulative amount of oil that has leaked from offshore infrastructure and into the Gulf of Paria and eastern coast of Trinidad to be between 55,740 and 260,120 gallons. That’s at least twice the size of the most recent oil spill in Huntington Beach, California and up to 10 times as large.
Our analysis used similar methods outlined in our 2012 and 2017 spill reports on the Taylor Energy spills. However, we did not use public pollution reports filed with the National Response Center to estimate our spill areas like our previous reports did. Instead, we used oil spills detected by SkyTruth’s project Cerulean. Cerulean uses artificial intelligence to identify patterns of oil pollution in satellite images. In our case, we used radar images from the Sentinel-1 satellites to detect oil spills. With much help from Jona Raphael, SkyTruth’s machine learning expert, we were able to pull these direct observations from Cerulean for our analysis.
Here’s how we did it:
- Using Cerulean’s detections, we computed an Average Daily Slick Extent (Area) estimate. Since no machine learning model is perfect, we paired the detections with Christian Thomas’s 2020 offshore infrastructure dataset to manually filter out spills that were potentially from sources other than offshore infrastructure.
- From the Area estimate, we derived an Average Daily Slick Volume. There was one key assumption we made while doing this derivation:
- For our assumed average slick thickness, we used a conservative standard of 1 micron (1 millionth of a meter) and an even more conservative estimate of 0.5 microns to reflect the low-end range of possible oil thickness.
- Using the Average Daily Slick Volume, we derived a Daily Flow Volume. There was one key assumption we made when doing this derivation:
- We assumed that one half of a given amount of a thin slick of oil on the surface of the ocean would degrade in three to seven days. We believe this is a conservative range based on published figures illustrating a range of half-lives for oil mixtures. Longer half-lives would mean less Daily Flow Volume and subsequently, a lower Cumulative Annual Emission.
- Since we’re interested in annual emissions, we multiplied our Daily Flow Volume by the number of days in a year. With this step, we have an estimate of our Cumulative Annual Emission.
After multiple calculations with different thickness and half-life scenarios, we calculated four values for estimated cumulative oil spilled:
Oceans play a vital role for island nations. Their lifestyle, culture, economy, and sense of place is heavily dependent on their surrounding environment. Continuous oil pollution events will place a long-lasting burden on Trinidadians. SkyTruth will continue to monitor Trinidad and Tobago. We hope that as our technological capabilities expand, we can perform more analyses to document oil pollution in Trinidad and Tobago and support concerned citizens and activists working to protect their local environment.