Send this article to a friend:

February
15
2017

Artificial Intelligence To Reveal The Biggest Secret In Oil
Tsvetana Paraskova

The EIA is, and will continue to be, the gold standard for reporting U.S. oil inventories, and the administration’s weekly reports have the power to immediately swing crude oil prices depending on builds or draws in oil stocks. But while the U.S. is a transparently reporting country, other nations are not revealing oil stocks data, or are rarely reporting—at best—opaque figures in which the markets have little faith.

No one really knows how much oil the countries around the world are storing, creating uncertainty in the supply side of the oil markets.

But one of the hottest new technologies may shed more light on oil storage around the globe, especially in countries that are keeping inventory figures to themselves. The tech is artificial intelligence (AI) - and an unlikely collocation to put in a sentence with oil and inventories.

Yet, U.S. geospatial analytics company Orbital Insight has been using a form of AI - convolutional neural networks (CNN) – to analyze satellite images and identify and quantify crude oil storage tanks. The tanks have floating roofs, so the volume of oil is visible. Orbital Insight is using shadow-detection technology and calculates how full a storage tank is by the size of the crescent-shaped shadow on the tank roof.

That approach to estimating oil storage capacity and volumes may reveal how much oil closed economies have or plan to have. One such nation is China, a huge consumer of oil and oil products—a closed economy not eager to share its reserves and storage figures with the world. But China’s storage has the potential to sway global oil markets.

Last year, Orbital Insight said that it had found that there were 2,100 commercial and strategic petroleum reserve tanks across China, with the capacity to store 900 million barrels of oil as of the end of 2014. That’s four times more than the 500 tanks reported in the industry-standard database of tank farms at TankTerminals.com. Orbital Insight’s estimates showed that China had around 600 million barrels of oil supply on its territory as of May 2016—and that’s not counting underground storage. Related: Is $60 Oil Within Reach?

Orbital Insight has tracked U.S. and China oil storage so far and is currently analyzing this data for the world. Its plans are to launch oil storage estimates for countries like Russia, Brazil, India, Venezuela, Angola, Nigeria, and Iran.

If AI analyses of satellite images can reveal how many oil storage tanks these countries have, and how full they are - above ground at least – they could shed more light on oil supplies around the world.

For now, tracking and reporting crude oil supply, storage, and flows is being done (if at all) in a variety of ways depending on the country: some estimate supply by calculating domestic consumption and tracking tankers, others just give some figures without revealing details, and a third group of nations report supply by collecting and aggregating data from companies.

“These are also prone to misinformation and are sometimes not retroactively reviewed in case some companies did not report in the survey,” Michael D. Cohen, an analyst at Barclays, has recently told Rigzone.

Although “satellite technology can tell you a lot, but not everything” - as Sandy Fielden, director of research for commodities and energy at Morningstar, told Rigzone – AI may have the power to increase, even if just a little bit, transparency in oil storage and inventories reporting. Because as sensitive as the oil industry is, there will always be countries that will not be sharing transparent and independent oil inventory figures.

By Tsvetana Paraskova for Oilprice.com

 

 

 

 

Tsvetana is a writer for the U.S.-based Divergente LLC consulting firm with over a decade of experience writing for news outlets such as iNVEZZ and SeeNews.

[Most Recent Quotes from www.kitco.com] [Most Recent USD from www.kitco.com] [Most Recent Quotes from www.kitco.com]

Send this article to a friend: