Related News

Surecomp integrates Cleareye onto trade platform

Banks must adapt to survive as commodity finance demand slumps

Sinopec executes ‘world-first’ first oil product eBL trade

Yap departs MonetaGo, returns to banking

GSBN, Cosco complete blockchain-based biofuel certification pilot

Singapore bank OCBC is trialling the use of deep learning satellite technology in a bid to reduce risks in the financing of crude oil.

The bank has engaged specialist data analytics firm ImageSat International (ISI) to estimate crude oil levels in tanks stored in undisclosed locations in Asia, using optical images collected via satellite. Data taken from the images, collected over a period of time, help build an algorithm which gives OCBC a 90% accurate gauge of the oil it has financed.

Since the information is linked directly to the commodity rather than the terminal at which it is stored, it eradicates the potential for fraud and inaccuracy in the reporting of the bank’s collateral position, the bank says.

GTR can reveal that OCBC has successfully completed a proof of concept with ISI and is looking to trial the technology further in other commodity verticals, where the bank provides storage and project financing against an underlying collateral position.

Phase two of the pilot, which will commence shortly, will add additional infra-red satellite imagery to boost the accuracy of readings to 95% since infra-red satellites can penetrate cloud cover and therefore sidestep potential adverse weather conditions.

“The first phase focused on oil inventory in Asia, looking to see if we could get an accurate independent estimation of oil in tanks that the bank is financing. That took some time, we had to identify where the bank would want to do this, and decided on a location in Asia. We worked with ISI over the course of months and came up with an accuracy level of 90% on the oil inventory we financed. We closed that off in January 2018. After this phase, we’re going to look at the second and third phase, focusing on metals and agricultural products,” Barend van IJsselstein, head of energy commodities at OCBC, tells GTR.

OCBC claims to be the first bank to apply this technology to the commodity finance sector. However, it has also been deployed to analyse oil inventory stocks by organisations including hedge funds. For instance, funds are using it in Cushing, Oklahoma – a major oil trading hub – to help predict the price of crude.

The use of satellite technology is also growing among commodity traders, many of which are using it as a means of tracking their commodity stocks from the source to the point of sale.

This technology combines advancements in satellite and smart technologies. Using shoe-box satellites, which come at a fraction of the cost of traditional, more cumbersome satellites, ISI takes pictures of the co-ordinates provided by OCBC, from space.

The company then uses its deep learning software to analyse the images. The algorithm scans the oil inventory for storage tank depletion, looking at the sinking level of the lid and the shadow cast by the sun on the inside of the tank. Detecting these patterns allows analysts to estimate how much oil is in the tank.

Furthermore, using bandwidth purchased from infra-red radar satellites, then analysing those images, the company can gauge the contrast in the temperature between the oil stored inside the tank, the tank itself and the gas-filled spaces in the tank to predict the volume of oil stored in the tank.

“We are providing the ability to monitor areas or locations from space,” Liron Vine, head of marketing at ISI, tells GTR. “We then apply our comparison analytics in which we identify the changes spotted in the area since the last picture that we sampled and over several periods of time. This enables us to assess the status of the pictured area. For example: if it’s a site – is it active or not, or are construction works progressing and what phase they are in? Another example: we can provide precise volumetric measurements of crude oil tanks with floating lids. By measuring the lid height – that is changing according to the oil volume – we can calculate exactly how much oil is in every tank.”

One of the main purposes of this sort of technology is to reduce the risk of fraud, which continues to plague certain commodity markets in Asia. High-profile cases in recent years include the Qingdao metals fraud, which exposed many banks to fraudulent warehouse receipts, obtained multiple times against metals stocks which may not have existed.

Last year, ANZ was exposed to a warehouse receipts fraud in the nickel sector, worth more than US$300mn. The bank was left with ownership of 83 fraudulent warehouse receipts which pertain to cargoes of nickel stored at Access World warehouses in Singapore and South Korea.

These instances, combined with the multitude of fraud cases that go unreported every day, have left banks looking to fintech for answers. One of the most frequently-cited pluses of blockchain technology is that it can help eradicate double financing, since data entered onto the blockchain cannot be altered. Other tools such as LMEshield are geared towards stopping warehouse receipt fraud.

Satellite technology that is able to track and monitor commodity stocks offers some assurances to banks that these cargoes do, indeed, exist and that they are what they claim to be.

“To my knowledge there hasn’t been a fraud case in oil in Asia for quite a long time. However, fraud in general in commodity finance remains a big risk factor, especially as the dollar amounts in oil and energy are quite large,” van IJsselstein says. “Anything a bank can do to reduce risk in financing is something that’s very useful. We see this as one of those additional tools. As we go along the process we are discovering extra benefits, such as reducing travel costs and CO2 footprint. These are side benefits, but the main benefit is that we can enhance our risk management and mitigation process.”

ISI says the technology is “highly scalable” and can be used across commodities, tracking ship traffic and following the stocks through the entire supply chain.

“We can quantify everything that is visible from containers, cars in parking lots and finished goods in factories and through that, create customised indexes,” Vine explains.