Deforestation accounts for up to 15% of global CO₂ emissions. As such, land conversion in biodiverse regions is recognised as a major contributor to climate change. More than half of the world’s deforestation is linked to the production of beef, soy, paper and pulp, and perhaps the most controversial, palm oil.
Palm oil – which can be found in more than 50% of all supermarket products, both in food and non-food categories – is a complicated commodity when it comes to stamping out deforestation. The large proportion of smallholder farmers working in the sector increases the complexity of supply chains and can make tracing the commodity back to its origins challenging.
In Indonesia for example, which together with Malaysia produces 85% of the world’s supply, 45% of palm oil is grown by smallholder farmers. This means food companies can often find themselves dealing with five or six-tiered supply chains.
FMCG heavyweight Unilever is committed to eradicating deforestation from its global supply chains. The company has pledged a net zero deforestation target by the end of 2020, covering the protection of high conservation value, high carbon stock, and tropical peat forests.
As the December 2020 deadline edges closer, Unilever is taking steps to improve transparency in its palm oil supply chains by looking beyond the use of satellite technology currently employed to monitor land-use change.
Through a partnership with California-based tech company Orbital Insight, Unilever will team satellite imaging with geospatial analytics and artificial intelligence (AI). This, the FMCG hopes, will help to shed light on the ‘ground truth’ and ultimately identify the individual farms and plantations mostly likely to be supplying the palm and soy mills in Unilever’s extended supply chain.
Addressing the ‘first mile’ challenge
The first mile – from farm or plantation to mill – presents a particularly unique challenge for sustainable sourcing, a Unilever spokesperson explained.
“Crops such as palm oil fruit or soy can be harvested from lots of different areas of land, belonging to several farmers, and mixed with raw material from other farms or plantations several times before reaching the mill.
“We want to help overcome the challenge presented by the first mile by reimagining traceability through a digital-first approach.”
To date, Unilever has employed an existing methodology used by the industry which relies on satellite imagine. While such systems ‘play an important role in monitoring land-use change’, the company is aware that ‘images alone’ won’t prevent deforestation.
The current technique uses satellite images to draw a 50km radius around the mills. It is assumed that the farms or plantations in those catchment areas are equally likely to be supplying the mills. The methodology is based on the need for palm oil to be processed within 48 hours, the spokesperson explained.
While ‘used’ and ‘accepted’ by much of the industry in recent years, Unilever believes it can now be improved upon, suggesting that a partnership with Orbital will do just that. Describing the move as a ‘step change’ from the current approach, the FMGC is excited at the prospect of modelling supply chain linkages at scale.
“With a clearer picture, it’s easier to estimate any risks, such as deforestation, and take action,” a Unilever spokesperson
The tie-up will kick off with pilot projects at a small number of palm oil mills in Indonesia and soy mills in Brazil.
Counting trucks provides insight into supply chain specifics
Orbital uses AI to transform millions of geospatial data points – from satellite images, location, connected cars and other IoT devices – into a ‘clear picture of what’s happening on the ground’, the US-based tech firm’s CEO James Crawford explained.
In doing so, the technology can ‘reveal the full extent of supply chain activity and relationships’.
“This allows CPG companies like Unilever to pioneer new approaches and understand if their supplier networks are operating sustainably. By tracing the relationships among farms, ports, silos and refineries, companies can see into the difficult ‘first mile’ of supply chains to identify potential issues.
“With this insight, they can make decisions that are good for the planet and good for business.”
To achieve farm-level traceability, Orbital identifies the producers supplying raw materials – in this case, palm oil and soy – to Unilever by geofencing the truck parking lots at known locations in its supply chain, such as a warehouse, refinery, port, mill or silo.
From there, using mobile phone geolocation to track the path of the trucks, Orbital creates a ‘detailed map’ that extends back to the start of the supply chain. “By counting trucks on the trail, Orbital Insight provides granular visibility into the frequency of deliveries and insight into what each farm is supplying,” Crawford told FoodNavigator.
“This way, Unilever can discover the farms, silos, refineries and other nodes in its supply chain, understand their significance and monitor how they are using the land.”
On an ongoing real-time basis, the CEO said Orbital’s computer vision algorithms can analyse satellite imagery and identify any deforestation that has taken place historically or recently at these suppliers.
For example, he continued, if virgin rainforest was levelled for planting or to build roads, buildings, or other infrastructure. In this case, Unilever could then incorporate this information into their sourcing and planning to ensure they fulfil their sustainability objectives.
Potential for expansion?
From Orbital’s perspective, there is potential for its technology to be ‘trained’ for different types of crops and terrain with the use of AI. “Once the supply chain is mapped, Orbital Insight’s AI can differentiate planted versus naturally occurring forest, identifying which farms are unsustainable from high-cadence satellite imagery.”
Further, by combining AI with geospatial data like satellite imagery and location data, the technology can continuously monitor planted and natural forests, grasslands and agriculture, water resourcing, buildings and roads – and importantly, the CEO added, “how this connects back to n-tier suppliers”.
If the pilots are successful, Unilever sees potential for Orbital’s tech to be rolled out across its entire palm oil supply chain. “We’re currently scaling the approach to cover all of Southeast Asia, including Malaysia, and have done some early explorations to see how it may be applied in other palm geographies,” we were told.
Unilever sees ‘immediate potential’ for the partnership to cover palm oil and soy, and revealed ‘the potential is there’ for Orbital to work on other crops in the FMGC’s network, such as cocoa.
Any modern smartphone device, anywhere in the world, can produce geolocation data without network coverage, highlighted Unilever’s spokesperson. “This is a powerful fact, and we are focused on what it means for the positive impact Unilever can make in palm.
“We’re also working to understand how the technology applies in other crops such as wood fibre, cocoa, and other crops where traceability is critical.”