Forests vs palm oil

While there have been improvements in deforestation rates in some countries in 2022, as reported by the World Resources Institute, notably in Brazil and Indonesia due to government policies, heightened awareness, and sustainable business practices, the problem persists.

Agriculture, logging, and mining remain the primary drivers of deforestation, with commodities such as palm oil, soybeans, beef, and wood products playing pivotal roles. Palm oil, a versatile ingredient in food, cosmetics, and biofuels, is in high demand, driving the expansion of palm oil plantations. Soybeans, a major crop for livestock feed and vegetable oil production, also experience increased demand, leading to expanded cultivation. Beef, a popular meat globally, sees rising demand, contributing to the expansion of cattle ranching. Lastly, wood, essential for furniture, paper, and construction materials, faces growing demand, spurring logging expansion. Addressing these challenges is crucial in combatting deforestation's global impact.

Explaining the regulatory demands

What is the EU Regulation on deforestation-free supply chains?

The EU Regulation on deforestation-free supply chain products (EUDR) is a landmark piece of legislation that aims to reduce the European Union's (EU) contribution to global deforestation. The regulation, which entered into force on June 29, 2023, applies to seven commodities: beef, cocoa, coffee, palm oil, rubber, soy, timber, and wood pellets.

What are the key requirements of the EUDR?

The EUDR requires companies that place these commodities on the EU market, or export them from the EU, to ensure that their supply chains are deforestation-free. This means that the commodities must not have been produced on land that has been deforested or degraded after December 31, 2020.

What do companies have to do to be compliant?

They have to carry out a risk assessment to identify and assess the deforestation and forest degradation risks in their supply chains. This can be done using satellite data to track deforestation and forest degradation.

They also have to take measures to mitigate these risks, such as working with suppliers to adopt sustainable practices.

And they have to keep records of their due diligence activities, as the regulation stipulates. Hence, they will have to document the tools that they used to ensure that no deforestation is linked to their supply chain products.

What are the enforcement mechanisms for the EUDR?

The EUDR includes a number of enforcement mechanisms, such as fines and injunctions. This means that companies that fail to comply with the regulation could face significant penalties.

What are the potential impacts of the EUDR?

The EUDR is expected to have a significant impact on the global fight against deforestation. It is estimated that the regulation could reduce the EU's contribution to global deforestation by 71.92 kha, or approximately 100,728 football pitches of forest per year and reduce CO2 emissions by 31.9 million metric tons per year (the State of the World’s Forests report 2020 estimates a global deforestation rate of 10,000 kha per year mainly driven by production of cattle, oil palm and soya (FAO, 2020)).

What are the success factors for this regulation?

Many have emphasized that the success of this legislation will ultimately depend on the effectiveness of its monitoring and enforcement mechanisms. To ensure that the regulation achieves its intended goals, rigorous monitoring and verification of supply chains will be crucial.

How can satellites help to implement the European Deforestation regulation?

Earth observation satellites have been used since the 1980s to detect changes in landcover and thereby also to detect and understand large-scale deforestation.

However, a significant transformation has occurred in the past seven years. Advancements in satellite technology, heightened global recognition of deforestation's environmental consequences, the enhanced transparency offered by accessible data platforms, and the cost-effectiveness of satellite-based surveillance have all combined to establish satellite technology as an essential instrument for monitoring and combatting deforestation trends. This development allows for better-informed conservation initiatives and responsible forest management practices.

By monitoring changes in tree cover, satellites can help to identify areas that are at risk of deforestation and take action to prevent it.

One way that satellites are used to monitor deforestation is by comparing images of the same area taken at different times. This can be used to see if there have been any changes in tree cover, such as if trees have been cut down or if forests have been degraded.

Another way that satellites are used to monitor deforestation is by using radar imagery. Radar imagery can see through clouds and plants, so it can be used to monitor deforestation in hard-to-reach places. Radar imagery can also be used to measure the height of the forest canopy, which can be a sign of forest degradation.

Satellites can also be used to track deforestation in real time. This is done by using satellites that are equipped with sensors that can detect changes in the forest cover on a daily or even hourly basis.

The use of satellite data to monitor deforestation is a powerful tool that can help protect forests and fight climate change. By using satellites, we can identify areas that are at risk of deforestation and take action to prevent it. This is essential for protecting the world's forests and the vital services they provide, such as climate regulation and biodiversity conservation.

The case of Malaysia

We looked into the situation in Malaysia as this was flagged as a country by the annual deforestation report of the Global Forest Watch (GFW), an initiative hosted by the World Resources Institute (WRI). In Malaysia, deforestation reduced drastically in very recent years due to strong policy action, but is nevertheless still persistent to some degree, the report said. Both Indonesia and Malaysia had been very critical of the EU deforestation regulation and also criticized it publicly.

The report by Global Forest Watch also acknowledges that strong political action brought down the numbers of logged forest. Observers have however also cynically noted that there is little left to deforest, as sizable areas of suitable land had already been converted into plantations.

The Round Table on Sustainable Palm Oil (RSPO) has also broadly contributed to a reduction in deforestation and a large part of palm oil plantations are now RSPO certified (the area is around ⅔ of the total forest area in Malaysia).

• Total forest area (ha): 260.000 ha (260 Kha)
• Total Tree cover loss as per GFW in 2022: 71.900 ha (71.9 Kha)
• Total RSPO concession (ha) in 2022: 176.083 ha
• Total alleged deforestation within 13 examined RSPO concessions: 5.561 ha
• Total alleged (but not confirmed) deforestation within 8 more closely examined concessions: 3.785 ha
• Total area investigate in this study: 381.000 ha

We examined around half of the RSPO certified area (381 out of 719 sqkm). Around 10% of that area was flagged by World Ressources Institute as a loss of primary forest (37sqkm).

To conduct the investigation, we used both the Tree Cover Loss algorithm (that is widely applied in such instances, also by Global Forest Watch) and we complemented it with a visual examination of very high resolution satellite images from commercial providers (Maxar and Airbus) that are made available open source on Google Earth Pro.

The different steps of our investigation are being described further down. The outcome had been as follows: What had been marked by WRI as forest loss turned out to be in fact a harvested palm oil plantation. The reason for that lies in the capability of the satellite-based algorithm that is widely used to detect deforestation. The “tree cover loss” algorithm can be mistaken for deforestation of primary forest while it actually sometimes is only a normal harvest of a palm oil plantation.

This is because satellite data cannot always distinguish between different types of tree cover. For example, optical imagery, which is the most commonly used type of satellite data for monitoring deforestation, can only detect changes in the amount of greenness in an area. This means that it cannot distinguish between a forest that has been cleared and a plantation that has been harvested and very high resolution satellite images need to be used to investigate the causes of a drop in the tree cover.

Step 1: We filtered the RSPO database for the largest palm oil concessions in Malaysia.

According to RSPO, a total of 907 minor and major palm oil concession sites exist in Malaysia that are certified as RSPO. We filtered out those 13 RSPO-certified concessions where the largest alleged deforestation occurred in 2022 as per GFW.

Step 2: We uploaded each of the 13 largest RSPO concession areas on the Global Forest Watch (GFW) website and searched for tree cover loss occurring over those concessions during the year 2022.

We applied the same algorithm that GFW is also using and confirmed the same tree cover loss that GFW has also found. In fact, we found that in all of the 13 examined plantations, a tree cover loss occurred. In total the tree cover area lost in those concessions during the year 2022 was 3785 ha. In the next step, we had to understand if Tree Cover Loss incidences had actually been deforestation, as GFW had been suggesting.

Step 3: We randomly picked 8 concessions for further investigation: We wanted to understand the causes for tree cover loss in those concessions - was it due to deforestation or due to palm oil harvest.

We investigated all eight sites with high resolution satellite images to understand whether the tree cover loss had been due to a loss of primary forest or due to a palm oil plantation harvest situation. For all eight sites, we found that a palm oil plantation was already in place much earlier than 2022 and that hence the tree cover loss has been due to a palm oil plantation harvest and not due to deforestation. One example is illustrated below.

This figure shows Lepar Hilir 8 out of 13 concession areas. According to the analysis conducted by Vertify, there was no deforestation between 2021 and 2023. What GFW is labeling as deforestation is actually the harvesting of palm oil crops. The observed changes only indicate harvested palm oil in various locations, contrary to Global Forest Watch (GFW) claims that these locations are deforested.

Conclusion

We learned in this investigation that a single satellite layer (here: Tree Cover Loss) is not sufficient to explain the local phenomena of deforestation. Such a layer can give an indication about changes on the ground, but the causes of such changes need to be investigated further. Our investigated case is representative for many others and Global Forest Watch data is now carrying a clearer disclaimer that Tree Cover Loss ought not to be equalized with deforestation as their original communication had been suggesting.

As a result, a good due diligence, as requested by the EU regulation, should involve the use of multiple data sources to illustrate site-specific challenges. In our case, the use of a single satellite-based algorithm led to an over reporting of deforestation; but the choice of another satellite-based index or algorithm could have also led to the opposite.

Next to satellite data, site- or ecosystem-specific contextual data is also important. Actors at the first mile of the supply chain are best-placed to provide that contextual understanding. They might also be able to provide ground-truthed data, such as geo-tagged pictures taken by a phone camera.

By using multiple sources of satellite data alongside ground-truth data, it is possible to reduce the risk of misclassification, such as mistaking tree cover loss for deforestation.

The next innovation frontier for EUDR-compliant deforestation monitoring will be the operationalisation of site-specific monitoring tools. But tools that include ground data are generally more resource-intensive to deploy: Local communities have to be aggregated by a trusted partner on the ground and community members have to be trained to collect local data. They will also have to be remunerated for their work in data collection.

An important precondition for all of that is however that supply chain are sufficiently traceable so that a due diligence can be conducted in the first place. This will be the first challenge for companies operationalising the EU Regulation.