Understanding Corporate impacts through supply chain transparency
In line with the global effort to reduce the impact of climate change and achieve net-zero targets, there is a growing need for companies to provide more transparent information around their business decision-making. Here at Space Intelligence, we are developing in-house products leveraging earth observation data to monitor companies impacts on environmental metrics like biodiversity and agricultural emissions, to provide insight on ESG performance.
ESG, Environmental, Social and Corporate Governance, encompasses the three broad categories or standards that are commonly used when assessing the sustainability of a potential investment or company. This information is very useful for investors who want to ensure they are investing responsibly in companies whose values and concerns align with their own metrics, such as sustainability and environmental impact. With companies improving their transparency, we can begin to offer company and regional level performance monitoring for environmental metrics, allowing for better informed sustainable investing.
In 2018, Unilever released the locations of its 1431 mills used for palm oil production. This location information is very useful for scientists and investors, as mill locations can be compared to other data in order to better understand landscape and biodiversity change, driven by palm oil production. At this point, Space Intelligence was hosting a NERC-sponsored scientist, Dr Penny How, who led an analysis comparing the locations of these mills to forest cover loss, with the objective of examining the impact of palm oil production on tropical forests. We’re re-publishing Penny’s blog 2018 report of the analysis now, to illustrate how we can provide environmental performance insights at scale.
We’re now working with companies around the world seeking to understand corporate impacts, and if you are interested in obtaining similar insights, please contact [email protected].
Global forest change
The map below shows the University of Maryland forest data. The map shows forest cover change between 2000-2014, with any change in forest cover (0-100%) marked as black on the map. Regions of large-scale forest change can be seen across South America, the south-east of North America, central Africa, Indonesia, Malaysia, China, Scandinavia, Russia, and Canada. However the processes driving change in each region are different: Scandinavian forests being subject to long term forest management for timber; whilst tropical forests continue to be cleared and degraded in order to be able to plant crops like soya (predominantly south America); Cocoa (predominantly west Africa); and Oil palm (predominantly Malaysia and Indonesia).
Space Intelligence have been using free, open-source data to visualise the impact of palm oil production on deforestation. We are fortunate to have this data available, since Unilever recently released mill locations across the world. We compare this to data on forest change hosted on Google Earth Engine. These maps of forest loss are created by scientists at the University of Maryland, using Landsat satellite imagery (Hansen et al., 2003).
Unilever’s palm oil mills are spread across the tropcs, as shown by the green points on the map. We previously commented about the available information on Unilever’s palm oil mills, with by far the highest number of mills in Indonesia and Malaysia (1275 mills) followed by Central and South America (99 mills). These two areas are highlighted, with Map A showing Central and South America, and map B showing Indonesia and Malaysia. At this scale, we can see forest change and individual mill locations more clearly. However, it is easier to observe impact when we focus on even smaller areas.
The maps above show forest change and mill locations on the island of Sumatra, Indonesia and on peninsular Malaysia. The finer scale of these maps provide more detail about forest change, with the colour ramp indicating the proportion of forest loss between 2000 and 2014. Blue areas have experienced little loss in forest cover (>1-15%), whilst the red areas have experienced substantial losses (75-100%). White areas indicate that the landscape has experienced little/no forest loss (<1%).
Here, we have focused on three particular regions in Indonesia and Malaysia. Map A and Map C are in Sumatra, outside two of the main cities – Dumai (Map A) and Palembang (Map C). Map B is an area of Penang, which is one of the district states within Malaysia.
There are regions of high percentage forest loss around Dumai and Palembang, as indicated by the widespread red patches in Map A and Map C. Forest loss is lower in Penang (Map B), with fewer red patches and many more white regions.
We also looked in finer detail at forest loss and palm oil mills in the Central and South America area, as shown on the map above, with particular interest in regions of Honduras (Map A), Ecuador (Map B) and Columbia (Map C). The three regions selected here have experienced forest loss to some extent (within the 0-15% category). For example, Map B shows a region of Ecuador near to the capital Quito, which has experienced large areas of forest loss of <30% between 2000 and 2014. This is nowhere near the scale of forest loss in Sumatra by comparison, which is clearly visible based on the number and extent of red patches on the maps.
Regions of high percentage forest loss are very small in Central and South America. These red regions tend to be located in close proximity to selected palm oil mills, such as in Map A (on the coast of Honduras) and Map C (in Columbia). It is difficult to link these regions of high-percentage forest loss to the presence of Unilever’s palm oil mills directly, however patches of intense forest loss do appear to be spatially co-located with the mills.
So is forest loss correlated with palm oil mills?
It is hard to attribute landscape change directly to the presence of Unilever’s palm oil mills given that there are many drivers of forest loss. This is a complex issue that cannot be encapsulated as a single cause-and-effect relationship. It is likely that palm oil production by Unilever has influenced the landscape to some extent, but it is challenging to quantify this impact without a more sophisticated analysis.
What is the value to forests of RSPO certification?
We can examine the relationship between mill location and mill RSPO certification status, and forest loss, by extracting the forest loss rates from the area near the mills. This is not unreasonable to do since the palm oil kernels need to reach the mills relatively quickly (within 24 hours) to be processed, so we assume that forest loss near the mills is associated with that mill’s activity. To do this we aggregated the Hansen forest change map to 3 km x 3 km pixels, and extracted the forest loss rates for each pixel where a mill was located.
Globally, the 3 km x 3 km pixel in which a mill was located experienced a mean forest loss of 26.3 % (median 19.0%). RSPO certified mills had a mean loss rate of 24.6 % (median 18 %) ; whereas non-RSPO mills had 26.7 % mean forest loss (median 19%). We are not controlling for any other variables, which we would need for a full impact assessment. However, on this statistically naive comparison, there appears to be a difference in forest loss rates between RSPO and non-RSPO mills. As more mill data becomes available, it may become a viable research project / paper to test the impact of certification on forest loss rates in a fully robust manner.
Unilever’s transparency in its supply chains is a laudable start to improving sustainability in oil palm production, and more companies with forestry activities in this region need to follow suit in order to achieve this.