“Spatial Data - getting down to (Google) earth”

What finance and business needs is not more, but accurate, consistent, timely and scalable data, integrated through automation, tools and processes. Does last weeks’ launch of Dynamic World[1] by Google in partnership with WRI start to change the game?

Financial firms using spatial data is not a new idea. The likes of Refinitiv, Bloomberg and many other commercial data providers have been providing spatial data for traders and investors to track ships, weather patterns, crop and commodity production, oil and gas shipments for 10 years or more. Insurance companies have been using spatial and historical data to map physical (flood, fire) risks for many years as well. There are many examples of NGO and/or government backed tools and platforms, such as?ENCORE (Exploring Natural Capital Opportunities, Risks and Exposure) created by the Natural Capital Finance Alliance[2] or Global Forest Watch and Ocean Watch, both launched from within the WRI or more recent platforms such as Global Mangrove Watch.

So what is new? Firstly demand for better data is rapidly increasing and demand for broader and deeper data sets. Companies (and investors, financing firms) need to have a much better understanding of their operations, to locate and assess physical risk from climate change, and assess nature and bio-diversity risk as these issues also become more mainstream to a company and investors risk management and sustainability approach.?Companies need this information not just for their direct operations but also their ‘upstream’ suppliers, and downstream distribution and customer usage.

Spatial Data converts to Spatial Finance when geo-location data is combined with financial asset data to assess a company’s dependencies and impact on the surrounding environment which can impact its enterprise value. The challenge – it’s not just about knowing the state of the earth, it’s about knowing your dependencies and impacts on the ecosystem at each relevant location where business activity interfaces with the ecosystem services provided by nature to business. There are a number of challenges being solved today:

Know your ecosystems - Having a comprehensive and up to date asset location map that pinpoints a company’s assets to a precise location, attributes about the assets, and the characteristics of ecosystems that intersect with these assets. Many companies have or are building this information, but most companies are not disclosing this information today.

Common, standard measures - A common terminology and lack of standard metrics and measures so data can be aligned, compared and benchmarked – TNFD[3] , which I co-chair, is a group trying to solve this, for the CBD[4] (Convention on Biological diversity) developing the GBF (Global Biodiversity Framework). ?Its primary priority is developing a common analysis and disclosure framework, so that impacts on nature can be consistently measured and assessed against specific targets at a global, national and local level. In the race to develop tools and data solutions, providers are implicitly developing and promoting their own methodology for ecosystem condition assessment.?Lots of different tool proponents are claiming to have ‘the answer’ to assessing the state/condition of nature and risk assessment for companies, yet there is today no consistent science based universal consensus on how to measure eco-system status. The UN SEAA framework is arguably the closest but there is some debate as to its practicality as a global standard

Saleability and automation - Many of the spatial data tools available today provide natural system geo maps allowing ‘bottom up’ assessments, that provided detailed and transparent information on the state of ecosystem services and natural capital. Investors need to analyse this and make decisions; assessing impact from the ‘top down’. Having bottom up meet top down is currently not a repeatable or scalable process, and requires many hours of manual human expertise. Automation will change this. As one example of new automation, a relatively new company Down Force[5] , originally from an Oxford based trust, is using data sources to model natural capital and create a “digital twin” of an ecosystem area (water, earth, erosion, bio-diversity health) to measure impact in a digital version of the earth, more quickly and more repeatable than the manual assessments do

Upstream & downstream gaps - Upstream supply chain and downstream distribution is still very difficult because of poor information linking and availability of supplier asset and operational data or standard metrics and disclosures. TRASE[6] (backed by a partnership between the Stockholm Environment Institute and Global Canopy) is one initiative, which uses publicly available customs and trade data to map sources of commodities such as wood pulp, soy, palm oil to environmental sensitivities and issues

Temporal consistency – Reliable time series datasets are key for observing and understanding environmental trends and patterns over time and can serve as the basis to forecast into the future. Temporal databases store data relating to time instances. They offer temporal data types and store information relating to past, present and future time. However many data sources are from ‘one off’ projects and studies, and do not provide satisfactory historical or in time monitoring and tracking. Google’s Dynamic World application, developed with WRI, uses AI to map 9 land use and cover types at 10m resolution in near real-time using Copernicus Sentinel-2 satellite imagery which takes daily global snapshots.?This is a step forward, others will undoubtedly use similar sources and measurement frequencies as the number of satellites and monitoring systems increase.

Accuracy of data – Understanding the accuracy and limitations of data is challenging and with many companies using the same data sets, can lead to systemic errors with multiple use of direct and indirect proxies. On the ground verification can enable estimates of accuracy and verify data but it is expensive to take physical measurements and difficult to scale. Recent advances in bioacoustics technology, including the development of autonomous cabled and wireless recording arrays, permit data collection at multiple locations over time. Natural State[7] is an example of an initiative using machine learning and microphone arrays to detect the level of species and animal activity in parts of Kenya

Measuring bio-diversity – it is well known that biodiversity itself is very hard to measure and quantify, but new tools are at least making this easier. ?The Integrated Biodiversity Assessment Tool (IBAT) is a tool allowing companies to evaluate whether their assets intersect with protected areas, key areas of high biodiversity, and red listed species. This allows companies to screen for risk of impacts on biodiversity The ICUN STAR[8] , which is now integrated into IBAT, goes one step further, providing?a static map of species bio-diversity index allowing companies to understand how they are contributing to biodiversity impacts, either positively or negatively. ?eDNA (Environmental DNA) offers the promise of accurate, inexpensive and rapid field-based measures of the biodiversity health in an area through eDNA barcoding. Whilst reliant on physical measurement on the ground, eDNA provides for bottom-up assessment and verification of higher scale spatial datasets. This is something being explored by Nature-Metrics[9]

Data confidence – data confidence and quality can be an issue. Many data sets are now quite stale, maybe 10 or even 20 years old. There is a common need for a confidence or quality stamp, as quality assessments today are arbitrarily made. As intimated earlier and outlined in the TNFD Proposed Technical Scope (TNFD 2021), there are keys areas of scope which must be assessed for decision-grade datasets, namely relevance, resolution and scalability, temporality, frequency of update, geographic coverage, accessibility, comparability, thematic coverage and authoritativeness.

In summary there is much to be excited about as spatial data comes down to earth, and the Google announcement of Dynamic World, whilst not answering all of the challenges, is bound to fuel interest and excitement as well as further innovation. New providers, new tools and analytics, are being created at almost the same speed as new low orbit satellites (1200 in 2020!) are being launched.

For those members of the TNFD and other companies and financial institutions now embarking on the nature risk assessment journey, this is of course great news. Not having enough data to manage nature-related risk is an argument that is hard to make, with more data from more providers than ever before. Standards and confidence in data quality, will become critical, and this is one of the goals of the TNFD Data Catalyst. We aim to help accelerate the development of better standardised data, analytics and workflow tools for measuring, assessing and reporting on nature related risks and opportunities.

As more companies and financial investors put time and energy into this space, I predict the demand for scale will lead eventually to a consolidation of data providers leading solutions which scale both geographically and across ecosystems. There will be more automation of company and invested assets to locations, dependency, impact and risk assessment. The scientists will argue automation can remove scientific rigour, but the technologist will argue the challenge is having both. If we are serious about re-directing financial flows to nature positive activities, we certainly need both.

The authors:

David Craig ?is co-chair of the TNFD alongside Elizabeth Mrema, and previously the CEO of Refinitiv.

Dan O’Brien is a Sustainability and Climate Change partner at PwC, and leads the TNFD Data Working Group

The TNFD is updating the Data Landscape Discussion paper[10] which will be available following the launch of the TNFD Data Catalyst Programme

[1] Dynamic World - https://blog.google/products/earth/dynamic-world-land-cover-data/

[2] The NCFA Secretariat is run by the UNEP Finance Initiative, Global Canopy and UNEP WCMC

[2] www.tnfd.global

[4] https://www.cbd.int/

[5] https://downforce.tech/

[6] TRASE https://resources.trase.earth/documents/Trase_supply_chain_mapping_manual.pdf

[7] Natural State, Kenya https://www.naturalstate.org/

[8] https://www.iucn.org/regions/washington-dc-office/our-work/species-threat-abatement-and-recovery-star-metric

[9] https://www.naturemetrics.co.uk/

[10] https://tnfd.global/publication/data-discussion-paper/