Benefits of high fidelity datasets for climate action

The future of climate projections... The EU's "Destination Earth" initiative created a Digital Twin (DT) of the Earth system. This DT is a kind of simulated "living" replica of the Earth system and was designed to provide more fine-grained climate projections. Powered by the first pre-exascale supercomputers in Europe, the Climate DT is able to provide climate impact data at scales of a few kilometres (current scale is around 100km, see image). Such local granularity matters, as climate change is a global but also very local phenomenon. The new Climate DT can bridge the gap between global (rather large-scale) climate projections and local climate impacts. Hopefully, this will support policy-making on climate adaptation and mitigation with a regional focus in mind. More info on the Climate DT: https://lnkd.in/dG3YV_kA Academic paper on Destination Earth: https://lnkd.in/dj_gjRNW #climatechange, #sustainability

We know A.I. and product-level data can be used to drive climate action. But what does that actually look like in practice? Here’s a great success story about Reckitt using CO2 AI to deliver on its scope 3 targets. Reckitt is a UK-based multinational that owns many household cleaning, health & hygiene brands including Dettol, Finish, Nurofen, Strepsils and Durex. The company has a science-based target to cut its scope 3 emissions by 50% by 2030 and achieve net zero by 2040. (They already achieved their target for operational emissions and are now fully focused on their supply chain). Working with CO2 AI and Quantis, Reckitt obtained more granular data on its 25,000+ products, improving data accuracy by 75%. That’s impressive. But the beauty of product-level emissions data is not about complete data or even choosing green products — it’s about companies being able to pinpoint emissions hotspots and opportunities for action. And that’s exactly what happened… They identified a potential 1.4Mt of additional CO2e reductions in their value chain, which had previously been overlooked due to a lack of granular data. That’s a 60% increase!! They were even able to drill down to the substance level, identifying the top 25 substances for raw materials and packaging that contribute most of their scope 3 emissions. This allows for very targeted supplier engagement and decarbonisation initiatives. I particularly love this quote: “The focus is shifting activity from measuring for reporting, to measuring for decision-making.” That’s exactly it. We need to focus less on reporting and compliance and more on making good, climate-informed business decisions. Thank you David Croft for your leadership on this! #ClimateAction #ProductLevelData #PACT

NVIDIA is one of the most successful companies today, leading the industry and propelling the development of the whole world. This year, the company set an example of how its cutting-edge technologies can mitigate the climate crisis. NVIDIA launched Earth-2, a digital clone of our climate system. This AI-powered tool can bring down the $140 billion cost of climate change-related natural disasters worldwide. Earth-2 is a cloud platform designed to predict weather and provide timely notification on extreme events and natural disasters. As these events become more and more frequent with climate change, Earth-2 can save many lives and enhance our understanding of the environment. The system runs various AI models to get high-resolution simulations. One of them is CorrDiff. It uses diffusion modelling to produce 12.5 times higher resolution images than current numerical models. In addition, using this system is 1,000 times faster and 3,000 times more energy efficient. The AI accelerates climate action and delivers warnings in seconds rather than minutes or hours. Another advantage of NVIDIA’s models is cost-effectiveness. Collecting high-resolution data and employing traditional simulation methods to achieve a kilometre scale require too many resources. AI allows information generation from lower-resolution input data and makes accurate and precise weather predictions. The Weather Company was among the first to adopt Earth-2 and is already using it for forecasts. In essence, Earth-2 is a set of tools available via cloud APIs. It is a full-stack open-source solution that allows any user to create emulations using its models. Startups and researchers worldwide can access Earth-2 data or use their datasets to train tailored versions of CorrDiff for specific use cases. It democratises access to climate information, promotes innovation, and goes hand in hand with NVIDIA’s other projects, such as the Inception program for startups. Earth-2 is used in Taiwan to predict the exact locations of typhoon landfall and evacuate people in time. Compared to traditional models with 25-km resolution, CorrDiff can work on a 2-km scale, helping to determine the intensity of storms and the location of the strongest winds. Currently, Taiwan loses $650M annually due to natural disasters. Accurate predictions can minimise casualties and economic losses. Using CorrDiff costs Taiwan’s government about $60,000 compared to nearly $3 million for the previous prediction system. In September, NVIDIA partnered with the UAE-based AI company G42 to work on precision and further develop the technology. They have established a new operational base and Climate Tech Lab in Abu Dhabi. The initial focus of the collaboration is building a square-kilometre-resolution weather forecasting model. In addition to saving lives and cutting losses, Earth-2’s accurate forecasts benefit the financial sector, allowing companies to improve risk assessment and asset management.