This image is astonishing for both the message it communicates and for its sci comm value. Incredible work by team Copernicus EU It shows global ocean temps for each year since 1940. The alarming, yet brilliant part is the pink shading which highlights the 2023-2024 jump 1/

This is live! Please tune in to this event with Executive Vice-President Margrethe Vestager now talking #DestinationEarth youtube.com/live/2gULcd_9y…

Margrethe Vestager sounding the bell for the launch of Destination Earth Margrethe Vestager European Commission Digital EU 🇪🇺 #DestinE youtube.com/watch?v=2gULcd…

🌍🌐 Today we celebrated a major milestone in #DestinE at the LUMI supercomputer EuroHPC supercomputer center in Finland with @Vestager. The launch kicked off the second phase of DestinE, the ambitious initiative to create a digital replica of our planet. ➡️ destine.ecmwf.int/news/destinati…

We often focus on global temperatures, but land regions where we all live have been warming much faster. For #showyourstripes day, here are land, ocean, and global climate stripes shown on the same color scale:

AI weather prediction models can perform global downscaling of low-resolution climate model data. Details in our new preprint: arxiv.org/abs/2406.17977 The image shows 2m temperature data from a 250km resolution climate model, downscaled using Pangu-Weather.

🔍We have taken #CMIP6 model output (left) and enhanced its resolution globally using an #AI weather prediction model (right). 🔎 More details in our preprint, led by Nikolay Koldunov : doi.org/10.48550/arXiv… ECMWF The animation shows 2-metre temperature in 2010 #downscaling

Blame Dr. Mathew Barlow for inspiring this cursed video 🥸 #python #ducks #ScienceVisualization

In Christian Lessig 's words, a well-trained weather forecasting model that has learned the behavior of the Earth system can make a big difference for a range of applications. Such general model can potentially give better results than a smaller model just trained for downscaling

Tantilizing new work led by Nikolay Koldunov. It shows that using a well-trained weather forecasting model--which has learned the general dynamics and behaviour of the Earth system!--can make a big difference for a wide range of application, in our case climate downscaling.

Emerging AI-based weather prediction models as downscaling tools ⚒️

Thomas Rackow 🧊 Christian Lessig I really believe this is the way & the 2 approaches have merits: 1️⃣ An IPCC trained downscaling (+bias-correct) using #AI has merits of using a large amount of existing readily available data. 2️⃣ A well trained #AIFS (foundation?) model natively at Km-scale is so cool 😎👍

Really cool results - i never cease to be amazing how #ML models break new ground every day, accelerating the weather & climate revolution at an unprecedented pace :) Stay tuned for more interesting work in this space ECMWF with our partners, also in the framework of #DestinE

Still proud to be part of this fantastic project, a flagship initiative of the European Commission #DestinationEarth