Advancements In Technology

🌦️ GenAI in Weather Forecasting: Decoding Unseen Patterns 🌦️ Imagine a world where weather predictions are so accurate, they can anticipate even the most subtle changes in the atmosphere. This is not science fiction—it's the power of Generative AI (GenAI) in weather forecasting. Why GenAI? 1. Decoding Satellite Images: Traditional weather forecasting relies heavily on interpreting satellite images. GenAI can process these images with unparalleled precision, identifying patterns and anomalies that human eyes might miss. 2. Unseen Patterns: The true strength of GenAI lies in its ability to detect unseen patterns in vast datasets. By analyzing historical and real-time data, it can predict weather events with greater accuracy. How Does It Work? - Data Processing: GenAI processes massive amounts of data from satellites, sensors, and historical records. - Pattern Recognition: It uses advanced algorithms to recognize patterns that indicate specific weather conditions. - Predictive Modeling: The AI generates predictive models that can forecast weather events with higher precision than ever before. The Impact 🌪️ Disaster Preparedness: More accurate predictions mean better preparation for natural disasters, potentially saving lives and reducing economic losses. 🚜 Agricultural Benefits: Farmers can make more informed decisions about planting and harvesting, leading to better yields and more sustainable practices. ✈️ Aviation Safety: Improved forecasts can enhance flight safety and efficiency, reducing delays and optimizing routes. The Future The integration of GenAI in weather forecasting is just the beginning. As technology evolves, we can expect even more refined and accurate predictions, leading to a safer and more efficient world. 🔍 Curious about the future of weather forecasting with GenAI? Let's explore it together! P.S. Have you experienced the benefits of advanced weather forecasting in your field? Share your story below! 🌍

So excited to share our new Nature paper on GenCast, an ML-based probabilistic weather forecasting model: https://lnkd.in/enzPUFbn It represents a substantial step forward in how we predict weather and assess the risk of extreme events. 🌪️ GenCast uses diffusion to generate multiple 15-day forecast trajectories for the atmosphere. It assigns more accurate probabilities to possible weather scenarios than the SoTA physics-based ensemble system from ECMWF, across a 2019 evaluation period. It’s vital that we ensure these new ML weather systems are safe and reliable. One thing I'm proud of is our range of evaluation experiments: per-grid-cell skill & calibration, spatial structure, renewable energy, extreme cold/heat/wind, and the paths of tropical cyclones (i.e. hurricanes). For example, we created a dataset of simulated wind power data at wind farm sites across the globe, and found that GenCast outperforms ENS by 10–20% up to 4 days ahead. This is promising, because better weather forecasts can reduce renewable energy uncertainty and accelerate decarbonisation. We also compared cyclone tracks from GenCast and ENS with ~100 cyclones observed in 2019. GenCast's ensemble mean cyclone track has a 12-hour position error advantage over ENS out to 4 days, and more actionable track probability fields out to 7 days. Cyclone maximum wind speeds are still generally underestimated (a common problem for ML weather models), but this performance on tracks is really promising. One recent devastating cyclone was Hurricane Milton, which caused >$85 billion in damages. GenCast predicted ~70% probability of landfall in Florida 8.5 days before the hurricane struck (and ~2 days before it even formed). A GenCast ensemble member takes 8 minutes on a TPU chip, versus hours on a supercomputer for physics-based models. This opens up the possibility of large ensembles (eg 1000s of members) which could better estimate risks of extreme events. We don't yet know how much value this will yield over conventional ensemble sizes (~50 members). Like its predecessor (GraphCast), the weights & code of GenCast have been made publicly available: https://lnkd.in/eg78dd7T. We’re looking forward to seeing how the community builds on this! It's been an honour to work on this study led by Ilan Price with such a talented team ✨: Alvaro Sanchez Gonzalez, Ferran Alet Puig, Andrew El-Kadi, Dominic Masters, Timo Ewalds, Jacklynn Stott, Shakir Mohamed, Peter Battaglia, Rémi Lam, & Matthew Willson

Satellites generate more data in an hour than we can download in a day. Here's why that's about to change. Modern satellites collect an overwhelming amount of information - far more than we can transmit back to Earth quickly. But this isn't just a technical problem. It's potentially costing lives. Here's what's happening right now: When wildfires threaten homes: ↳ Satellite images showing their spread sit trapped for hours During hurricane season: ↳ Vital storm trajectory data reaches emergency teams late - when every minute counts Military operations rely on several-hour-old satellite intelligence ↳ In situations where seconds matter Think about that: We have the data to: • Protect lives • Mitigate disasters • Optimize operations But much of it's stuck in space, waiting to be downloaded. This is why AI-powered satellites are transforming space operations. Take the European Space Agency's new Φsat-2 satellite. Instead of blindly collecting and slowly transmitting back to Earth, it: • Processes images in orbit • Identifies what's actually important • Only sends down actionable intelligence The early indications are game-changing: • 80% reduction in transmission needs • Real-time disaster monitoring • Faster threat detection • Rapid weather pattern analysis Of course, AI in space faces challenges: → Cybersecurity risks → Regulatory constraints → Complex international coordination But the potential rewards are immense for those focusing on: • Reducing data transmission bottlenecks • Providing real-time, actionable insights • Solving critical infrastructure and monitoring challenges This goes beyond a “tech upgrade”. It's a powerful transformation in how we protect communities, save lives, and understand our planet. The old approach: Collect everything, transmit slowly, analyze later. The emerging reality: Think in orbit, send what matters, act immediately. Earth’s early warning systems are getting smarter. P.S: Join high-growth founders and seasoned investors getting deeper analysis on emerging tech trends and opportunities on my newsletter (https://lnkd.in/e6tjqP7y) ____________________________ Hi, I’m Richard Stroupe, a 3x Entrepreneur, and Venture Capital Investor I help early-stage tech founders turn their startups into VC magnets Building in space tech? Let's talk

You might have seen news from our Google DeepMind colleagues lately on GenCast, which is changing the game of weather forecasting by building state-of-the-art weather models using AI. Some of our teams started to wonder – can we apply similar techniques to the notoriously compute-intensive challenge of climate modeling? General circulation models (GCMs) are a critical part of climate modeling, focused on the physical aspects of the climate system, such as temperature, pressure, wind, and ocean currents. Traditional GCMs, while powerful, can struggle with precipitation – and our teams wanted to see if AI could help. Our team released a paper and data on our AI-based GCM, building on our Nature paper from last year - specifically, now predicting precipitation with greater accuracy than prior state of the art. The new paper on NeuralGCM introduces 𝗺𝗼𝗱𝗲𝗹𝘀 𝘁𝗵𝗮𝘁 𝗹𝗲𝗮𝗿𝗻 𝗳𝗿𝗼𝗺 𝘀𝗮𝘁𝗲𝗹𝗹𝗶𝘁𝗲 𝗱𝗮𝘁𝗮 𝘁𝗼 𝗽𝗿𝗼𝗱𝘂𝗰𝗲 𝗺𝗼𝗿𝗲 𝗿𝗲𝗮𝗹𝗶𝘀𝘁𝗶𝗰 𝗿𝗮𝗶𝗻 𝗽𝗿𝗲𝗱𝗶𝗰𝘁𝗶𝗼𝗻𝘀. Kudos to Janni Yuval, Ian Langmore, Dmitrii Kochkov, and Stephan Hoyer! Here's why this is a big deal: 𝗟𝗲𝘀𝘀 𝗕𝗶𝗮𝘀, 𝗠𝗼𝗿𝗲 𝗔𝗰𝗰𝘂𝗿𝗮𝗰𝘆: These new models have less bias, meaning they align more closely with actual observations – and we see this both for forecasts up to 15 days, and also for 20-year projections (in which sea surface temperatures and sea ice were fixed at historical values, since we don’t yet have an ocean model). NeuralGCM forecasts are especially performant around extremes, which are especially important in understanding climate anomalies, and can predict rain patterns throughout the day with better precision. 𝗖𝗼𝗺𝗯𝗶𝗻𝗶𝗻𝗴 𝗔𝗜, 𝗦𝗮𝘁𝗲𝗹𝗹𝗶𝘁𝗲 𝗜𝗺𝗮𝗴𝗲𝗿𝘆, 𝗮𝗻𝗱 𝗣𝗵𝘆𝘀𝗶𝗰𝘀: The model combines a learned physics model with a dynamic differentiable core to leverage both physics and AI methods, with the model trained directly on satellite-based precipitation observations. 𝗢𝗽𝗲𝗻 𝗔𝗰𝗰𝗲𝘀𝘀 𝗳𝗼𝗿 𝗘𝘃𝗲𝗿𝘆𝗼𝗻𝗲! This is perhaps the most exciting news! The team has made their pre-trained NeuralGCM model checkpoints (including their awesome new precipitation models) available under a CC BY-SA 4.0 license. Anyone can use and build upon this cutting-edge technology! https://lnkd.in/gfmAx_Ju 𝗪𝗵𝘆 𝗧𝗵𝗶𝘀 𝗠𝗮𝘁𝘁𝗲𝗿𝘀: Accurate predictions of precipitation are crucial for everything from water resource management and flood mitigation to understanding the impacts of climate change on agriculture and ecosystems. Check out the paper to learn more:  https://lnkd.in/geqaNTRP

AI Revolutionizes Weather Forecasting: A New Era of Accuracy AI is transforming weather prediction, with tools like Google DeepMind’s GraphCast leading the way. GraphCast delivers 10-day weather forecasts with up to 90% accuracy, processing data in under a minute—a task that traditionally takes hours. By analyzing 39 years of historical weather data, GraphCast has already outperformed conventional models, accurately predicting extreme events like Hurricane Lee’s landfall three days earlier than other forecasts. However, despite its impressive capabilities, AI in weather forecasting still requires human oversight. Meteorologists play a crucial role in interpreting AI-generated data, ensuring the predictions are accurate and actionable. This collaboration between AI and human expertise enhances disaster preparedness and decision-making across industries, marking a turning point in meteorology. #AI #WeatherForecasting #TechInnovation #ClimateTech #DeepLearning #Meteorology #HumanAICollaboration

GenCast is out in Nature Magazine! It's the first high res ML ensemble weather forecast which outperforms the operational state of the art. And a few more things have happened since the preprint was first released ⬇️ ⬇️ ⬇️ The paper shows GenCast provided better probabilistic weather forecasts, including better forecast of extreme weather, than the operational gold standard over our year-long evaluation period. This could mean earlier preparation for extreme events, more reliable wind power, and much more. Alongside publication in Nature Magazine, we are making the code and model weights available to the community (incl. a mini version of the model which gives passable results and can run in a free colab). And soon we'll share an archive of model historical and current forecasts. We also recently fine tuned the model to work on operational inputs so that it can be run live. We conducted a retrospective analysis of this model's forecasts of the track Hurricane Milton 🌪️ . GenCast predicted 60-80% probability of landfall in Florida already from 8.5 days before landfall eventually happened - a couple of days before Milton even formed - and more than 90% from 5.75 days before. *caveats on this example* a) individual examples should always be taken with a pinch of salt - we need rigorous evaluation over an extended period, see the cyclone track section of the paper. b) these are track predictions, not intensity predictions. Overall, GenCast marks something of an inflection point in the advance of AI for weather prediction, with SOTA raw forecasts now coming from AI. I think we can expect them to be increasingly incorporated operationally alongside traditional models (and to continue to improve!) Check out the paper: https://lnkd.in/dsasGbNb The code: https://lnkd.in/dHSjfW-3 And the blog: https://shorturl.at/NPvwL Work with a remarkable team: Alvaro Sanchez Gonzalez, Ferran Alet Puig, Tom Andersson, Andrew El-Kadi, Dominic Masters, Timo Ewalds, Jacklynn Stott, Shakir Mohamed, Peter Battaglia, Rémi Lam, & Matthew Willson

Most AI weather models are trained once and then left to run, making predictions based on what they’ve seen before. But the atmosphere doesn’t work that way. It’s constantly evolving, with each weather event being uniquely different from anything in historical data. The future of forecasting isn’t just about accuracy at initialization. It’s about reinforcement learning—AI that continuously validates itself, adapts in real time, and improves with every new observation. At Tomorrow.io, we’re enabling this shift with real-time, high-resolution satellite observations that feed directly into AI models, continuously validating predictions as events unfold and learning to better represent the physical processes driving extreme weather. Because the best AI forecast isn’t just the most accurate today. It’s the one that learns and improves with every observation to better predict tomorrow.

Why is MTG-S1 a nowcasting game-changer? 1. New data: until now, Europe's fleet of geostationary weather satellites has relied exclusively on imagers. The Infrared Sounder on board MTG-S1 introduces an entirely new capability by measuring infrared spectra in each of its pixels. Using this information, scientists can detect atmospheric properties such as temperature and humidity vertically as well as horizontally. The technology used – called imaging Fourier-Transform Spectrometry – detects the unique ‘fingerprints’ created on infrared light waves when gases in the atmosphere emit or absorb infrared light. 2. Faster warnings for severe weather: MTG-S1’s ability to revisit Europe every 30 minutes and provide a vertical profile of temperature and moisture means forecasters get near real-time updates on atmospheric conditions. The Infrared Sounder can be used to identify vertical air movements, including temperature inversions, where a warm layer traps cooler air below. Inversions can suppress storm activity until the inversion collapses, releasing energy in the form of heavy downpours or hail. Without vertical data, inversions are invisible to traditional imagers – but they become clear using MTG-S1’s hyperspectral sounder. 3. Making skies safer for air traffic: turbulence is a perennial hazard for air traffic – causing both discomfort and danger for pilots and passengers. Sometimes the cause is visible, but sometime turbulence occurs even during clear weather. MTG-S1’s vertical profiling of temperature and humidity, combined with wind estimates, allow meteorologists to pinpoint these hidden hazards. 4. Supporting climate and the environment: meteosat satellites have been providing large-scale weather datasets since 1977. Now MTG-S1 will add infrared sounding data to this body of information, enriching the climate record. 5. Designed for the future: MTG-S1 will set a new standard as its Infrared Sounder is among the most complex and powerful hyperspectral sounders ever built for space.

In this week's column, I look at NVIDIA's new generative foundation model that it says enables simulations of Earth’s global climate with an unprecedented level of resolution. As is so often the case with powerful new technology, however, the question is what else humans will do with it. The company expects that climate researchers will build on top of its new AI-powered model to make climate predictions that focus on five-kilometer areas. Previous leading-edge global climate models typically don’t drill below 25 to 100 kilometers. Researchers using the new model may be able to predict conditions decades into the future with a new level of precision, providing information that could help efforts to mitigate climate change or its effects. A 5-kilometer resolution may help capture vertical movements of air in the lower atmosphere that can lead to certain kinds of thunderstorms, for example, and that might be missed with other models. And to the extent that high-resolution near-term forecasts are more accurate, the accuracy of longer-term climate forecasts will improve in turn, because the accuracy of such predictions compounds over time. The model, branded by Nvidia as cBottle for “Climate in a Bottle,” compresses the scale of Earth observation data 3,000 times and transforms it into ultra-high-resolution, queryable and interactive climate simulations, according to Dion Harris, senior director of high-performance computing and AI factory solutions at Nvidia. It was trained on high-resolution physical climate simulations and estimates of observed atmospheric states over the past 50 years. It will take years, of course, to know just how accurate the model’s long-term predictions turn out to be. The The Alan Turing Institute of AI and the Max Planck Institute of Meteorology, are actively exploring the new model, Nvidia said Tuesday at the ISC 2025 computing conference in Hamburg. Bjorn Stevens, director of the Planck Institute, said it “represents a transformative leap in our ability to understand, predict and adapt to the world around us.” The Earth-2 platform is in various states of deployment at weather agencies from NOAA: National Oceanic & Atmospheric Administration in the U.S. to G42, an Abu Dhabi-based holding company focused on AI, and the National Science and Technology Center for Disaster Reduction in Taiwan. Spire Global, a provider of data analytics in areas such as climate and global security, has used Earth-2 to help improve its weather forecasts by three orders of magnitude with regards to speed and cost over the last three or four years, according to Peter Platzer, co-founder and executive chairman.

The first generation of AI weather models is now approaching maturity and utility, thanks to European Centre for Medium-Range Weather Forecasts - ECMWF and its release of the #opensource AIFS model, which contains many more useful weather parameters, updates faster, is mostly quite accurate, and can be run by users on their own small hardware. This first generation is slightly improving on something that physics-based models already did very well at - making a global gridded forecast at medium resolution (~10-50km grid; 3-6hr steps), which updates a few times per day. An amazing achievement in just two years to nearly replicate physics-based models that were refined for 5 decades. Some researchers and companies will continue to iterate on this first generation of models, until they get to full feature parity and resolution with the physics-based models. A better mousetrap. Others will push the envelope, and take #ai weather model development to the places where the physics-based models really struggled: ⏳Much faster updates: From 1/3/6/12 hourly updates, to updating every 5/10/15 minutes. 🛰️Data assimilation: Actually using all the observational data, rather than only the data that the physics-based models can work with. 📐Finer resolution: Being able to resolve the weather that matters, which can be as small as a tornado, going down to 10 metre to 1 km scales There is already some great early progress on these avenues. What does this mean for the consumers of weather data? I would say to push your data and software suppliers hard on what they are doing to leverage these new model capabilities in the products they make. I would also say for #datascience teams, be wary of “let’s use one of these open source models and do it ourselves” - since at best you end up with just one of the raw ingredients that your supplier(s) use. What does this mean for the #weather enterprise? Distribution models and business models and moats will change very quickly. Value will accrete to primary data collection/archival and management, and to end use cases, rather than the middle of the stack. At the very least you should be deeply thinking and openly debating what it means for you, and making investments. Stephan Rasp Daniel Rothenberg Ryan Keisler Bas Steunebrink Dr. Jesper Dramsch