AI for accurate weather prediction in rural areas

Thrilled to unveil our latest work: multi-modal machine learning to forecast localized weather! We construct a graph neural network to learn dynamics at point locations, where typical gridded forecasts miss significant variation. Paper: https://lnkd.in/eBmfsJin Weather dataset: https://lnkd.in/ejCG8bKs Code: https://lnkd.in/eQg-JzQJ AI weather models have made huge strides, but most still emulate products like ERA5, which struggle to capture near-surface wind dynamics. The correlation between ERA5 and ground weather station data is low due to topography, buildings, vegetation, and other local factors. In this work, we forecast near-surface wind at localized off-grid locations using a message-passing graph neural network ("MPNN"). The graph is heterogeneous, integrating both global forecasts (ERA5) and historical local weather station data as different nodes. What do we find? First off, ERA5 interpolation performs poorly, failing to capture local wind variations, especially in coastal and inland regions with complex conditions. An MLP trained on historical data at a location performs better than ERA5 interpolation, as it learns from the station's past observations. However, it struggles with longer lead times and lacks the spatial context necessary to capture weather patterns. Meanwhile, our MPNN dramatically improves performance, reducing the error by over 50% compared to the MLP. This is because the MPNN incorporates spatial information through message passing, allowing it to learn local weather dynamics from both station data and global forecasts. Interestingly, adding ERA5 data to the MLP does not improve its performance significantly. The MLP struggles to integrate spatial information from global forecasts, while the MPNN excels, highlighting the importance of combining global and local data. Large improvements in forecast accuracy occur at both coastal and inland locations. Our model shows a 92% reduction in MSE relative to ERA5 interpolation overall. This work showcases the strength of machine learning in combining multi-modal data. By using a graph to integrate global and local weather data, we were able to generate much more accurate localized weather forecasts! Congrats to Qidong Yang and Jonathan Giezendanner for the great work, and thanks to Campbell Watson, Daniel Salles Chevitarese, Johannes Jakubik, Eric Schmitt, Anirban C., Jeremy Vila, Detlef Hohl, and Chris Hill for a wonderful collaboration. Thanks also to our partners at Amazon Web Services (AWS) for providing cloud computing and technical support!

Announced today in the WSJ!  Our AI-based long-range weather forecasting model, NeuralGCM is helping 38 million farmers in India, enabling them to make more informed decisions about when to plant their crops for the season. Hundreds of millions of smallholder farmers across the tropics depend on information about when the monsoon season will come each year. However, accurate forecasting of when the monsoon will begin, especially at long lead times and at local scales, has remained a century-old challenge. When we open-sourced NeuralGCM, we hoped the community would use this new model to power their own innovative applications. The University of Chicago, in collaboration with India's Ministry of Agriculture and Farmers’ Welfare, did just that – they used our NeuralGCM model in combination with models from ECMWF to text forecasts directly to farmers each week. Here's how this can help farmers: -- Beyond traditional weather models: For years, accurate long-range monsoon forecasting has been a challenge. AI-driven models like NeuralGCM learn from decades of historical weather data, making them more efficient to run. -- Tangible economic and social impact: Existing research from the University of Chicago shows that accurate monsoon forecasts can almost double a farmer's annual income. By putting actionable information directly into the hands of tens of millions, this initiative is helping farmers strengthen their resilience against climate variability. This project demonstrates the immense potential of AI to create solutions that directly benefit communities. It’s a huge achievement by the Indian Ministry of Agriculture and a model for how to put people first in the age of AI. I am so proud and thankful to our team – Olivia Graham, Stephan Hoyer, Shreya Agrawal, Mansi Kansal–  whose passion and commitment has kept us focused on the impact we can drive in agriculture.  Yet another testament to the value that AI can bring to our earth. Read the full story: WSJ: https://lnkd.in/ge9ujZTt Google blog: https://lnkd.in/gDqsPYvS

Other companies talk about their “new AI model.” Tomorrow.io's has been AI operational, globally, for years, turning cutting-edge technology into real-world insights for millions. And the results speak for themselves - in partnership with One Acre Fund and TomorrowNow.org, Tomorrow.io delivered hyperlocal weather forecasts via SMS to farmers in Kenya. The impact: - 12% higher yields on average with the best delivery model - 88% adoption and trust in the service - Among the most cost-effective interventions One Acre Fund has ever measured Farmers like Jeanne Mushimiyimana, pictured here, are turning insight into impact with Tomorrow.io’s Resilience Platform™, protecting their livelihoods and building climate resilience where it’s needed most. This is resilience in action - a proven, scalable model ready to reach millions more farmers across Africa and beyond. Proud of the teams that made this possible and even more excited for the impact ahead. Learn more: https://okt.to/7xKORr #AIForGood #WeatherIntelligence #Resilience

Google just released it's AI-powered weather Nowcasting on Search in Africa, 🌍 🎉 reshaping the future of weather forecasting in Africa! 🌦️ For years, many parts of Africa have faced significant challenges due to a lack of reliable weather data, with the continent having only a fraction of the radar facilities available in other regions. This data gap has made it difficult for people to make informed decisions about everything from farming to daily activities. 🌾 Now, thanks to Google Research’s MetNet AI model, short-term precipitation forecasts are becoming a reality for millions across Africa. By using satellite data and advanced AI, Google is able to deliver highly accurate weather predictions every 15 minutes, even in areas with limited infrastructure. This breakthrough is a game-changer for farmers, coastal communities, and anyone who relies on accurate weather information to plan their day or protect their livelihood. 🛰️🌧️ With this new technology, Google is helping to bridge the gap in weather data, empowering local communities to better prepare for extreme weather events like floods and droughts. The result? A more resilient continent better equipped to tackle the challenges of a changing climate. ☀️ Google’s commitment to AI innovation is helping Africa not only improve weather forecasting but also build long-term resilience, making this advancement crucial for communities and economies across the continent. 🌍 #AIForGood #GoogleResearch #TechForAfrica #WeatherForecasting #GoogleInAfrica #ClimateAdaptation

🚀 How Can We Use AI for Climate Resilience: Scaling Solutions for Farmers 🌱🌍 The Financial Times recently highlighted how AI is revolutionizing agriculture, sustainability, and climate resilience—and we’re honoured that TomorrowNow.org was featured as an example of how AI-powered weather intelligence is already making a difference for farmers in Africa. The unpredictability of extreme weather is not an abstract challenge—it’s a daily reality affecting yields, food security, and livelihoods. That’s why AI-driven climate solutions are critical, and why our work—backed by the Gates Foundation—is helping smallholder farmers become more climate-resilient with AI-powered weather advisories delivered via SMS. 💡 What AI Can Do for Farmers: ✅ Precision Agriculture – AI-driven insights can boost farm yields even in the face of climate shocks. ✅ Weather-Based Advisory – TomorrowNow’s AI-powered weather forecasts from Tomorrow.io are helping farmers in Kenya make smarter planting, irrigation, and harvest decisions. ✅ Water Efficiency – With water scarcity on the rise, AI-driven irrigation can reduce waste and improve crop resilience. 🌎 The Bottom Line? Agriculture must embrace AI to adapt, survive, and thrive. AI has the potential to support farmers, enhance food security, and drive more sustainable agriculture worldwide. **The key is ensuring these technologies are accessible and farmer-first.** 🔗 Read the FT article featuring our work here: https://lnkd.in/ezKQNATw #TomorrowNow #FinancialTimes #AIforGood #ClimateResilience #SustainableAgriculture #WeatherIntelligence #FarmersFirst