An Operational Deep Learning System for Satellite-Based High-Resolution Global Nowcasting
Shreya Agrawal; Mohammed Alewi Hassen; Emmanuel Asiedu Brempong; Boris Babenko; Fred Zyda; Olivia Graham; Di Li; Samier Merchant; Santiago Hincapie Potes; Tyler Russell; Danny Cheresnick; Aditya Prakash Kakkirala; Stephan Rasp; Avinatan Hassidim; Yossi Matias; Nal Kalchbrenner; Pramod Gupta; Jason Hickey; Aaron Bell · 2025-10-10
The paper introduces Global MetNet, a global machine learning model for real-time precipitation nowcasting that predicts rainfall up to 12 hours ahead using satellite and global weather data rather than radar. Operating at high spatial (∼5 km) and temporal (15-minute) resolution, it significantly outperforms traditional numerical weather prediction models—especially in data-sparse regions of the Global South—offering rapid, accurate, and equitable forecasts already deployed to millions of users via Google Search.
Shreya Agrawal; Mohammed Alewi Hassen; Emmanuel Asiedu Brempong; Boris Babenko; Fred Zyda; Olivia Graham; Di Li; Samier Merchant; Santiago Hincapie Potes; Tyler Russell; Danny Cheresnick; Aditya Prakash Kakkirala; Stephan Rasp; Avinatan Hassidim; Yossi Matias; Nal Kalchbrenner; Pramod Gupta; Jason Hickey; Aaron Bell. 2025-10-10. An Operational Deep Learning System for Satellite-Based High-Resolution Global Nowcasting. https://arxiv.org/pdf/2510.13050 (AI & Environment Resource Hub; record paper-401; collection snapshot 2026-09-15).
Classification and context
Included in the original Scientific Paper collection. Labels below are metadata based suggestions. They are not verified findings, claims of effectiveness, or endorsements.
Shreya Agrawal; Mohammed Alewi Hassen; Emmanuel Asiedu Brempong; Boris Babenko; Fred Zyda; Olivia Graham; Di Li; Samier Merchant; Santiago Hincapie Potes; Tyler Russell; Danny Cheresnick; Aditya Prakash Kakkirala; Stephan Rasp; Avinatan Hassidim; Yossi Matias; Nal Kalchbrenner; Pramod Gupta; Jason Hickey; Aaron Bell
Publisher
arXiv
Publication date
2025-10-10
Date precision
day
Geographic scope
Unknown
Language
unknown
Peer review
unknown
Source status
verified
Last verified
2026-09-15T16:44:50.045Z
Snapshot import
2026-09-06
Legacy domain
Cross-Cutting Sustainability
Legacy subdomain
Global MetNet: A Global Machine Learning Model for Real-Time Precipitation
journal
arXiv
Provenance and review
Source sheet: Scientific Paper · Row 405 · Original ID: paper-401.
A paper listing is not a quality assessment. Peer review and findings require source-level confirmation. Source link reachable · checked 2026-09-15. Source identity and required metadata verified. The import date is not the original date added.
Philip Johnston is the co-founder and CEO of Starcloud, the company building data centers in space. In November 2025, Starcloud launched an Nvidia H100 GPU into orbit and trained the first large language model in space. They've since raised $200 million, hit a billion-dollar valuation just 17 months after YC demo day and filed with the FCC to deploy 88,000 more satellites. In this episode, Philip walks us through their wild origin story, the engineering challenges behind the Starcloud-1, why they booked a SpaceX launch before they even knew what they were building and how data centers in space make sense both economically and politically.
Source metadata available
Shared topics: Cross-cutting sustainability; Governance and society; Research tools and geospatial methods
Trust is a cornerstone and enabler of human civilization, determining the very nature of how people interact with each other. The swift integration of artificial intelligence (AI) into daily life poses grand societal challenges and necessitates a reevaluation of trust. Our bibliometric literature review calls for scientists and stakeholders to cross traditional academic boundaries to address emerging and evolving societal challenges arising from AI. We propose a transdisciplinary research framework to understand and bolster trust in AI and address grand challenges in domains as diverse and urgent as misinformation, discrimination, and warfare.
Humanities and Social Science Communications
Shared topics: Cross-cutting sustainability; Research tools and geospatial methods; Weather, hazards, and adaptation
This study showcases a machine learning hackathon in France where a CNN-based model accurately predicted flood risk evolution using geospatial and climate data, even without streamflow input, emphasizing trustworthy AI in data-scarce regions.
ICLR
Shared topics: Cross-cutting sustainability; Governance and society; Research tools and geospatial methods
