Google DeepMind has made an artificial intelligence system for tracking global rainfall available to researchers and developers, a move that could improve flood and drought monitoring in areas with limited weather infrastructure.
The company announced on Nov. 13 that users can now access historical records and continuously updated data from PARDIG, its AI-powered global precipitation monitoring system.
The technology combines satellite observations with atmospheric data to estimate rainfall every 15 minutes across a global grid at roughly 10-kilometer resolution.
Its release could broaden access to rainfall intelligence in developing countries and remote regions where sparse weather radar coverage makes storms, flooding, and prolonged dry spells difficult to track.
Google said PARDIG learns from precipitation measurements collected by NASA’s Global Precipitation Measurement satellite and uses additional satellite imagery and atmospheric information to estimate rainfall worldwide.
Unlike conventional monitoring systems that depend heavily on ground-based weather stations, PARDIG can generate estimates over oceans and regions with limited observation networks.
The dataset also supports Google’s WeatherNext 3 forecasting model, which uses PARDIG-generated rainfall information during training.
Researchers can access historical rainfall records dating back to 1999, along with a separate dataset covering 2016 to the present.
Google has released both under a Creative Commons license that permits commercial reuse with attribution, potentially allowing companies to incorporate the information into agricultural monitoring, water management, and disaster-risk applications.
Google’s AI outperforms NASA rainfall benchmark by more than 30%
DeepMind said PARDIG achieved more than 30% higher hourly precipitation detection skill than NASA’s IMERG-Final dataset when measuring light-to-moderate rainfall.
The company evaluated the system against radar networks in the United States, Europe, and Japan, as well as observations from more than 36,000 weather stations worldwide.
Its research paper also reported improved identification of intense rainfall events, which existing global precipitation datasets can struggle to capture accurately.
These capabilities could help researchers develop more detailed models of extreme weather events, including flash floods, landslides, and drought conditions.
Google highlighted the July 2025 Texas Hill Country floods as one example, comparing PARDIG’s estimates with ground-based radar measurements and NASA’s existing rainfall dataset.
However, the technology has limitations that could constrain its use during rapidly developing emergencies.
Although PARDIG estimates rainfall at 15-minute intervals, Google publishes its near-real-time data in six-hour batches, creating an average delay of about five hours between observation and availability.
Fine time resolution does not mean immediate delivery.
Google’s documented delivery schedule, not a measured uptime or speed test. Latency is approximate at publication. PARDIG is experimental precipitation estimation and does not replace official warnings.
This means emergency services cannot use the dataset as an immediate substitute for live weather warnings.
Google also acknowledged that the system can underestimate snowfall in mountainous regions and is less reliable in certain polar areas.
Meanwhile, the company still classifies PARDIG as experimental and advises users to rely on official meteorological agencies when making public-safety decisions.
Also Read
- Bitcoin Slides to Four‑Month Low Near $62,000 Amid Surge in Liquidations and Dollar‑Market Sell‑Off
- Sirius Real Estate Extends Dividend Growth Streak to 26 Consecutive Years
- Bank of Japan’s Takata Urges Faster Rate Hikes as Weak Yen and Ultra-Low Real Rates Raise Inflation Risks
- National Bank of Pakistan Publishes Updated Foreign Exchange Rates

