Monday, 16 November 2026
BTC $64,621 +0.15% ETH $1,823 +0.60% SOL $71.13 −0.43% XRP $0.957 +0.66% BNB $589.61 −0.17% DOGE $0.059 +1.05% ADA $0.153 −0.30% TON $1.05 +0.63% AVAX $7.14 −0.76% BTC $64,621 +0.15% ETH $1,823 +0.60% SOL $71.13 −0.43% XRP $0.957 +0.66% BNB $589.61 −0.17% DOGE $0.059 +1.05% ADA $0.153 −0.30% TON $1.05 +0.63% AVAX $7.14 −0.76%
GPreferred sourceon Google
Advertise
AdvertisementInvestIn.News: Bullish or bearish? Vote on live crypto flashes
Latest
Crypto Newsroom

Google DeepMind opens PARDIG global AI rainfall dataset

Google DeepMind opened access to PARDIG on Nov. 13, saying researchers and developers can now use its AI system for estimating rainfall worldwide, with historical records stretching back to 1999. The announcement, as reported by CryptoSlate, positions the dataset as a tool for tracking floods and droughts in regions where ground-based weather stations and radar coverage are sparse.

Google DeepMind’s PARDIG estimates global rainfall every 15 minutes on a roughly 10-kilometer grid. Google says it achieved more than 30% higher detection skill than NASA’s IMERG-Final dataset for light-to-moderate rainfall, but near-real-time data arrives in six-hour batches with an average delay of about five hours.

How PARDIG estimates rainfall without ground stations

The system learns from precipitation measurements collected by NASA’s Global Precipitation Measurement satellite, then uses additional satellite imagery and atmospheric data to produce estimates across the planet. That approach differs from conventional monitoring that leans heavily on ground-based weather stations, letting PARDIG generate figures over oceans and regions with thin observation networks.

DeepMind said it evaluated the system against radar networks in the United States, Europe and Japan, and against observations from more than 36,000 weather stations worldwide. Its research paper also reported improved identification of intense rainfall events, which existing global datasets can struggle to capture accurately.

The company pointed to the July 2025 Texas Hill Country floods as one example, comparing PARDIG’s estimates with ground-based radar readings and NASA’s existing rainfall dataset. PARDIG-generated rainfall information also supports Google’s WeatherNext 3 forecasting model during training.

The five-hour delay and other limits

Fine time resolution does not mean immediate delivery. While PARDIG produces estimates every 15 minutes, Google publishes them in six-hour batches, so the oldest frame in a batch is roughly eight hours old and the newest roughly two hours old when it appears. The average delay of about five hours means emergency services cannot treat the dataset as a live substitute for official weather warnings.

Google also acknowledged that the system can underestimate snowfall in mountainous regions and is less reliable in certain polar areas, citing those shortcomings in its documentation.

AdvertisementInvestIn.News: Bullish or bearish? Vote on live crypto flashes

Why it matters

Rainfall intelligence shapes decisions in agriculture, water management, insurance and disaster response, and much of that data has historically been scarce where weather infrastructure is thinnest. Google’s Creative Commons licensing removes a common barrier for companies and researchers who want to build monitoring products on top of the estimates, even if the licensing does not eliminate the latency problem. The >30% skill gain over NASA’s benchmark, if it holds up under independent scrutiny, would give researchers a sharper tool for studying flash floods, landslides and prolonged dry spells.

The dataset also arrives in a year when AI-driven weather and climate research has accelerated, and the comparison to established public benchmarks like IMERG-Final gives researchers a concrete way to judge the new entrant rather than take marketing claims at face value.

What to watch

Independent validation will determine whether DeepMind’s benchmark advantage survives comparison with NASA’s own datasets in operational settings. Google’s documented delivery schedule, rather than its estimation interval, is the figure to check when assessing whether PARDIG can support faster disaster response in the future. Whether Google moves PARDIG out of experimental status will also matter for how official agencies treat its output.

Frequently Asked Questions

What is PARDIG?

PARDIG is Google DeepMind’s AI-powered global precipitation monitoring system, which combines satellite observations and atmospheric data to estimate rainfall every 15 minutes on a roughly 10-kilometer grid.

How accurate is PARDIG compared with NASA’s rainfall data?

Google DeepMind says PARDIG achieved more than 30% higher hourly precipitation detection skill than NASA’s IMERG-Final dataset for light-to-moderate rainfall, tested against radar networks in the US, Europe and Japan and observations from over 36,000 weather stations.

Why can’t PARDIG be used for live flood warnings yet?

Although estimates are produced at 15-minute intervals, Google publishes near-real-time data in six-hour batches with roughly two to eight hours of latency and an average delay of about five hours. Google also labels the system experimental and says users should defer to official warnings.

Can the PARDIG data be used commercially?

Google has released both the historical records dating back to 1999 and the dataset covering 2016 to the present under a Creative Commons license that permits commercial reuse with attribution.

Source: CryptoSlate

Not investment adviceBitcoinWorld publishes news and analysis for information only. Nothing here is a recommendation to buy, sell or hold any asset. Digital assets are volatile and you can lose your entire capital. Consider your own circumstances and speak to a regulated adviser before acting. Read the full disclaimer.

Keshav Aggarwal

Co-Founder & Responsible Editor

Keshav Aggarwal is the Co-Founder and Responsible Editor of BitcoinWorld, a digital publication covering cryptocurrency, blockchain and Web3 since 2020. He has worked in the digital-asset space for over six years and helped build one of India's most active crypto communities. At BitcoinWorld he sets editorial direction, reviews every AI-assisted article before publication and reports on the business of crypto and Web3: funding rounds, product launches and the regulatory shifts shaping digital finance.

Spotted an error in this article? We correct openly and log every change.Report a correction

More in Crypto Newsroom

See all →