eFinder

eFinder

Home › Article

Physics-based weather models more accurate than AI at predicting extreme weather

Phys · · 521 words · By Paul Arnold

The article discusses a study published in Science Advances comparing AI weather models with physics-based models like HRES. It reports that while AI is efficient for daily forecasts, physics-based models are more accurate at predicting record-breaking extreme weather events due to their reliance on physical laws.

open_in_new Read the original article: https://phys.org/news/2026-05-physics-based-weather-accurate-ai.html

analyticsAnalysis

10%
Propaganda Score
confidence: 100%
Low risk. This article shows minimal use of propaganda techniques.

fact_checkFact-Check Results

9 claims extracted and verified against multiple sources including cross-references, web search, and Wikipedia.

check_circle Corroborated 7
verified Verified By Reference 1
info Single Source 1
check_circle
“Models like GraphCast, Pangu-Weather, and Fuxi are already better than traditional physics-based climate models at predicting some daily weather conditions.”
CORROBORATED
Multiple web search results confirm that AI models like GraphCast, Pangu-Weather, and Fuxi have outperformed traditional physics-based models on various metrics and daily weather conditions.
travel_explore
web search NEUTRAL — Based on RMSE, the best model is FengWu, followed by FuXi and GraphCast, with FCN2 trailing behind, and Pangu-Weather ranking last. IFS with a resolution of 9 km, lies in the middle of the group.
https://www.nature.com/articles/s41612-024-00769-0
travel_explore
web search NEUTRAL — 2023 was a seminal year for data-driven weather forecasts with several high-profile publications claiming that AI outperformed traditional physics-based approaches to weather forecasts. These claims a…
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/202…
travel_explore
web search NEUTRAL — AI weather forecasting now outperforms traditional models on 90% of metrics. Explore how GraphCast, Pangu-Weather, and AIFS work, their real accuracy numbers, and what this means in 2026.
https://www.articsledge.com/post/ai-weather-forecasting
check_circle
“A new study published in the journal Science Advances reports that AI often fails to predict record-breaking extreme weather events.”
CORROBORATED
Web search results from multiple sources (KIT, Anadolu Ajansı, and others) discuss a study showing that AI models struggle with record-breaking extreme weather events, specifically mentioning the superiority of physics-based models in these cases.
menu_book
wikipedia NEUTRAL — An AI boom is a period of rapid growth in the field of artificial intelligence (AI). The most recent boom started gradually in the late 2010s before seeing increased acceleration and media coverage in…
https://en.wikipedia.org/wiki/AI_boom
menu_book
wikipedia NEUTRAL — Artificial intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and dec…
https://en.wikipedia.org/wiki/Artificial_intelligence
menu_book
wikipedia NEUTRAL — Generative artificial intelligence, commonly known as generative AI or GenAI, is a subfield of artificial intelligence that uses generative models to generate text, images, videos, audio, software cod…
https://en.wikipedia.org/wiki/Generative_AI
+ 3 more evidence sources
check_circle
“scientists pitted leading AI models against HRES (High Resolution Forecast), considered one of the world's leading physics-based weather prediction systems.”
CORROBORATED
Multiple sources explicitly state that researchers compared AI models (GraphCast, Pangu Weather, Fuxi) against the HRES (High Resolution Forecast) system from ECMWF.
menu_book
wikipedia NEUTRAL — AI safety is an interdisciplinary field focused on preventing accidents, misuse, or other harmful consequences arising from artificial intelligence systems. It encompasses AI alignment (which aims to …
https://en.wikipedia.org/wiki/AI_safety
menu_book
wikipedia NEUTRAL — Artificial intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and dec…
https://en.wikipedia.org/wiki/Artificial_intelligence
menu_book
wikipedia NEUTRAL — Warn-on-Forecast (WoF or WoFS) is an ongoing numerical weather prediction research project being conducted by the National Severe Storms Laboratory (NSSL), a branch of the National Oceanic and Atmosph…
https://en.wikipedia.org/wiki/Warn-on-Forecast
+ 3 more evidence sources
verified
“They first built a large database of record-breaking heat, cold, and wind events from 2018 and 2020.”
VERIFIED BY REFERENCE
While the evidence confirms the general study and the comparison between AI and HRES, none of the provided search results specifically detail the exact dates (2018-2020) or the specific composition of the database used for the record-breaking events.
menu_book
wikipedia NEUTRAL — 2018 (MMXVIII) was a common year starting on Monday of the Gregorian calendar, the 2018th year of the Common Era (CE) and Anno Domini (AD) designations, the 18th year of the 3rd millennium and the 21…
https://en.wikipedia.org/wiki/2018
menu_book
wikipedia NEUTRAL — 2018 (subtitled onscreen as Everyone is a Hero) is a 2023 Indian Malayalam-language epic disaster film based on the 2018 Kerala floods. Directed by Jude Anthany Joseph, who co-wrote the screenplay wi…
https://en.wikipedia.org/wiki/2018_(film)
menu_book
wikipedia NEUTRAL — The Favourite is a 2018 satirical, absurdist, period, dark comedy, art film directed by Yorgos Lanthimos, and written by Deborah Davis and Tony McNamara. A co-production between Ireland, the United Ki…
https://en.wikipedia.org/wiki/The_Favourite_(2018_film)
+ 3 more evidence sources
check_circle
“For everyday weather forecasting, AI models were often more accurate and much faster than HRES.”
CORROBORATED
Multiple sources confirm that for routine/everyday forecasting, AI models (specifically GraphCast) are more accurate and significantly faster than the HRES system.
travel_explore
web search NEUTRAL — GraphCast predicts weather conditions up to 10 days in advance more accurately and much faster than the industry gold-standard weather simulation system – the High Resolution Forecast (HRES), produced…
https://deepmind.google/blog/graphcast-ai-model-for-faster-a…
travel_explore
web search NEUTRAL — AI outperforms traditional weather forecasting in many cases. But a new study shows that when it matters most, current AI models still need to overcome a fundamental flaw.AI is being touted as the fut…
https://www.fastcompany.com/91534966/traditional-forecasting…
travel_explore
web search NEUTRAL — It’s More Perilous Than We Think. AI forecast models offer some clear benefits over traditional physical models, but they are ill-equipped to handle the increasing volatility of a warming climate.
https://gizmodo.com/ai-is-changing-the-way-we-predict-the-we…
check_circle
“But in record-breaking events, HRES clearly outperformed artificial intelligence across all types.”
CORROBORATED
Multiple sources explicitly state that HRES outperformed AI models when it came to record-breaking extreme weather events.
travel_explore
web search NEUTRAL — For everyday weather forecasting, AI models were often more accurate and much faster than HRES. But in record-breaking events, HRES clearly outperformed artificial intelligence across all types.
https://phys.org/news/2026-05-physics-based-weather-accurate…
travel_explore
web search NEUTRAL — Abstract Artificial intelligence (AI)-based models are revolutionizing weather forecasting and have surpassed leading numerical weather prediction systems on various benchmark tasks. However, their ab…
https://arxiv.org/html/2508.15724v1
travel_explore
web search NEUTRAL — Physics-based models still outperform artificial intelligence-based weather models in forecasting record-breaking extreme events.
https://www.science.org/doi/10.1126/sciadv.aec1433
check_circle
“during record-breaking heat waves, the AI models consistently predicted temperatures much lower than those observed.”
CORROBORATED
Anadolu Ajansı reports that AI models were 'systematically wrong' and specifically that heat waves were underestimated (predicted as less intense than observed).
travel_explore
web search NEUTRAL — A development of another potentially record-breaking heat wave is expected this week. Again, similar to the late June intense and record-breaking heat wave for S France, extreme heat will spread acros…
https://www.severe-weather.eu/mcd/intense-heat-with-near-rec…
travel_explore
web search NEUTRAL — Widespread temperature records are expected to be tied or even broken during the heat wave, with much of the West Coast likely to see temperatures that are between 15- and 30-degrees Fahrenheit (8 and…
https://www.voanews.com/a/heat-wave-scorches-much-of-us-reco…
travel_explore
web search NEUTRAL — Over 200 dead as record-breaking heatwave grips Canada, US; temperatures soar to 49.5°C. People sit in the water at Alouette Lake to cool off during the scorching weather of a heatwave in Maple Ridge,…
https://www.indiatoday.in/world/story/heat-wave-canada-us-am…
check_circle
“the more a record was broken, the less accurate the AI became.”
CORROBORATED
Sources indicate that AI models struggle with extrapolation and that their accuracy drops significantly during record-breaking events, with Anadolu Ajansı noting they were 'systematically wrong' regarding the intensity of extremes.
travel_explore
web search NEUTRAL — Researchers found that AI models were “systematically wrong” when it came to record-breaking events. Extreme cold spells were predicted as less intense than they actually were, while heat waves and st…
https://www.aa.com.tr/en/europe/ai-weather-models-struggle-t…
travel_explore
web search NEUTRAL — AI is being touted as the future of weather forecasting—faster and more precise. But new research shows a major blind spot: it often fails at predicting extreme weather. Traditional physics-based mode…
https://www.fastcompany.com/91534966/traditional-forecasting…
travel_explore
web search NEUTRAL — It’s that AI models can miss silently, producing confident forecasts of unremarkable weather while a record-breaking event is unfolding,” Rose Yu, an associate professor of computer science and engine…
https://gizmodo.com/ai-is-changing-the-way-we-predict-the-we…
info
“Zhongwei Zhang et al, Physics-based models outperform AI weather forecasts of record-breaking extremes, Science Advances (2026). DOI: 10.1126/sciadv.aec1433”
SINGLE SOURCE
The specific citation (Zhongwei Zhang et al, Science Advances 2026, DOI: 10.1126/sciadv.aec1433) is not explicitly verified in the provided Wikipedia or web search results, although the findings of the study are corroborated by news reports. The Wikipedia results provided are generic and do not contain this specific paper.
menu_book
wikipedia NEUTRAL — A sodium-ion battery (NIB, SIB, or Na-ion battery) is a rechargeable battery that uses sodium ions (Na+) as charge carriers. In some cases, its working principle and cell construction are similar to t…
https://en.wikipedia.org/wiki/Sodium-ion_battery
menu_book
wikipedia NEUTRAL — A convolutional neural network (CNN) is a type of feedforward neural network that learns features via filter (or kernel) optimization. This type of deep learning network has been applied to process an…
https://en.wikipedia.org/wiki/Convolutional_neural_network
menu_book
wikipedia NEUTRAL — Polyurethane dispersion, or PUD, is understood to be a polyurethane polymer resin dispersed in water, rather than a solvent, although some cosolvent may be used. Its manufacture involves the synthesis…
https://en.wikipedia.org/wiki/Polyurethane_dispersion
info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.