BIG news from #ECMWF: a portable version of our global forecasting model is now open source! Development of #IFS started in 1987 and just a year before entering its 4th decade it is now available to all.

This continues our journey that started a few years ago, when we made a first and growing number of several key model components available.

See the news article for more details: https://www.ecmwf.int/en/about/media-centre/news/2026/openifs-open-source

Code repository: https://github.com/ecmwf-ifs/openifs

#NWP #Weather #OpenSource

ECMWF’s portable global forecasting model OpenIFS now available for all

OpenIFS is now open source, expanding global access, accelerating innovation and strengthening collaboration.

ECMWF

ECMWF annual seminar 2026 is on "Advancing the assimilation of Earth system observations with new methodology and Machine Learning"

Registration is open:
https://events.ecmwf.int/event/513/overview

#DataAssimilation #MachineLearning #NWP #WeatherForecasting

Annual Seminar 2026

Advancing the assimilation of Earth system observations with new methodology and Machine Learning Data Assimilation continues to underpin all operational weather forecasting, initialising physics-based models as well as initialising and training data-driven models. The 2026 Annual Seminar will look at how the deployment of Machine Learning (ML) technology and innovative methodology is changing the way we assimilate observations for numerical weather prediction, atmospheric composition and...

ECMWF Events (Indico)

@tinoeberl #ECMWF recognized “AI” #NWP is accurate for large scale ~2,000 km features only. Let’s put some science based constraints to the “AI” hype.

https://www.ecmwf.int/en/newsletter/184/news/hybrid-forecasting-nudging-large-scales-ifs-deterministic-aifs

Hybrid forecasting: nudging large scales of the IFS to the deterministic AIFS

 

ECMWF

Je propose des cartes de divers paramètres clés du modèle numérique de prévisions #meteo (#PNT #NWP) GFS entraîné par apprentissage automatique (Machine Learning en anglais) via GraphCast de Google Inc.
Les données binaires initiales sont fournies par la #NOAA
Licence CC-BY 4.0.

https://www.irizone.net/meteo/para/aigfs/1.0.x/index.php

#EUMETSAT via #businesswire:
"
Revolutionärer Metop-SGA1 überträgt bereits Instrumentendaten
"
"Weniger als drei Wochen nach dem Start von Metop-Satellit A1 der zweiten Generation (Metop-SGA1) am 13. August überträgt der Satellit bereits Daten von zwei seiner sechs Instrumente."

https://www.businesswire.com/news/home/20250902406888/de

https://www.eumetsat.int/revolutionary-metop-sga1-already-transmitting-instrument-data

2.9.2025

#EO #Erdbeobachtung #ESA #Europa #NWP #MetOpSG #MetOpSGA1 #MWS #Raumfahrt #RO #Satelliten #SpaceFlight #Wettervorhersage #Wettersatellit

Machine Learning is making a big impression in Numerical Weather Prediction these days. However, explicitly physics-based models still outperform AI models when it comes to extreme events. #meteorology #nwp

https://arxiv.org/abs/2508.15724

Hey sometimes my LinkedIn feed isn't all Facebook for office grunts.

Numerical models outperform AI weather forecasts of record-breaking extremes

Artificial intelligence (AI)-based models are revolutionizing weather forecasting and have surpassed leading numerical weather prediction systems on various benchmark tasks. However, their ability to extrapolate and reliably forecast unprecedented extreme events remains unclear. Here, we show that for record-breaking weather extremes, the numerical model High RESolution forecast (HRES) from the European Centre for Medium-Range Weather Forecasts still consistently outperforms state-of-the-art AI models GraphCast, GraphCast operational, Pangu-Weather, Pangu-Weather operational, and Fuxi. We demonstrate that forecast errors in AI models are consistently larger for record-breaking heat, cold, and wind than in HRES across nearly all lead times. We further find that the examined AI models tend to underestimate both the frequency and intensity of record-breaking events, and they underpredict hot records and overestimate cold records with growing errors for larger record exceedance. Our findings underscore the current limitations of AI weather models in extrapolating beyond their training domain and in forecasting the potentially most impactful record-breaking weather events that are particularly frequent in a rapidly warming climate. Further rigorous verification and model development is needed before these models can be solely relied upon for high-stakes applications such as early warning systems and disaster management.

arXiv.org

Basics of Numerical Weather Prediction (NWP):

1. THE HORIZONTAL MOMENTUM EQUATION:
\[
\frac{d\mathbf{V}}{dt} + f\hat{k} \times \mathbf{V} = -\nabla \phi + \frac{\sigma}{p_s} \frac{\partial \phi}{\partial \sigma} \nabla p_s + \mathbf{F}
\]

2. THE CONTINUITY EQUATION:
\[
\frac{\partial p_s}{\partial t} + \nabla \cdot (p_s \mathbf{V}) + \frac{\partial}{\partial \sigma}(p_s \dot{\sigma}) = 0
\]

3. THE THERMODYNAMIC ENERGY EQUATION:
\[
\frac{1}{R} \frac{d}{dt} \left[ \sigma \frac{\partial \phi}{\partial \sigma} \right] + \frac{RT}{C_p p} \left[ p_s \dot{\sigma} + \sigma\dot{p_s} \right] = -Q
\]

4. HYDROSTATIC EQUATION:
\[
\frac{\partial \phi}{\partial \sigma} = -\frac{RT_v}{\sigma}
\]

5. SURFACE PRESSURE TENDENCY EQUATION:
\[\displaystyle
\frac{\partial p_s}{\partial t} = -\int_{0}^{1} \nabla\cdot (p_s \mathbf{V}) \, d\sigma
\]

6. MOISTURE EQUATION:
\[\displaystyle
\frac{\partial}{\partial t} (p_s q) + \nabla\cdot (p_s q \mathbf{V}) + \frac{\partial}{\partial \sigma} (p_s q \dot{\sigma}) = p_s S
\]

The six primary unknowns are: \(\mathbf{V}\) (horizontal wind velocity), \(p_s\) (surface pressure), \(T\) (temperature), \(q\) (specific humidity or moisture), \(\phi\) (geopotential), and \(\dot{\sigma}\) (sigma velocity or vertical velocity in \(\sigma\)-coordinates).

#NWP #Weather #NumericalWeatherPrediction #Meteorology #Climate #ClimateScience #Earth #EarthScience #ClimateChange #ClimateSciences #Science #WeatherPrediction #Humidity #Moisture #Pressure #Velocity #SurfacePressure #HydrostaticEquation #WeatherPrediction #Ocean #Atmosphere #AOS #ClimateDynamics #WeatherDynamics #Geopotential #SigmaVelocity #VerticalVelocity #MoistureEquation #Thermodynamics #Dynamics #NavierStokes

A largely cloud free satellite image of the central Northwest Passage shows that the prior melt ponds along the southern route have drained, and that melt ponds have now formed on the northern route through McClure Strait:

https://GreatWhiteCon.info/2025/05/the-northwest-passage-in-2025/#Jun-22

#Arctic #SeaIce #NWP

This experimental model uses ML as an adjunct to an atmospheric physics model, which is equation-driven. The continuous corrections provided through ML modeling help the physics-driven model achieve significant gains, up to half a day of previsibility at 7 days forecast.

Note: I am no longer affiliated with the Meteorological Service of Canada. Any enquiry should be directed as specified in the announcement.

https://comm.collab.science.gc.ca/mailman3/hyperkitty/list/[email protected]/thread/D3ELXLOLV2F2OORRNNI746H2ZDF3RWVD/

#weather #NWP #MachineLearning
#meteorology
2/2

[NOUVEAU] : Version expérimentale du SGPD piloté spectralement (IA) // [NEW] : Experimental version of the spectrally nudged GDPS (AI) - dd_info - Comm.Collab.Science.Gc.Ca

Short thread

You may or may not be aware that since the early 1960s Canada has developed and operates a world class Numerical Weather Prediction program and that most of the operational data products are available for free. I rarely share technical announcements, but this one is for you if your work and/or research involves using weather models.

#meteorology #NWP #Weather #machineLearning

1/2