Synthetic intelligence (AI) make progress on the earth of climate forecast … European heart for the forecast makes its AI mannequin absolutely purposeful

Paul Dorian

overview

Artificial intelligence (AI) is a collection of technologies that can use computers to perform tasks that typically require human intelligence, and is increasingly affecting the world of weather forecast. The European Center for Medium-sized forecast (ECMWF) has made progress with its artificial intelligence forecast system (AIFS) because it has recently become fully functional and has now been led side by side with its traditional physical integrated forecast system (IFS). According to the ECMWF, the AIFS exceeded the physical model for many measures, including, for example, tropical cyclone traces. In addition to the ECMWF AIFs, there are at least four other well-known AI-trained weather models, including NOAA/Google Graphcast, Microsofts Aurora, Nvidia's FourCast and Huawei's Pangu-Weather.

A part-time comparison of the conventional European forecast model (left) and the Euro AI version (right) of “total snow amounts” in the continental USA for the period of Friday, April 4,,. Cards with friendly approval from ECMWF, central weather

discussion

The traditional approach to the weather forecast was to use the numerical weather forecast (NWP), which is based on current conditions, physical models and the solution of complex equations for powerful supercomputers to output parameters such as temperature, pressure, wind and precipitation. KI -KI models for artificial intelligence (AI), especially machine learning, are increasingly being used to improve the weather forecast by learning from large data sets with weather data to identify patterns and trends. AI models can process data faster and identify complex patterns, which may lead to faster and more precise forecasts. The increasingly important role of the AI ​​in the weather forecast will be to supplement and improve conventional NWP models.

The European Center for Center for Center (ECMWF) has made its forecast system for artificial intelligence (AIFS) the first such fully functional weather forecast for machine learning and artificial intelligence. Praising such a system means being openly available and supporting the meteorological community around the clock. These AIFS can generate a wide range of output parameters such as winds, temperatures and details on precipitation types from snow to rain. The AIFS currently have a network distance of 28 km and, according to the ECMWF, can exceed the physical counterpart for certain measures by up to 20%.

A co-side comparison of the conventional European forecast model (left) and the Euro AI version (right) of “500 millibar Height Anomalies” in the continental US COM in the continental USA. Cards with friendly approval from ECMWF, central weather

The AIFS uses the same initial atmospheric conditions for its forecasts as the IFS. These are based on the combination of an earlier short -term forecast with around 60 million high -quality observations of satellites as well as many other streams, including from airplanes, boats, sea boys and many other earth measuring stations. Every six hours, these initial conditions lead to the AIFS. The machine learning model, which was trained on how the weather has developed in the past, evaluates how the initial conditions will influence the weather in the coming days. In contrast, the IFS uses physics -based functions to make a forecast with a network spacing of 9 km in the globe, whereby the laws of physics are integrated into its computer code.

Meteorologists rarely use complete trust in a single model, but see a number of model forecasts. Most global models calculate the 10-day prospects every six hours, and a disposable forecasting measure the reliability of a model by checking its consistency from run to run. The Euro-AI model was good at this test for the hurricane Francine in September 2024, as it ended up every run from 96 hours until its landfall near Morgan City, LA. Tickets with friendly approval from foxweather.com

The first operational version is called AIFS “single”. A single forecast is carried out, which is known as a deterministic forecast. However, ECMWF pushes this model to create a collection of 50 different forecasts with minor deviations at a certain point in time in order to provide the entire area of ​​possible scenarios. This is known as an ensemble modeling, a technology that was developed and implemented by ECMWF more than thirty years ago. According to the ECMWF, the start of AIFS “Single” as an operational service is the first step to improve its artificial intelligence forecast skills. The next step will be to make ensemble forecasts available through the use of artificial intelligence and extended range (seasonal) forecasts, and we will continue to monitor progress here at Arcfield Weather.

Meteorologist Paul Dorian
Arcfield
arcfieldweather.com

Follow us on Facebook, Twitter, YouTube

Like this:

How Load…

Related

Do you discover more from watts?

Subscribe to the latest posts to your e -mail.