Thank you very much for the introduction, Georg.

Thank you very much for the invitation.

Also, I had a chance to look at DERMOS.

It's astonishing to see it there.

I was just surprised because I made my diploma thesis on the DERMOS with a computational

fluid dynamics parallelization.

At that time, you could get a diploma with a parallelization.

Times are changing.

Okay, good.

So today I would like to report about what's going on with machine learning and weather

prediction.

So at first a little bit of historical review.

Then I would like to give you an introduction into how weather prediction is done in general,

what the components are algorithmically, what data they use and all these things.

And then we have a look about the integration of machine learning that started, let's say,

the first time in 2018-19.

But then there were some quite astonishing papers in the year 23.

And since then, it accelerates.

And of course, Google is involved in that.

Google DeepMind.

We will have a look at this Graphcast as one of the representatives of machine learning

based weather prediction concepts.

And then we have a look at the ECMWF.

So ECMWF is the Center for Medium Range Weather Forecast that we have, reading based in the

past, now moved to Bologna and Bonn.

And finally, I go back to something that is not yet machine learning based.

That is the ICON model that is used by DWD, because I would like to show you one of the

visualizations that the Max Planck colleagues did a few years ago.

So of course, history of weather prediction.

Obviously, we do this for millennia.

We always want to know what will be the weather for the next time.

So for the crop and all these things.

And in the end, we would like to know what is the atmosphere, let's say, temperature,

precipitation, things like that, at a given point of time, at a given location.

And in the past, all that was done qualitatively.

People made observations.

Then they had some special kind of calendar.

And it started to become formalized in the late 19th century.

Then they had the first attempts of putting together the formulas for that.

And then they started to do a little bit of hand computation.

What you need is, of course, then quantitative data about the state of the atmosphere.

If you want to make it in your area, if you want to make a global weather prediction,

then of course, you also need to know the state of the ocean.

So, meteorology as a science developed.

And then we had the first weather report in 1st of August 1861.

That was in England.

I always think that weather prediction as a topic is more for those countries where

the weather is notoriously bad.

Because I think it could not have developed in Spain or so.

Nobody's interested because you say, okay, tomorrow the weather will be the same as today.