Logo
Empowering weather and climate forecast

Just to give you some idea.

Machine learning does great jobs wherever fuzzy, chaotic data must be understood, especially such data coming in vast amounts. And what's more fuzzy and chaotic than our entire planet with oceans, continents, atmosphere, clouds, plants, animals, and 8 billion people messing with it?

Weather & climate: Basis of life. Sometimes our nemesis.

Weather has always been so dominant in man's life that almost all cultures worshipped weather gods. In Central America alone,
0 weather gods
were known: Chaac, Coatrisquie, Cocijo, Ehecatl, Guabancex, Guatauva, Huracán, Juracán, K'awiil, Q'uq'umatz, Kukulkan, Quetzalcoatl, Tezcatlipoca, Tlaloc, Tohil, Tupã, Yopaat.

Compare this to:

That's
0
times all Christian gods.

That's
1.4
times all Greek gods on Olympus.

Extreme weather events caused over
0
casualties in the past 20 years.

Compare this to:

That's
5.3
times the people killed in German traffic in the same period.

That's
4.8
times the number of casualties in the Bosnian war.

Extreme weather events caused over over
0 Billion US$
economic damages in the past 20 years.

Compare this to:

That's
2.4
times the GDP of India in 2005.

That's
0
times the GDP of Haiti in 2020.

World Bank expects up to
0
climate refugees by 2050, who will have to find new homes because of global warming.

Compare this to:

That's
2.6
times the global number of refugees in 2020.

That's
3
times the projected population of Germany in 2050.

Is it really difficult to predict weather and climate?
Yes it is.

Earth's atmosphere contains about
0
molecules, which pretty much do what they please.

Compare this to:

That's
0
times the number of stars in the universe.

That's
0
times the age of the universe in seconds.

To predict just the weather in Europe, around
0
observations must be processed. Every day!

Compare this to:

That's
0
times the number of observations a person could read in a day (assuming 10 per minute).

A garden-variety weather-related machine learning problem deals with about
0
input variables for one single sample.

Compare this to:

That's
0
times the input variables in a machine learning app that can distinguish a cat from a dog.

But we are better equipped for the task than ever before.

The fastest supercomputer in the world performs
0 operations/sec
at peak performance

Compare this to:

That's
0
times the Saturn V Apollo Guidance Computer.

That's
0
times a Pentium 4 Processor, 3.2 GHz.

That's
0
times an Xbox Scorpio GPU.

The European Centre for Medium-Range Weather Forecasts more than
0 MB
of weather data stored.

Compare this to:

That's
0
times free iCloud storage (as of 2021).

That's
0
times the capacity of a regular computer hard drive.

Degrees of freedom are a measure for the complexity of statistics. Our high-resolution prediction model has almost
0
DoF in a single simulation.

Compare this to:

That's
0
times the degrees of freedom a person could update manually in one day (assuming 5 per minute).