AI Weather Forecasts not for tomorrow,
but for months ahead

A new paradigm

Accurately and reliably forecasting the weather months ahead is an enormous challenge. Like most complex problems, solving it requires a new way of thinking.
AI: Data driven

In contrast to the classical weather forecasting models, our model forecasts by learning from historical connections in our climate system. Our AI model is specifically designed to extract the signal from the slow-moving parts of our climate system and to use these for forecasting. This enables a much more accurate and reliable forecast weeks up to months ahead.

Climate Expertise

We know that creating a skillful data-driven weather forecast is not enough. We have the climate expertise to explain which physical processes are the drivers of our AI weather forecast. Combined with a new branch of AI research, explainable AI, we can identify where and when our forecast is reliable. These are essential elements for creating trustworthy predictions.


We can co-develop weather forecasts, enabling us to fine-tune a forecast to specific needs. Classical weather services rely on general weather forecasts that often fail to meet desired accuracy and specificity that are essential for operational use. Our custom-made weather forecast can  greatly simplify the decision making process.

Direct impact

As the weather is omnipresent, almost anything is impacted by the weather. To further help with decision making, we can easily use our methods to not only predict weather, but also its impact. Our latest research shows high performance for impact forecasting. By utilising the concept of 'windows of opportunity', we can identify where and when our forecast is reliable and thus where impact is to be expected.

About S2s

Sub-seasonal to seasonal (S2S) weather forecasting refers to predicting weather 2 weeks up to 3 months ahead. Up to twenty years ago it was deemed impossible to predict the weather on S2S timescales. In literature, it was even dubbed: the prediction desert. The growing availability of climate data and new computational power, with the help of AI, have now paved the way to make reliable predictions weeks up to seasons ahead.

Why Now?

Our Team

Our team has a strong grounding both in leading research on climate science and AI.
Photo credits to Peter Valckx. For symposium A Better World @ VU.


We offer accurate and reliable S2S weather forecasting services tailored to your business needs.
Air temperature

An air temperature forecast for 1-3 months ahead for an area of your interest. You will get the AI and climate system expert knowledge from leading scientists in the field.

Get in touch
Check out our latest work

The forefront of S2S research

Contact us now!

De Boelelaan 1111, 1081HV Amsterdam