With new supercomputer, KNMI can predict even better weather: ‘4,000 trillion calculations per second’

A new supercomputer is helping the KNMI to calculate short-term weather forecasts with increasing precision. This will also allow warnings for extreme weather to be issued in increasingly more detail per region in the future.

An extreme downpour that lasted for hours hit the Achterhoek and Twente on Sunday 21 and Monday 22 July. It produced apocalyptic images of cars completely submerged on the highways, the fire brigade having to rescue people from roofs and streets that were flooded. Locally, one and a half times the average amount of rain for the entire month of July fell in a few hours.

Warn earlier

Heavy downpours like these seem to come out of nowhere. One moment you look at Buienradar and there is not a cloud in the sky, the next moment a strip of dark red clouds suddenly approaches you. Climate change has increased the chance of downpours ninefold, resulting in flooding risks. The warming of the surface of the oceans causes more water vapour in the air and therefore more rain. If clouds move slowly, a lot of rain can fall locally in a small area, with all the consequences that entails.

That is why the KNMI is working on techniques to announce sudden weather extremes earlier, says Maarten van Aalst, director of the KNMI. “The trick is to warn people in time when the weather becomes really extreme and dangerous. Even two hours in advance, a warning is still very valuable for people to take measures.”

See also

More computing power

“The closer we get to an extreme weather event, the better we can predict it with models,” Van Aalst explains. “But we can also observe it earlier with increasingly better satellites and rain radars.” The KNMI has also invested in computing power to be able to expect the weather even earlier and more accurately. “We have just bought a beautiful, new supercomputer, together with Iceland, Denmark and Ireland. An even more powerful machine with which we can calculate the weather,” says Van Aalst.

This supercomputer is located in the Icelandic capital Reykjavik and can make 4,000 trillion calculations per second. “With this weather data, we can not only look a bit further ahead, but we can also make more precise and therefore smaller-scale forecasts,” says Van Aalst. In particular, short-term weather forecasts are improved by it. The computing capacity of the supercomputer is almost ten times greater than the previous computer.

Smaller areas

It is fed with the enormous amounts of data from satellites, observations at home and abroad, from radars on ships and on buoys at sea. The weather forecasts that the computer makes are received almost immediately by the KNMI in De Bilt. With the new device, the institute wants to make weather forecasts and warnings possible at a regional level, instead of just at a provincial level.

“The fact that we can now only issue the weather alarm (code red) and the weather warnings (code yellow and orange) per province actually limits us,” Van Aalst explains. “For example, for heat, the forecast often applies to large parts of the Netherlands and then such an alarm or warning is fine. But in the case of very extreme rainfall, you often see that one part of the province is hit very hard and another part of the province is not.”

See also

More precise warnings

He sees that people don’t like receiving weather alerts that they don’t notice. “Then we get the phone calls again from people who say: ‘guys, code orange, I can’t see it!’. So it will help us if we can warn more precisely and that is what we are working on now.”

For example, the new computer helped calculate the chance of heavy showers in the east of the country at the end of July, says KNMI meteorologist Rob Groenland. “But we did not anticipate that it would involve so much precipitation. We expect to be able to improve the models in the near future, but calculating the chance of the location and time of such heavy weather is the most difficult.”

Message on your phone

According to the law, code yellow may only be issued 48 hours in advance and code orange and red only 24 hours in advance. “That is determined by politics, we have no say in that,” Groenland adds, “but we can help people more to be less vulnerable to expected weather extremes.”

That is why the KNMI is working on a major update of its own weather app in September. Users will then receive a push message on their phone if a warning is issued in the specified province, and that is not all, says Groenland. “Equally important is that with the renewed app, together with our partners such as Rijkswaterstaat and the Safety Regions, we let people know what they can do themselves if an extreme weather situation is approaching.”

Concrete warnings

“People don’t necessarily want to know how many millimeters of rain are going to fall, or what the wind speed is going to be, but what the expected impact is,” Greenland thinks. “What is the expected weather going to do on the ground?”

That is why the KNMI wants to use the app to give people practical instructions. “We will work with concrete warnings in the event of extreme heat or extreme precipitation: the chance of flying debris or fallen trees, train traffic that is disrupted, or the message not to go on the road, to stay home, to take water with you, etc. In this way we hope that the damage on a personal level remains limited.”

embed Place this item on your own site

A new supercomputer is helping the KNMI to calculate short-term weather forecasts with increasing precision. This will also allow warnings for extreme weather to be issued in increasingly more detail per region in the future.

Questions? Ask!

Do you have any questions or would you like to respond? Then send us a message in our chat here. Every Thursday we will tell you in the Do-me newsletter what we do with all the responses. Do you want that in your mail? Then sign up here.

Share:

Facebook
Twitter
Pinterest
LinkedIn

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.