Use case On track for gold

This summer, some 10,500 athletes from 204 countries will compete in Tokyo for 962 Olympic medals. A number of those medals will be awarded in a discipline where the Netherlands traditionally belongs to the contenders; sailing. In order not to leave anything to chance, the TU Delft is working hard on models that can accurately predict from which angle a golden wind will blow.

The sailing races are held in the Bay of Sagami. Due to the unique topographical conditions (a complex landscape and the proximity to the island of Oshima), the meteorological characteristics of this bay are quite complex and not well documented. The research team is therefore mapping the wind conditions of the bay and is also developing a forecasting system that will provide the Dutch sailing team with accurate and reliable wind data prior to each race.

"TU Delft has a contract with the Sailing Innovation Centre, a foundation affiliated with the Watersportverbond," says Wouter Stiphout. "They are looking for a kind of computerized chessboard for the sailors, with which you can better predict the local weather by adding or removing different elements and circumstances. Sailing athletes, however, are not only interested in predictions, but also in recognizable patterns. For this purpose, we use extremely high-resolution GPU-based models, which we have housed at Professor Harm Jonker's company WHIFFLE in Delft".

Real-time forecasts

"Since the advent of computers, work has been done on weather forecasts," Sukanta Basu adds, "especially to predict the extremes in the weather. Wind has always been a neglected element. Who is actually interested to know if tomorrow will be wind force 2 or 3? And from which direction will it come? When sailing, this type of information is of course very important. The prediction of wind in a small area, surrounded by complex topographical factors around the coast, is particularly challenging. Moreover, we work with models that look at a high resolution of 100 metres. Of course, this requires extra computer time and power." By way of comparison: a weather forecast at KNMI has a resolution of 3 to 25 kilometres. The finer the resolution, the more accurate the weather forecast becomes.

Publicly accessible (ERA5) data is used in the research. This data goes back to 1979 and provides hourly estimates of a large number of atmospheric, terrestrial and oceanic climate variables. The data cover the earth on a grid of 30 km and view the atmosphere using 137 levels from the surface to an altitude of 80 km. This data forms the basis, but will be refined and extended to a higher resolution in Delft so that the athletes will soon be able to develop a tactic and strategy for every conceivable pattern on Olympic waters.

Sukanta Basu: "The tricky thing is that Japan has only one local weather station that releases additional data into the public domain. That's why the people at the Sailing Innovation Centre also measure in and around the area where they sail".

"Many different countries," Basu continues, "have published similar research results after the previous games. It began around Sydney in 2000. We learn from each other's literature, so we don't make the same mistakes. And, of course, we are not the only country preparing for Japan. The difference lies in the approach that is chosen for each country and which facilities are used. Thanks to the resources at SURF, we can hopefully distinguish ourselves. The closer we can get the desired data for our simulations to the starting times, the better and more accurate they will be. So we have asked SURF for dedicated support - without waiting times - for our real-time forecasts during the Olympic Games. SURF is also helping to analyse the data afterwards in order to clarify which parameters worked best for the various scenarios.

Image: complex flow pattern around Sagami Bay on August 18, 2019 around 5 am (local time). Strong circulation and coastal wind gradients are visible. Also, the blocking effect of the Oshima island can be seen near the south west corner of the image. This real-time wind forecast was performed with the Weather Research and Forecasting (WRF) model.

Douwe Broekens: "With the help of TU Delft, we are trying to stay ahead of the competition in Japan. The tricky part is that it's all about forecasts and we have to deliver top performance in a small area within an hour. So how do you interpret the forecasts and which characteristics are most important in this respect?

"Circumstances in Tokyo are very different from those of the previous games in Rio. It's probably even warmer in Japan. Last year we had a test event at 40 degrees and 80 percent humidity. There are mountains in the area, the water is warmer and in the vicinity are some under water volcanoes. All these factors influence the weather. Typhoons are not uncommon in the region. Of course there will be no sailing when that happens, but we will even go as far as to investigate the effect of those typhoons before or after the Olympic tournament".

"As a top athlete, when you have a lot of talent and experience, you can sail very fast in Japan", concludes Broekens, "but if - during frequently changing weather situations - you get the wrong information, you can also sail very fast in the wrong direction. Then you miss out because, for example, the wind has unexpectedly turned. In short, having the availability of reliable forecasts will literally be as valuable as gold".

Text: Edwin Ammerlaan
Photo: Richard Langdon