Femmy's student experience

predicting water levels

Femmy is working on her module 3 final project

Predict the water levels in the Bocht van Walsoorden (Zealand), for tomorrow, a week from now and two weeks from now. When will the water level be 1.5 metres above Amsterdam Ordnance Datum (NAP in Dutch)? That was the assignment our teacher gave us. An interesting challenge at first sight, but how do you actually do it? Which factors affect those water levels, and to what extent?

We started by looking up which factors influence water levels in general. We found quite a few, but many of them were not useful, because their precise impact was unclear or too small. The information exchange within our group was less than perfect as well: we often shared information without listening to each other. This came up during a course we were following on teamwork. We were all given different pieces of information, along with a problem we had to solve together. First, we all talked at the same time, trying to share the different bits of information we had – but of course that didn't work. After that, when we had analyzed what the actual question was and identified which information we would need to solve it – and when everyone started to listen to each other – we came up with a solution pretty quickly. It goes to show that thinking about what information is necessary works better than just starting randomly and seeing where you end up.

Our project took off slowly. The main reason was that we were still learning the mathematics needed to make an accurate water level prediction. It was only after we had mastered Fourier Transforms that we could really get to work. We then used the MATLAB programme to make our predictions. On the basis of previous water levels, our function was able to predict the next two weeks.

Once we had what seemed like a reasonable prediction, it was time to find out whether it was as good as we thought it was. That’s where our Probability class came in. Unfortunately, it turned out our prediction was not actually that good: at some points, the gap between our prediction and the actual water level was pretty big. Thankfully we soon realized what the problem was. Our prediction focused mainly on the hours at which the water exceeded 1.5 metres, rather than on the water heights at the maximum (flood) and minimum (ebb) points. After a lot of puzzling, we finally resolved the issue. We finished off the project with a presentation on how we carried out the project, our prediction and the quality of the prediction.

Because of the way the project was structured (forcing you to immediately apply what you have learned), you gain a better understanding of the subject matter and you get a taste of its possibilities in practice. It makes the subjects a lot more tangible and funner.