A Virtual River-in-a-Box = Problems!

If you’ve been to an event and played one of our games, you will have most likely also seen the River-in-a-Box mini-flume. This is a big box of plastic sand, through which water flows, building miniature rivers.


Xuxu setting up River-in-a-Box for Hull’s Freedom Festival in 2015

I’ve been trying to recreate this in a computer model called Caesar-Lisflood – the model that is built into TideBox. Caesar-Lisflood not only simulates water flows, but is designed to simulate the movement of sediment (mud, rocks, stuff like that) to show how geomorphology processes – erosion and deposition – change the landscape. My latest attempt is below, and there are many, many problems with it.

Let me explain why this is an issue. Caesar-Lisflood was designed to simulate changes to large areas (eg, whole river basins) over a long time (more than 1000 years often). If it were an athlete, it would be Mo Farah – lean and quick, keeps a steady pace, and although capable of a sprint when required, it’s there for the long-haul. Trying to use it to simulate the River-in-a-Box is like trying to make Mo Farah compete in the 100 m sprint – he is not optimised to do this in way another athlete, say Usain Bolt, is.

The main problem we have is that this video shows nearly two weeks of processing on the computer. That is slow – slower than the time it is trying to simulate (a few hours’ worth of changes in the River-in-a-Box*). One of the purposes of computer models is that they are much quicker than real-life so the fact this is far slower means, scientifically, it isn’t much use.

There is also instability – you will see areas in the flow which look like a chequer’s board and this is too much water being moved downstream that the physics in the model then immediately moves it back upstream, and this continues, back and forth. It’s a bit like when you were a kid when you ran down a hill and went so fast your feet couldn’t keep up so you tumbled over – we can help the model to stop doing this by instructing it to slow down in certain areas, such as restricting the amount of water it can move from one place to another in one go.

As it is, this is a pretty (yes, it is pretty) rubbish piece of modelling (my fault, not the model’s), but there is potential here. We use flumes, which are like River-in-a-Box but bigger and more advanced, to better understand how landscape change. We use computer models in a similar way, and the physics we learn from the flumes helps us develop the models. The ability to simulate the flume environments in a computer model would be a useful one as we would learn more about how our experiments work, what their weaknesses are, and how we can make them better. This in turn will improve our ability to simulate the real world and, for example, forecast risks like flooding with better accuracy.

I hope to share more of this experiment with you as it develops. Thanks for reading.

Dr Chris Skinner – @floodskinner

*Actually, technically, it is still quicker than real-life. As Caesar-Lisflood is like Mo Farah, to help it out I made the course more like the 10,000 m. All dimensions and times in the model have been multiplied by 100, so for each centimetre in the River-in-a-Box the model is told it is a metre. Likewise, to simulate an hour the model is simulating 100 hours (the video shows more than 28 days of actual simulated time). The only thing not scaled in this way is the size of the sediment, which is kept at 0.0003 m.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s