One size fits all?

The resolution of your model - finding the right balance

We sometimes get the question "How should I set the resolution of my calculation grid?"

As a general rule, simply press the "Set to map resolution" button to set x and y spacing to the same resolution as the imported bathymetry, but remember that there is no information in the bathymetry file regarding the depth resolution of your imported bathymetry, so you'll have to set that yourself.

Sometimes, however, the imported bathymetry can be very large and you might end up with calculation grids that are impractically large (>2 GB in Figure 1, even before results are stored).

Figure 1. Three different resolutions for a sample scenario. The lefternmost will take up very little memory, while the one on the right has been set to match the resolution of the imported bathymetry. All have 1m depth resolution.

While it's tempting to "go big or go home" in the quest for increased accuracy, it often does not make sense to do that, given the sharp increase in resource costs and lack of increased accuracy.

Always keep in mind the uncertainties related to your input as well as the expected variability in environmental conditions, leading to limited precision in your modelling. Simply cranking up the settings might produce "nice" looking, smooth outputs, but it's arguably better to reflect the quality of the input (or lack thereof) in the output.

Further; remember the scale of the real scenario that the results are to be applied to. Yes, you can set the calculation to happen over one-meter increments, but if the scenario covers e.g. 10x10 km the detail will be lost.

Figure 2. Solution for the three scenarios from Figure 1. notice how the two results on the right look very similar, while the lefternmost one looks very "rough".

In the above scenario (Figure 2), I have modelled the same source, but with three different resolutions (actual resolution of the bathymetry, the calculation and the results grid). Note that the range step is set to equal the number of points in the scenario diagonal. 

To make the comparison of the different models easier, I have exported the results to a GIS package and turned the levels into curves of equal level (Figure 3. below).

Figure 3. Note the relative similarity between the middle and right solution, while the left (smallest) solution stands out as being coarse and with higher levels.

Back in the real world, where we're interested in assessing the impact of a noisy activity in a large environment, the small difference between the medium and large scenario doesn't matter, while the increased accuracy from the "Very low resolution" model to the "medium resolution" offers a huge leap in assessment quality.

From doing many of these types of comparisons I now tell you to keep your resolution between 200 and 300 points, and if you find you need more spatial detail, crop you bathymetry smaller and make a (spatially) smaller model.

Making smaller models has the added benefit of being much faster to re-run for varying environmental factors (e.g. tides, salinity or animal presence).