Chilean earthquake: impact of the tsunami

The National Oceanic and Atmospheric Administration (NOAA) has a page with some interesting information about last week's earthquake in Chile, but what really stood out for me was this chart of the predicted wave heights around the globe resulting from the associated tsunami:

 

Click to enlarge: it's a fascinating chart. Although labelled a forecast, from the explanations on the page it appears to be based on observed wave heights at various monitoring stations (with model-based interpolations between them, I assume). This really was a hemispheric event, with impacts around the entire Pacific basin. Clearly though, the impact on the Chilean coast was extreme. Unfortunately NOAA doesn't have a similar chart for the devastating Boxing Day 2004 tsunami; it would be interesting to compare them. (By the way, although you can easily create charts like this in R, I'm not sure whether R was used for this one or not.)

West Coast/Alaska Tsunami Warning Center, NOAA/NWS: Offshore Maule, Chile Tsunami of 27 February 2010

Comments

[Neil]

I'm also unsure as to whether the chart is observation or prediction. I saw the same chart at several news websites (e.g. ABC News Australia) before the tsunami waves had reached many of the areas on the map.

[Nicholas]

I hope people don't make more charts like this in R. The colors of the scale are not visually discernible beyond twenty. It is one of those horrible rainbow scales, that do not convey an increase in magnitude well, ...., yech.

But one can make good maps in R, look at the spatial ctv.
http://cran.r-project.org/web/views/Spatial.html

[David Smith]

Not sure I agree, Nicholas. It seems likely that the colour-scale on that chart was histogram-normalised, good practice for showing detail at all scales. With a linear scale, you'd lose all the interesting detail currently shown in the yellows and greens. On the other hand, it does obscure what may be the most important feature: truly devastating waves (Chilen coast) versus merely large waves (Alaskan coast). But it all depends what the purpose of the graph was: to track the progression of the waves across the globe, or to identify the critical spots. This graph is clearly more suited to the former purpose.

[Nicholas]

Hmm, log2 intervals would probably linearize things better, even if labels are at the original scale, they should not be equally spaced. Also the scale is one showing increasing magnitude, I don't like the rainbow scale for that. Cartographer's have been complaining about that for a long time. But I think we agree one can do really cool figures in R, we are just arguing about execution.

[David Smith]

Yes, on that we can definitely agree :)