Posted by therabbithole 3 days ago
I think the main issue is colour and scale. The magnitude is logarithmic, but the scale of the circles are not.
We experience earthquakes in New Zealand all the time. Th last one was 10 hours ago [1] with a total of three yesterday. So in places you would expect large quakes, you also get lots and lots of small ones.
On a side note, Google beta-tested their early earthquake warning system in New Zealand and it was opt-out. I had students diving under desks in a Deep Learning class because the warning sound emitted was very much like the govt. emergency SMS messages. It was a very minor quake, I am not sure we even felt it.
https://itnext.io/using-kepler-gl-to-visualise-over-35-000-e... for the curious
The artifacts change randomly if I zoom and pan around, but it isn't easy to control. For example, in the default global view when opened, it only seems to show quakes in the South Pacific near NZ.
After pointing out that annoyance I hate to mention too that the data could use some cleaning of duplicate entries related to slight differences in focal locations.
One set of triplicates that demonstrates this is located east of Greenland and is visible when parameters are set to display >6.5 magnitude quakes. The event is a 6.8 quake which occurs on 11/8/2018 at 7:49:40 pm UTC for an event at 10 km depth and there are three locations for the same event (one of them has a time stamp two seconds earlier but is unlikely to be a different event of the same magnitude).
I suspect that the display uses information from multiple sources that have not been cleaned for duplicates. I believe it would be difficult to identify all instances of duplication in a dataset like this for quakes of lower magnitudes especially.
Anyway, I like it but I did stumble on or against a couple of things that I thought noteworthy.
Good job overall, the easy part is done and it works and is a useful display.
I think if I had a feature request it would be an option to have a fixed interval(or window) of time visible, rather that the current method of a fixed start time. for example one year behind current. This can sort of be achieved by moving the start and end bugs in sync, but that was less than satisfying in practice. The thing that would make sense is to be able to drag the illuminated part of the time line.
If I had a second feature request it would be to assign what feature(depth, age, magnitude) is mapped to the color axis, nothing wrong with depth here, I just noted that it was redundant in the 3d view.
(Mid-Atlantic Rift Zone.)
Historically virtually all cities were located either along coastlines or major river transport, as shipping was far and away the cheapest way to move large volumes (and masses) of goods. Even today that pattern remains strong.
Tectonic movement is also associated with factors that often produce economically-critical natural resources, from minerals to simply fertilising soil. Australia, which sees little seismic activity, has famously infertile farmland, in which even minuscule additions of mineral fertilisers --- not the Big Three of nitrogen, phosphorus, and potassium, but trace minerals such as iron, copper, selenium, zinc, and manganese. Again, that's where cities tend to form.
And coastlines are strongly associated with earthquakes, especially those along subduction zones (Western Americas, Eastern Asia). Not only do those have many earthquakes, but some of the largest and most destructive, along with tsunamis which can further the devastation.
Note that coasts nearer to rift zones (eastern Americas, western Europe, both west & east Africa) have fewer earthquakes. Several of those also have major populations.