Text from you tube "A simulation of draining the world's oceans by pulling a hypothetical "plug" in the Marianas Trench, inspired by the work of Randall Munroe of XKCD: https://what-if.xkcd.com/53/ "
Text from this reddit string of posts "A few years ago I bought the book What If? by Randall Munroe of XKCD fame, which included a map showing what would be left of the world's oceans if you pulled a 10m plug in the Marianas Trench. He did a post about it online here as well. I decided to build a model (likely similar to what he built, given the similarity of the final result) that would allow me to animate it.
I used the ETOPO1 Global Relief Model,
which has a pixel size of about 1.8km across. It's the equivalent of a
233MP photograph in size and includes of the earth's elevation and
bathymetry (i.e. ocean elevation). I also used the Natural Earth Oceans
and Lakes + Reservoirs data sets from here. Much of this data was combined or converted between vector and raster using various GDAL tools, whose scripts you can see here.
Most of the processing was done using GeoTrellis in Scala. Full source is here. The key was a scanline flood fill algorithm (a fancy term for the paint bucket tool you might have used in MS Paint) that was adapted from Lode Vandevenne's work, which was modified to support wrapping at the edges so that it could be used on a globe. It's here in the source.
The math for calculating the time is based on a simplified version of the Bernoulli incompressible flow equation. Basically, the volume of the water that drained divided by the flow rate calculated from this equation. It can be seen here. It slows down near the end because each frame represents a 10m drop in the ocean, which takes less time as the volume of accessible water decreases.
If you're interested in a less brief and easier to understand explanation of all this, let me know and I'll put a blog post together covering the technical side in a more digestible way.
Also, for those asking:
Most of the processing was done using GeoTrellis in Scala. Full source is here. The key was a scanline flood fill algorithm (a fancy term for the paint bucket tool you might have used in MS Paint) that was adapted from Lode Vandevenne's work, which was modified to support wrapping at the edges so that it could be used on a globe. It's here in the source.
The math for calculating the time is based on a simplified version of the Bernoulli incompressible flow equation. Basically, the volume of the water that drained divided by the flow rate calculated from this equation. It can be seen here. It slows down near the end because each frame represents a 10m drop in the ocean, which takes less time as the volume of accessible water decreases.
If you're interested in a less brief and easier to understand explanation of all this, let me know and I'll put a blog post together covering the technical side in a more digestible way.
Also, for those asking:
- Here's a video version to make it easier to view for some
- It's going to Mars
- Other oceans aren't necessarily deeper than the Pacific, they just get landlocked and stop draining
- Many lakes will likely stick around since they are fed from rivers and direct rainfall
- The continents drift, but not that much, over this time span
- The plug is 10m in diameter (that's metres, for the Americans)
- No, there's not actually a plug
No comments:
Post a Comment