Severe Storms, Big Data, and Pretty Pictures
Severe Storms, Big Data, and Pretty Pictures

Visualization is Hard


Still compiling...

Yeah, it's been a month since last time. Quiet doesn't necessarily mean nothing has been going on, but things are currently in the less "flashy" stages of getting spun up, like trying to get custom software compiled in order to load subsets of petabyte simulations, or reading papers. Due to my advisor being on travel out of the country for some conferences, along with some architecture issues with compiling the software on Blue Waters, I haven't been able to load any meaningful volumetric data until just this last week. Thankfully those issues have been resolved, and we can finally look at what a tornado looks like in one of these simulations! It's not exactly ground breaking science just yet, but you've gotta start somewhere... and generally being able to see your data is a decent enough start.
Extended range view of tornado and mesocyclone, looking due West parallel to the forward flank
Close-up of the tornado, with an anticyclonic funnel visible to the left

Tools of the Trade

While half the battle has been playing the waiting game on getting the visualization tools compiled and working, when it comes to addressing actual research topics and methodologies, it requires a different set of tools. Now, I don't know how most people pick their research topics for their Masters/PhDs, but I've decided to take what feels like a slightly less intuitive approach on picking a topic. Instead of immediately trying to address a scientific question or knowledge-hole, I've been trying to direct my journal-reading more on learning analysis techniques with the idea of building up a toolbox of methods for approaching scientific questions first. Frankly, I have no idea how you even approach a scientific question or come up with an experiment when you're dealing with this much data. However, so far this approach has yielded some fruitful ideas, such as this paper on a technique for extracting jet core lines in three dimensions. Next up on the reading list (thanks Greg!) is using Potential Vorticity analysis techniques on supercell thunderstorms, which could also yield some potentially interesting methods using PV invertability. If anyone out there happens to have some cool papers that involve vorticity budgets, trajectory analysis, or anything involving 3D/4D feature detection, please send them my way!

A Parting Gift...

If you've ever wondered what it might potentially look like to look directly up at a tornado from underneath, but don't want to risk, you know, death... this is for you! Please, don't try this at home! I got to witness this in real life in Louisiana earlier this year, and I do not recommend ever trying it :).
Tornado from below... don't try this at home, kids!