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

2018 Conference on Severe Local Storms

Surrounded by my people!

I haven't been to SLS since my very first time back in 2012, and honestly, I had completely forgotten about how invigorating the entire week is when you're surrounded by people excited by the same stuff you are. And then, at the end of the day after you've heard about all of the cool stuff going on, the conversations continue into the early morning hours with your friends and peers. It's such a cool experience, and I don't know anything else quite like it. While at the end of this week I am mentally and physically exhausted, and frankly, ready to go home, I can't quite help but feel a little sad that something like this is hard to come by. I suppose that leaves something more to look forward to.

Leigh Orf and the Giant Death Banana

As part of this year's opening session, Leigh closed things out by showing everyone some of his findings with our temporally averaged data that shows some unique features in the low-level updraft. Well... this resulted in the birth of the "Death Banana". He told us he wouldn't actually say that in public, but... here we are...
In addition to that, Leigh also teased some of the results from the simulation initialized off of the April 27 2011 JAN sounding that produced an EF-4 tornado in a 30 meter simulation. That's right, we've got 2 long-track, significant tornado simulations. Consider me pumped.
If you haven't already seen it, the talk is uploaded to his YouTube channel and is definitely worth a watch! The talk with the bulk of the April 27 2011 simulation belongs to Dr. Cathy Finley, and will not be up until the recorded talks are posted to the AMS website. I'll more than likely share that here when it is available.

The GPU Parcel Code Does Its Job

All of my efforts prior to this conference were focused on finishing this GPU parcel trajectory code that's gradually been discussed since February through multiple blog posts. Once we got our glorious card from NVIDIA, development took off exponentially and the testing and validation began. I'll gloss over the details for now, but we were able to verify the integrity of the calculations through some control simulations of a supercell at lower resolutions, and the parcel code works. Below is a brief example of a run where I dropped a "log" of parcels (50k parcels) into the SVC and let it integrate for about 2 minutes of simulation time (720 time steps). The great thing about this code, and the reason we've spent so much time on it, is this process only takes about 10-20 minutes and then dumps it all to a file. No re-running of CM1 required to get the parcel trajectories integrated at the model time step, on the staggered mesh. I'm pretty pumped about that.

Some Quantitative Analysis

The great thing about this conference is that we've finally been able to present at least the beginnings of some quantitative work being done with these simulations. I do want to stress that it's only the beginnings, and there really isn't much in the way of results quite yet, but a story is beginning to unfold in the data. I guess the judges of the Student Poster Contest must have agreed, because I was one of the poster award recipients for the conference!

A lot of literature talks about the tilting of horizontal vorticity into the vertical and subsequent stretching, but it's beginning to become more clear that a lot of parcels that end up in the tornado or the low level mesocyclone undergo long periods of horizontal stretching of horizontal vorticity. We wanted to quantitatively show this through the use of the vorticity tendency equation, and this served as the basis for the poster I presented at SLS. I've included the core figures from the poster below, but I'll also post a link to the full poster.
The top row is the streamwise vorticity magnitude, the middle row is the horizontal stretching rate of horizontal vorticity along the X axis for the first two and Y axis for the last, and the bottom row is the tilting rate for horizontal vorticity into the vertical axis. For many parts of the cold pool, horizontal streamwise vorticity is increasing through horizontal stretching for 3-4 km of path length before any sort of vertical velocity is encountered. That's pretty interesting.
The full poster can be found here!

Looking Forward

So now what? What does it even matter that horizontal streamwise vorticity is increasing through stretching? That's kind of what we're asking ourselves too. It's come to our attention that this doesn't necessarily tell us about where the streamwise vorticity originates or how it's created, what the physical processes are that make it even matter for the storm, or what theoretical frameworks we can exploit to gain information on dynamic balances and imbalances in the flow. Is vorticity even the right framework to view this through? There are a lot of questions going forward, but I'm most excited to start looking at cases where tornadogenesis failed, because that's where we might be able to start teasing out some patterns. We'll see.