How we measure SST – a graphical thingy

For years, I had a slide that I used to explain the wide range of ways in which SST was measured. It was animated in a very simplistic way, but it always went down well, or at least better than the other slides.

A video of a powerpoint slide what I made to show how we measure sea surface temperature.

I also had this one, which is static and often preferred – inexplicably – by people who had to give presentations on my behalf. It is more sober and actually covers a wider range of methods.

However, I direct your attention to the bottom right of the image. In the metadata these are listed as APBTs – Autonomous Pinniped BathyThermographs. Not a problem, except I put this slide in a presentation for my boss’s boss and forgot about it. Some time later, he wandered casually over to my desk and asked me what an Autonomous Pinniped BathyThermograph was.

“They stick sensor packets on the heads of walruses and things of that kind”, I told him, wondering vaguely why, out of all the people he might have asked, he chose to ask me.

So, when he was giving the presentation to my boss’s boss’s boss’s boss’s boss’s boss, he indicated the APBT and said, “and walruses”. Unfortunately for him, there was a marine biologist in the room who noted that while there might be many things of interest in the photo, a walrus wasn’t one of them. After which, I received a stern lecture on the importance of taxonomic precision and I put the slide in storage never to be used again.

I needed a replacement. I wanted something that had the artistic coherence of the animated slide and the diversity of the static slide. I was also playing around with Blender at the time. If you’ve not heard of it, Blender is a free 3D modelling, animation and rendering program. I’d downloaded it with a bunch of other software to my laptop to stave off boredom on a long internet-less break from work deep in the French countryside. I found it to be completely incomprehensible¹ and I would have abandoned it completely², but by chance the local supermarche had a book “introduction to blender” which taught me just enough to stay interested.

The incomprehensible interface of blender.

In many ways blender is perfect. I can happily spend hundreds of hours painstakingly modelling individual items and making them go woosh! Eight years ago, it was already stuffed with features (the ocean in the image above was created with a few clicks, for example). Now it’s considerably more accessible and the range of features is expanding rapidly. You can also script addons in Python and C++ and it has its own inbuilt graph-based modular programming system³, so it can in principle do almost anything.

Anyway, I wanted an animated slide, which bought together lots of different ways of measuring SST. Unfortunately, my computer at the time couldn’t deal with an animation, so I ended up with this:

Rendering it still took a few hours on my lousy desktop machine and kept my office warm for a few days while I got it looking approximately right. I’ve added to it as people have made requests.

The satellite model in the background was from NASA who have a nice archive of 3D models. The seal was one of a pair of comedy seals. The TAO buoy, drifting buoy and Argo float I modelled based on technical drawings I found online. The giant foram on the righthand side of the image, just above the bucket was from was from this extraordinary collection. The lighthouse is modelled after the Tybee Island lighthouse, GA. The rubber duck is a rubber duck.

-fin-

1. I later discovered that it is famously incomprehensible. Once you get used to it though, you feel like a wizard. ↩︎
2. Probably not. When I was a teenager, I got a copy of POV ray on a disk that I mail ordered from an ad in the back of a magazine. It was a ray tracing programme that would take a text file as input and, after several hours of computation, spit out an image. The input file had to be completely correct. The slightest error would cause the program to give up without saying why. Even if the format was right, lots of things like working out which way to point a camera, or how close together items needed to be for the light sources to illuminate them required an iterative approach. The documentation was negligible so I had to work lots of it out by trial and error. ↩︎
3. It works by connecting nodes together. Nodes perform a variety of tasks. Having programmed for so long in the regular way, I find the node approach weirdly counter-intuitive. Nonetheless, it’s hyper-optimised so it’s often worth the hassle to do it this way rather than in python. ↩︎