Here is an intermediate breadcrumb of one of the topics I'm currently working on, that's analysing the dynamics of Jovian weather systems on the basis of JunoCam image pairs:
Click to view attachmentThis is an excerpt of an analysis of pixel displacement fields, here applied to PJ17 Oval BA.
The upper left image is describing a steady vector potential, based on the two maps in the bottom row. This can be interpreted as the solenoid (divergence-free) component of a steady 2D-flow. The upper right is the amount of the first derivative of the vector potential, hence describes kind of a velocity. The other tiles try to visualize the vector potential and the derived velocity field.
Some disclaimers: I'm going to translate this into physically meaningful units. The velocity maps will look a little different after considering map distortions and proper scaling. So, be careful, when trying to interprete these intermediate products. The results are also subject to various systematic and statistical effects, which need to be specified and quantified for any scientifically meaningful interpretation. Nothing of this is substantially peer-reviewed. All errors are mine.
The answer to the frequently asked question of which software I'm using: A C++ compiler. Almost everything is proprietary and implemented from scratch on the basis of C-standard libraries like stdio.h or math.h. Runtime for the small example above was on the order of 100 CPU core hours. (I know, that GPUs exist. But I'm ready to write shader code only for simple algorithms.) For access to SPICE kernels, I'm saving s/c trajectory position data to text files using the NAIF/SPICE utility spy.exe.