Juno, perijove 12, April 1, 2018 Options

[PhilipTerryGraha...]

So, according to the voting page, apparently the spacecraft's being spun around to get the instruments pointed directly at the planet. Io and Ganymede will also be imaged during this pass, as they'll come into view of JunoCam two hours before and twelve hours after closest approach, respectively. For Io, the team are "planning to take two pictures - one exposed nominally and one that over-exposes Io to look for volcanic plumes extending above the surface." The Great Red Spot is also expected to come into view during the spacecraft's departure.



Here's a logo for Perijove 12 I threw together, by the way.

[fredk]

Now I see - that is extremely simple. Thanks, Gerald, for the description. So indeed the crucial assumption is of a steady flow, ie constant flow velocity at each point. Indeed looking at more that two frames would help, in at least two ways. First, by giving you a sense of the time dependence of the flow field f(x,y). This could be fitted to simple polynomials or splines to approximate the time dependence (though substantial extrapolation is still risky). But more crucially making several estimates of the flow field would also mean beating down the noise, which has to be large when you only use two frames.

Are there triple or more frame sets that you could use in practice?

My comment about the div wasn't to say you should make that assumption. I think it would simply be interesting to impose div f = 0 and see what difference it makes. A large difference would mean that this is an important consideration and should be investigated further. So it would simply be a test.

[Gerald]

Yes, thanks! That's indeed the same kind of visualization for the flow lines of Earth's wind field.

Are there triple or more frame sets that you could use in practice?

I guess, that this might be possible closer to the poles. There, we have time series of several images. and may be up to a handfull within a respective series that might be of a sufficient quality to determine velocity fields. During PJ13, for the first time, JunoCam has taken a sequence of TDI 3 images of the south polar region. I think, that those images will be the best basis, thus far, for an attempt to find areas of unsteady flow.

Regarding div, laplace, etc, I've run statistical tests, too, whether the retrieved values are significant, i.e., above noise induced by the choice of Monte Carlo samples used for stereo correspondence. In this repect, all these derived entities appear to deviate significantly from zero, for at least some areas. But there are still statistical errors induced by the structure of the data to be ruled out.

There are various issues to beat down. The first limits I tried to test have been Jupiter's latitude, respectively the time from closest approach. It's the harder to get useful image pairs the closer we get to a perijove. That's due to the very rapid change of perspective. Currently, my limit to get useful results is somewhere near 5 minutes between two images taken from very different perspectives. For the equatorial zone, or later, for the perijove anticipated to migrate towards north, I should push the limits to below two minutes. This requires a very accurate global camera calibration and pointing. Therefore, my primary goal for the next three months is developing families of camera models more suitable for JunoCam calibration than the straightforward Brownian approach.
At the same time, I'll run tests with the south polar PJ13 sequence, in order to be able to provide feedback for planning future flybys (assuming them to take place, and with healthy instruments). One of the tests will include the feasibility of an inference of a displacement field between velocity fields, or at least comparisons between velocity fields derived from image pairs within a sequence of south polar images.
We'll then see, whether or which of those higher-order properties of the velocity field can be determined above noise level.