[X] My Assistant

[Bjorn Jonsson]

This is image PJ39_16 (northern circumpolar cyclones) in approximately true color/contrast and enhanced versions:

[Bjorn Jonsson]

The PJ39 images are really spectacular. Highly photogenic areas are now imaged at considerably higher resolution than earlier in the mission. This is processed from image PJ39_23:



Lots of cloud shadows are visible. The solar elevation angle near the center is ~30 degrees.

I recently made several improvements to my JunoCam processing pipeline following fairly extensive photometry from various JunoCam images and various experimentation with photometric parameters and photometric functions. The result is that the enhanced images now require considerably less post-processing and should be slightly better, especially at high latitudes. It turns out that it makes a big difference to use photometric parameters that vary as a function of latitude. Photometrically, the polar regions seem to differ significantly from areas closer to the equator. This is not unexpected (and has been known for a long time) but seems to make a bigger difference than I was expecting.

[Brian Swift]

I recently made several improvements to my JunoCam processing pipeline following fairly extensive photometry from various JunoCam images and various experimentation with photometric parameters and photometric functions. The result is that the enhanced images now require considerably less post-processing and should be slightly better, especially at high latitudes. It turns out that it makes a big difference to use photometric parameters that vary as a function of latitude. Photometrically, the polar regions seem to differ significantly from areas closer to the equator. This is not unexpected (and has been known for a long time) but seems to make a bigger difference than I was expecting.

Nice. Care to elaborate? Did you develop your own model or use a standard model? Have any pointers to discussion/analysis/research on the latitude variation?

At some point, knowing that Jupiter isn't lambertian, I started looking at creating a model based on computer graphics folks' subsurface scattering and volumetric rendering models, but didn't get too far, and currently just use a low degree polynomial fit to brightness.

[Bjorn Jonsson]

Nice. Care to elaborate? Did you develop your own model or use a standard model? Have any pointers to discussion/analysis/research on the latitude variation?

At some point, knowing that Jupiter isn't lambertian, I started looking at creating a model based on computer graphics folks' subsurface scattering and volumetric rendering models, but didn't get too far, and currently just use a low degree polynomial fit to brightness.

This is all still highly experimental. Lambert works well in some cases and in some cases not. Rather unexpectedly to me, I have found that combining slightly modified versions of Lambert and Lommel-Seeliger works rather well - this surprised me somewhat since I had assumed (maybe incorrectly) that Lommel-Seeliger was mainly used for solid bodies (e.g. the Moon). I experimented a bit with this a few years ago, then reverted back to modified Lambert combined with a limb darkening function but now I'm back to using Lambert/Lommel-Seeliger and a limb darkening function. I also experimented with a Minnaert function but it didn't work very well and much worse than it does when processing e.g. Voyager images. I am not quite sure why but possibly it is because the Voyager images are narrow angle images and the phase angle is (almost) uniform across the images. In contrast, the phase angle near e.g. the left edge of a JunoCam image can differ quite a bit from the phase angle near the right edge. It might be possible to correct this with a phase function (e.g. Henyey-Greenstein) and I've been experimenting a bit with that.

I still haven't experimented with polynomials but that's something I probably really should do - it might work even better than what I'm currently using. It certainly seems to work very well in e.g. Gerald's images except for very close to the limb.

I'm familiar with the Mendikoa et al. paper mentioned above but each set of photometric parameters I am using covers a much bigger latitude range than 1 degree. I'm currently using 6 sets, one for the equatorial region, three in the northern hemisphere and two in the southern hemisphere. The biggest change between parameters sets probably occurs near latitude 65 degrees N/S. The number of parameter sets will probably increase. In particular, the number will increase to three in the southern hemisphere once I start processing southern hemisphere PJ39 images. Which reminds me: Mike, do you know if there have been problems with the PJ39 downlink? No images of the equatorial and southern latitudes have appeared at the missionjuno website (I'm assuming the imaging was similar to what it has been for recent perijoves).

[mcaplinger]

Mike, do you know if there have been problems with the PJ39 downlink?

I'm not authorized to discuss problems. You can look at DSN Now or https://twitter.com/dsn_status to see what Juno downlink passes have looked like recently. The Earth-Jupiter range over 5.8 AU right now. As a general rule, we put up the images shortly after we receive them in their entirety.