Pluto Atmospheric Observations: NH Post-Encounter Phase, 1 Aug 2015- TBD Options

[scalbers]

Pretty dramatic flyby from Bjorn 2 posts back. Interesting to see the asymmetry in the atmospheric ring start to increase near the end as the sun pulls away. I was able to turn up my monitor brightness enough to see the Charonshine.

[Gennady Ionov]

Here is an animation I did of the New Horizons flyby that includes atmospheric effects:
https://vimeo.com/136223988
The scale height of Pluto's atmosphere is ~60 km according to pre-NH measurements. I'm using a scale height of 55 km for the aerosols which may or may not be correct. I tested other values; 30 km was visually different from the NH high-phase images and it was also not possible to increase the values to much more than 60 km without getting bad results.

Nice video!
If you make time of Charon pass more slowly, it will look spectacular.
I'm probably wrong somewhere, but I have obtained similar to NH haze images at a scale of less than 30 km...
Yes! My mistake has a copy/paste nature. Haze scale variable does not used at all. Instead Rayleigh scale variable are used.
Its fit value is 48 km:

[Bjorn Jonsson]

I'm probably wrong somewhere, but I have obtained similar to NH haze images at a scale of less than 30 km...

I should mention that my NH image measurements were rather crude and other things in the atmospheric models might be different (and I also cannot completely rule out subtle bugs in my code ). Also the haze/aerosols probably do not decrease uniformly with height, there could be one or more discrete layers of haze and the horizontal distribution of the haze could be non-uniform.

[ZLD]

Bravo Bjorn! The animation is really great and helps put some things in a better perspective, namely the later shots.

[fredk]

Here is an animation I did of the New Horizons flyby that includes atmospheric effects

Beautiful.

You say Charonshine is exaggerated - by what factor? And is the (far) post-encounter Mie scattered light intensity correct relative to the pre-encounter directly sunlit surface of Pluto? Ie, does the simulation use a constant "exposure" for each frame?

I had the impression that the scattered light must be much fainter than the sunlit surface, but now that I look at the LORRI exposures I see that they are comparable, so that means that the scattered light is remarkably bright.

Anyway, this leads me to wonder what the sky would look like from the surface (as others have speculated here already). But what I'm thinking about is the sky near the sun, which would correspond to Mie scattering at large phase angles, where we have a good handle on the phase function from LORRI. We now know that the sky near the sun would be quite bright - in particular, near sunrise/set (optical depth half what LORRI saw) the sky intensity would be comparable to that of the sunlit surface.

Of course our large uncertainty in phase function at smaller phase angles means we wouldn't be very certain how large the glow around the sun would be, and how faint it would get at large angles (eg 90 deg) from the sun. But pick a phase function out of the blue and you could render it...

[Herobrine]

I took the code I wrote to get the value vs radius combined plots of the 2 full-disk backlit LORRI shots and applied it to the 3 partial-disk backlit LORRI shots (lor_029920671, lor_0299206715, lor_0299206716). These are much closer but also much lower quality.
Data from 550,016 pixels across the three images are plotted here. This covers 8.6 times the number of pixels from the other one I posted, and at a much better spatial resolution, so the plot is much richer.

The horizontal axis is the pixel's distance from Pluto's center (left edge is 600 pixels; right edge is 740 pixels).
The vertical axis is the pixel's sample value (bottom edge is 88; top edge is (just below) 256). The sample values of the first two LORRI shots were scaled to match the brightness scaling of lor_0299206716.
All apparent large-scale contours/concentrations apparent here were also apparent in each frame's data individually; they were very consistent.
Artefacts near the bottom are from value-stepping in the original low-quality JPEG data.

Here's the same data plotted the same way, but with each point rendered with a hue based on the pixel's angle from the center of Pluto.

I had to write a different program to make this, and didn't bother with subpixel rendering. Instead, each pixel's data was written into a 560x256 array of lists (2-D binned, in other words) and each list of data was averaged to yield the pixel value. It was then scaled to 280x256 and cropped.

While I was at it, I took some of my new code and rewrote a much better atmosphere unwrapper that's binned and processes each pixel in the images exactly once and so isn't susceptible to a lot of the artefacts that turn up with resampling. I also applied an array column-averaged correction to the sample values of the radius vs angle data before rendering to compensate for the uneven lighting around the disk.

The bottom edge is the surface of the planet; the top edge is 78 pixels above the surface.
The left edge is an angle of 178 degrees; the right edge is an angle of 288 degrees (I might have that backwards...and you might need to subtract them from 360...)
The above version has been contrast-enhanced. The original output is below.

[Bjorn Jonsson]

Beautiful. You say Charonshine is exaggerated - by what factor?

I don't know . I simply added a light source with brightness=0.25 (typically the brightness of light sources is between 0 and 1). If I remember correctly, Gennady worked out the approximate true brightness of Charonshine a few days ago.

And is the (far) post-encounter Mie scattered light intensity correct relative to the pre-encounter directly sunlit surface of Pluto? Ie, does the simulation use a constant "exposure" for each frame?

Yes, I'm using the same set of parameters for almost everything throughout the animation. The only "cheat" is that I altered Pluto's phase function to make it brighter at high phase angles.

I had the impression that the scattered light must be much fainter than the sunlit surface, but now that I look at the LORRI exposures I see that they are comparable, so that means that the scattered light is remarkably bright.

This is true if the JPGs are not contrast stretched (and I don't think they are but I'm not 100% sure). The scattered light is remarkably bright.

Anyway, this leads me to wonder what the sky would look like from the surface (as others have speculated here already).

Here is a quick and dirty test render using the parameters used in the animation. The altitude above Pluto's surface is 200 meters and the field of view is 60 degrees. This should give a very crude idea of what Pluto's sky might look like.

[fredk]

Cool, thanks for that!

About the stretching, you can see clear signs of stretching in some images, eg this Charon image:
http://pluto.jhuapl.edu/soc/Pluto-Encounte...0x632_sci_3.jpg
Not only is banding visible by eye and the black regions quite bright, but the histogram shows discrete pixel values, indicating a stretch of about 4:1:

For this post-enc image:
http://pluto.jhuapl.edu/soc/Pluto-Encounte...0x632_sci_3.jpg
The histogram also indicates stretching, though not as severe (about 3:1):

But for this post-enc image:
http://pluto.jhuapl.edu/soc/Pluto-Encounte...0x630_sci_2.jpg
There is no sign of stretching:

(Though of course jpegging post-stretch could reduce the signs somewhat.)

[Gennady Ionov]

This is true if the JPGs are not contrast stretched (and I don't think they are but I'm not 100% sure). The scattered light is remarkably bright.

We can preliminary estimate stretching with assessment of the background stars magnitude, if they identify ...
I can evaluate the equatorial grid and put it into the frames...
By the way! We can try to measure the parallax of stars in comparison with images from the Earth. But only for stars closer than 50 pc

[Gennady Ionov]

About the stretching, you can see clear signs of stretching in some images, eg this Charon image:
lor_0299175721_0x632_sci_3
Not only is banding visible by eye and the black regions quite bright, but the histogram shows discrete pixel values, indicating a stretch of about 4:1:
For this post-enc image:
lor_0299206714_0x632_sci_3
The histogram also indicates stretching, though not as severe (about 3:1):
But for this post-enc image:
lor_0299323899_0x630_sci_2
There is no sign of stretching:
(Though of course jpegging post-stretch could reduce the signs somewhat.)

In my simulation I used the next gains table (stretching):
lor_0298996664_0x630_sci_2 = 1.5 ....... lor_0298996694_0x630_sci_2 = 1.5 ....... lor_0298996724_0x630_sci_1 = 1.5
lor_0298996974_0x630_sci_1 = 1.0 ....... lor_0298997004_0x630_sci_1 = 1.0 ....... lor_0299075349_0x632_sci_1 = 1.0
lor_0299123689_0x632_sci_3 = 0.85 ...... lor_0299124574_0x632_sci_1 = 0.95 ...... lor_0299135299_0x632_sci_3 = 1.0
lor_0299135484_0x632_sci_3 = 1.0 ....... lor_0299147641_0x632_sci_3 = 1.0 ....... lor_0299148119_0x632_sci_3 = 0.8
lor_0299148167_0x632_sci_3 = 1.0 ....... lor_0299148215_0x632_sci_3 = 0.75 ...... lor_0299148263_0x632_sci_3 = 0.7
lor_0299165499_0x632_sci_3 = 1.0 ....... lor_0299165548_0x632_sci_3 = 1.0 ....... lor_0299165597_0x632_sci_3 = 1.0
lor_0299165646_0x632_sci_3 = 1.0 ....... lor_0299165695_0x632_sci_3 = 1.0 ....... lor_0299165744_0x632_sci_3 = 1.0
lor_0299174665_0x632_sci_3 = 1.0 ....... lor_0299174713_0x632_sci_5 = 0.8 ....... lor_0299174857_0x632_sci_3 = 0.8
lor_0299174905_0x632_sci_8 = 1.0 ....... lor_0299174953_0x632_sci_3 = 1.7 ....... lor_0299175097_0x632_sci_3 = 1.8
lor_0299175145_0x632_sci_7 = 1.3 ....... lor_0299175604_0x632_sci_3 = 4.0 ....... lor_0299175721_0x632_sci_3 = 4.0
lor_0299175838_0x632_sci_3 = 4.0 ....... lor_0299206714_0x632_sci_3 = 3.0 ....... lor_0299206715_0x632_sci_3 = 3.0
lor_0299206716_0x632_sci_3 = 3.0 ....... lor_0299323619_0x630_sci_2 = 1.0 ....... lor_0299323649_0x630_sci_2 = 1.0
lor_0299323899_0x630_sci_2 = 10 ........ lor_0299323929_0x630_sci_2 = 10 ........ lor_0299324179_0x630_sci_2 = 1.0

In the case of lor_0299323899_0x630_sci_2 stretching may have been produced in 12-bit representation, so no effect on the histogram. (Possible there is stretching in 16 times, so 4 bit were cut)
High background noise and at least 7 visible stars favor this supposition.

[Gennady Ionov]

I calculate the average over the angle values of haze brightness in dependance on the height above the Pluto surface on lit side limb (on the illuminated side the haze glow is practically independent from angle).
Such calculation I made for the original LORRI images and for simulated pictures. Also I estimate PSF of LORRI from far Pluto snapshots (it can be verified from the images of Charon and other satellites).
Now I take PSF as radial function exp(-r*r/1.44)+0.0008*exp(-r*r/100) divaded by norm.

You can get the results in
ftp://gionov:NG@46.45.15.20/_Data/_LORRI/Haze/

For example:



For simulate of haze I used scale 48 km, Rayleigh scale is 60 km.

There is definitely scattered light around a bright target like Pluto but I wouldn't completely rule out a very faint layer of aerosols/haze in addition to the scattering. Also one problem with the glow around Pluto in the farther-out images days before closest approach is that it may be caused partially by JPG compression artifacts in addition to scattering. Things really do not become completely clear until we see much higher resolution images of Pluto's limb taken before closest approach (i.e. not at high phase angles). Also let's not forget that a haze layer is visible in images of Triton where the phase angle isn't very high.

It can be seen that only on the distant image lor_0298996664_0x630_sci_2 haze glowing a little less than light scattering inside the optics.
The picture lor_0299148263_0x632_sci_3_Haze shows appear as the effect of scattering near the limb, and haze glowing far away from the limb.

[fredk]

It's not clear that a psf measured from distant images would capture well the actual scattered light in the optical system in the near shots. Which far shots did you use?

Generally, there's going to be a strong degeneracy between optical system scattered light and atmosphere, so I think it would be very hard to reliably distinguish the atmosphere.

[Gennady Ionov]

It's not clear that a psf measured from distant images would capture well the actual scattered light in the optical system in the near shots. Which far shots did you use?

Mainly I used lor_0298996664_0x630_sci_2 and lor_0299148263_0x632_sci_3, and visual compare a bluring effect with original images.

Generally, there's going to be a strong degeneracy between optical system scattered light and atmosphere, so I think it would be very hard to reliably distinguish the atmosphere.

On *714-716, *899, *929 frames optical system scattered light is negligible, so we can get haze scale and optical depth from this images.
Atmospheric effects on the day lit images remains ambiguous only due to the particle phase function.

[fredk]

Yes, absolutely, I meant it will be hard to distinguish the atmosphere on the pre-encounter images.

[Gennady Ionov]

New simulation shows New Horizons as it went into orbit around Pluto https://youtu.be/zB4Ypa3-T_E
Glowing of the atmosphere too much to make it difficult the observation of the Pluto surface at Charonshine...