[X] My Assistant

[Nirgal]

among the many new pancams at exploratorium there seem to be interesting
images showing what must be clouds, clearly defined as never seen before
at the Gusev site.

http://qt.exploratorium.edu/mars/spirit/pa...HEP2680L7M1.JPG

Interestingly, those are high-res pancam multi-filter sequences.

Questin for Dan (slinted) refering to our recent discussion about sky color:
would't this be wotrth a try to do interesting sky color composites.
Maybe you could use the brightest pixels
in the clouds as a white point reference ?

[slinted]

Well, I've been working the last couple weeks, trying to come up with a "whole sky image". It seems to me like there are two ways to approach this, both of which have significant problems.

First would be the method most familiar to the what we've been doing with the other MER images, simply mosaicing together enough images taken of the sky on a given sol at a certain time into a single view of the sky. Unfortunately, a full 360 mosaic of the sky would require at least as many frames as the 360 degree mosaics of the ground. Neither rover has come even close to doing this. The greatest coverage of a single time on a specific sol are the sequences similar to what was done by Opportunity on sol 124, namely ~ 8 color images taken over the course of a half hour. Here's a upward looking polar projection of these 8 images (zenith is in the center, north is up, west is left):


As you can see, this isn't nearly enough image coverage to create a complete sky mosaic from just these frames. So, instead of using a single sequence to make the all sky mosaic, I had hoped to combine images taken on different days (hopefully around the same time) into the full sky. Unfortunately, the coverage doesn't exist to do this either, even if time of day were ignored. Here is a projection of *all* the sky images taken by Opportunity, over the first 270 sols. Color wise, it's complete nonsense as it combines images taken at different times of day, but it does show which portions of the sky have been imaged by Pancam and more importantly, which portions have not.

Let me qualify this a bit before I go on. For an image to be included in the above projection, it must have been taken in at least 3 filters all from the same pointing. Some sky images are done in just L2/L7, or just a single filter, and aren't included in that projection, but they don't fill in the gaps completely. Also, navcam coverage will probably fill in the holes in the pancam coverage, but wouldn't do much to help us determine the color at that position.

The holes in the coverage in the southern sky, it seems to me, pose the greatest problem. On sols with relatively little dust in the air, the position at approximately the same altitude as the sun, but opposite azimuth might well be the darkest point in the sky at certain times of the day, and without pancam coverage of that position we don't really know experimentally what color it will be.

So, the mosaicing approach won't be able to come up with a whole sky image, for lack of coverage.

The second approach would be to use a mathematical model of atmospheric scattering and absorption to determine the color and brightness for the whole sky at a given moment in time. As edstrick pointed out in his excellent post in this thread, this is tremendously complicated and requires a VERY thorough understanding of the martian atmosphere in order to render the colors and brightnesses accurately. I had hoped to avoid the models as much as possible, and present a view strictly derived from the images we have of the sky, but it appears that the only way to get the whole sky view will be to use the images (and/or minites data) to come up with the parameters to feed the model, and rely on the model to 'fill in the blanks' in the image coverage. I'm concerned with how consistent the model sky will be vs. those locations which have been directly measured, or even if we have enough data to feed the models for accuracy great enough to put side by side with actual images.

Sadly enough, even with 270 sols of calibrated data, we can't put forward an image and say "here is the whole sky as seen by the rovers", but given the limits of data bandwidth, and the relatively low priority of sky images vs. geologic targets on the ground, I guess we should be happy that we have as many sky images as we do.

[4th rock from th...]

Here's a upward looking polar projection of these 8 images (zenith is in the center, north is up, west is left):

Not bad at all.

Let me see if I can create a gradient in Photoshop that matches those images and can be used to fill the gaps.

Could you please post some more versions of the bottom mosaic? For example, just using the setting sun or high sun images?

[tedstryk]

Would it be possible to reconstruct the missing portions through cloning if that doesn't work?

Not bad at all.

Let me see if I can create a gradient in Photoshop that matches those images and can be used to fill the gaps.

Could you please post some more versions of the bottom mosaic? For example, just using the setting sun or high sun images?

[/quote]

[4th rock from th...]

Would it be possible to reconstruct the missing portions through cloning if that doesn't work?

Here are 3 images extrapolating the full sky from Slinted's images:





I had to use cloning and rotation, as the gradients alone couldn't account for all of the brighness changes.

The 2 last ones are diferent versions of the "all images" information, one with the sun at Zenith, the other with a lower sun and some darkening in the south(?).
It looks as though the haze might make the sky brighter close to the horizon in the opposite direction from the Sun. But no indication of a dark zenith from this data!