Europa's Color Options

[Bjorn Jonsson]

I have been trying to make (true ?) color images of Europa using mainly Galileo data. I am using red (IR756 in some cases), green and violet images. The violet images have considerably more contrast than the green images. Using Voyager images as a crude guide it seems the blue images have a contrast somewhere between the contrast in the green and violet images. Because of this I also made synthetic blue images as B=(G+V)/2.

I am attaching two preliminary versions of my images, one image using G1 data (IR756, G and V) and a C10 image (R, G, V). Images using V and synthetic blue are included. I should be able to get more realistic results in the future, the color balance is just a crude guess and it would be more accurate to use synthetic red instead of IR756. The red images have a bit higher contrast than the IR756 images so using synthetic red would be more realistic. Also it is probably not very accurate to use the exact average of green and violet as blue. A weighted average should be more accurate.

I eventually plan to use this to colorize a huge (roughly 10000x5000 pixels) global grayscale map of Europa I am working on. When complete it should be of slightly higher quality in terms of coverage than the USGS map. To colorize I need to remove limb darkening etc. because it varies with wavelength, making the color 'change' towards the limb. My biggest problem is that I do not know the wavelength-dependent photometric parameters for Europa. Does anyone know if Hapke parameters as a function of wavelength (or Galileo filter) are available somewhere ?

BTW I also did something similar (synthetic blue) for images of Io, see http://www.mmedia.is/bjj/3dtest/io/index.html for details.

Attached thumbnail(s)

 

[ugordan]

Somehow the IR-G-V images look more "natural" to me than the ones with a synthetic blue channel. I find the orangish colors more natural than the pinkish ones using a synthetic blue channel.
That preference may have its roots in the Voyager imagery so one comes to expect an orange colored Europa -- not neccessarily its true color at all.

[Big_Gazza]

Err.. this may be stupid question (here it comes...) but if my $150 digital camera can take a true(ish) colour picture in a single take, why can't space probes? (well, at least the modern ones built with digital technology).

(this is where the image-processing experts roll their eyes and curse the ignorant noobs )

[edstrick]

A consumer digicam usually has one CCD, and except for the Foveon CCD now being used in a few cameras, it cannot see all three colors with each hardware pixel. Instead, each pixel has either a red, green or blue single-pixel color filter in front of it. The standard design has a checkerboard pattern with 1/2 of the squares being green pixels and the rest are alternating red and blue.

When the data is read out from the CCD, smart software compares adjacent pixels and local brightness trends and makes a smart educated guess for what the missing two colors's brightness SHOULD be at the location of each pixel. It works. Pretty well. If you're not fussy.

That's why Pro-sumer video cameras often have 3 CCD chips and use special "dichroic" filters to let one color through and reflect another directly to that color's CCD. They're more sensative, since no photons are being absorbed, and each matched set of 3 pixels detects all the red, blue and green light that aren't randomly lost in reflection and transmission.

Spacecraft cameras, like Mariners, Viking Orbiters, Voyagers, Galileo and Cassini use one or two filter-wheels in front of cameras to swap a wide variety of filters in front of the single CCD, which you CAN'T do with a color-mosaic CCD like the ones in a consumer camera.

Some, like the MER camera filters split up the entire spectrum entering the cameras into fairly regularly spaced segments so a full set of narrowband MER filter images adds up to a 11 measurement spectrum of light entering the camera from violet to near infrared.

Galilleo and Cassini cameras also have narrow band filters that are specially designed to measure brightness in a narrow part of the spectrum. In their case, they measure light reflected from Saturn and Titan in bands where methane strongly absorbs, and you can use a set of methane and non-methane band images to measure the height <or depth> of clouds and hazes in the atmosphere. Cassini also has a set of polarizing filters to measure the polarized scattering of light by gasses and cloud/haze particles, which depends strongly on size and shape of cloud particles.

To take "TRUE-color" images, you have to design a set of color filters who's transmission varies with wavelength so that when combined with the transmission of the optics and wavelength dependent response of the camera CCD <or vidicon on old missions> you fairly precisely match the red, blue and green responses of "standard" color vision. To do that, you end up with a set of filters that are less ideal for scientific purposes, like estimating a surface spectrum as in the MER cameras.

MSR is planeed to have "hi-def" camera capabilities in color, and I suspect those detector/filters will be mosaic types, as in consumer cameras and camcorders. Even if color data from them won't be "full scientific grade" in quality, it will still be more useful for navigation and spotting "interesting" targets to look at closer than the monochrome hazcam and navcam images.

[Bill Harris]

Good explanation, Ed. Photography is all smoke and mirrors, whether it be black and white, grayscale, color, silver halide or doped silicon.

--Bill

[Bjorn Jonsson]

Somehow the IR-G-V images look more "natural" to me than the ones with a synthetic blue channel. I find the orangish colors more natural than the pinkish ones using a synthetic blue channel.
That preference may have its roots in the Voyager imagery so one comes to expect an orange colored Europa -- not neccessarily its true color at all.

The C10 (global) images look more 'natural' to me than the G1 images. This may be because I altered the color balance to make the area around Pwyll whitish whereas for the G1 image I didn't change the color balance. I should probably have changed it to something more similar to the C10 image. That said, I don't think Europa is as reddish as in the C10 RGsB image but I plan to use Cassini CB1GB images where both Europa and Jupiter are visible to see what Europa looks like when I process the image to make Jupiter's color balance similar to what one sees through a telescope.

[ugordan]

That said, I don't think Europa is as reddish as in the C10 RGsB image but I plan to use Cassini CB1GB images where both Europa and Jupiter are visible to see what Europa looks like when I process the image to make Jupiter's color balance similar to what one sees through a telescope.

You might have problems there because Cassini only used CB1 when it was still far away from Jupiter. Once the imaging plan changed to a 2x2 coverage footprint, they reduced the filter set and dropped CB1, leaving only CB2 as a close alternative. OTOH, I can't remember if they even kept the GRN filter coverage.
As a result, CB1 images of Europa are very distant and color differences will probably be unresolvable on such a small disk. My experiments with CB1 always resulted in Europa being more or less white.
There are a few RED/CB1 images taken near C/A IIRC, but they're severely underexposed, see my topic on Cassini Galilean images.
Note that in no way am I saying those colors I got are even remotely accurate (in fact, they're probably very wrong), that attempt was primarily driven by me wanting to know how well Cassini could resolve the moons from such a distance.

[vexgizmo]

Does anyone know if Hapke parameters as a function of wavelength (or Galileo filter) are available somewhere ?

Unfortunately, not yet for the Galileo data. This is the only photometric study I know of from the Galileo Europa data:

Helfenstein, P., N. Currier, B. Clark, J. Veverka, M. Bell, R. Sullivan, J. Klemaszewski, R. Greeley, R. T. Pappalardo, J. W. Head III, T. Jones, K. Magee, K. Klaasen, P. Geissler, R. Greenberg, A. McEwen, C. Phillips, T. Colvin, M. Davies, T. Denk, and M. Belton, Galileo obervations of Europa's opposition effect, Icarus, 135, 41-63, 1998.

The best from Voyager data is:

Domingue, D. L. and B. Hapke, Disk-Resolved Photometric Analysis of Europan Terrains, Icarus, 99, 70- 81, 1992.

[jrdahlman]

edstrick's explanation made me think of a question of spaceprobe construction:

Why even use filter wheels at all?

Now that 1000 x 1000 pixel sensors are so cheap, we can load probes with as many cameras as we want. Look how many the MERs have. So instead of possibly-sticky mechanical wheels, or prism arrangements (imagine a gigantic 8-prism for 8 filters!), why not just have an array of 8 separate cameras all pointing the same way, each with its own dedicated filter?

Granted, for landers or other close-up features the parallax would be unacceptable, but for planetary flyby/orbits, parallax would be too tiny to see. (Wouldn't it?) Then it could take pictures with ALL the filters all at the same time--no motion shift between them! And you wouldn't even have to worry about 8x the data coming back, because you could still transmit them sequentially, or even decide which filtered pictures are worth sending.

How come they're not being built this way now?

[paxdan]

edstrick's explanation made me think of a question of spaceprobe construction:

Why even use filter wheels at all?

Now that 1000 x 1000 pixel sensors are so cheap, we can load probes with as many cameras as we want. Look how many the MERs have. So instead of possibly-sticky mechanical wheels, or prism arrangements (imagine a gigantic 8-prism for 8 filters!), why not just have an array of 8 separate cameras all pointing the same way, each with its own dedicated filter?

Granted, for landers or other close-up features the parallax would be unacceptable, but for planetary flyby/orbits, parallax would be too tiny to see. (Wouldn't it?) Then it could take pictures with ALL the filters all at the same time--no motion shift between them! And  you wouldn't even have to worry about  8x the data coming back, because you could still transmit them sequentially, or even decide which filtered pictures are worth sending.

How come they're not being built this way now?

Because the cost of the sensor isn't the limiting factor, it is the size and weight of the camera(s). Mass and volume constraints are very tight on space craft. A filter wheel is a far more elegent solution to capturing a variety of wavelengths than duplicating the entire camera. A filter wheel weighs very little and occupies very little space, far far less than duplicating the sensor, optics and camera body multiple times.

The problem of motion shift is an interesting one for orbiters/flyby craft. There has been recent discussion regarding the use of 'sport mode' on the MER PANCAM. This cuts the interval between frames to 3 seconds from 10-20. While the interval between frames for Cassini is longer, one wonders wheather the use of subframe images or better routines will be able to squeeze a higher frame rate from the camera.

It is important to remember that spacecraft are primarily scientific instruments. Thus when spacecraft are designed, and choices made between returning scientifically calibrated data or pretty pictures the former must take priority and thus we should be thankfull that the latter is so often acheived as well.

[JRehling]

I have been trying to make (true ?) color images of Europa using mainly Galileo data. I am using red (IR756 in some cases), green and violet images.

Great stuff.

The area in the left image is an example of terrain where bright ice was seen to vary in IR brightness even though it was fairly uniform in visible. That might mean that it's impossible to interpolate/extrapolate between 756nm (near IR, actually just on the fringe of the visible spectrum) and 650nm (more of a central "red"). The differences between the "whites" can be seen vaguely in your image, and might not in "true" color. But beautiful stuff, and probably otherwise very on target, IMO.