Here is an animated GIF of the May 5th 2009 star occultation sequence from the CL1 VIO image set:

Click to view attachment
(click to animate)

For this stack, the limb of Titan was used as the reference for alignment.
The contrast has also been balanced between the images of the sequence.

Sweet. You should probably be asking me this rather than the other way 'round, but what's that star in occultation?

- Jason

That's Alpha Eridani, a 0.45 mag blue-giant star in the constellation Eridanus.

Layer stack aligned:

Click to view attachment

(Not sure of the reason for the slight curve in the track, it might be due to the limb size changing due to changing distance)

Control points added to contrast balance the images (within 1/255 units):
Click to view attachment

Two of the images were selected and averaged, and the region from around the star grabbed from the image without a star to back a background control. This will make the Background Model for the CL1/VIO images.
Click to view attachment

Subtraction of the Background model from the balanced image gives only the starlight coming through the haze layers.
Click to view attachment

For a spatial reference point, I'll use the outer edge of the detached haze layer. This feature seems to show up in both the CL1 VIO and the CB2 CL2 filtered images.

So the mapping of the haze layers will be downward from the outer edge of the detached haze layer.
(Anyone got a reference for the altitude of the outer edge of the haze layer from the surface?)

Using the recent Planetary Photojournal images (and reported pixel scales), here are two crops of the haze structure from PIA11485 and PIA11468.

Lining them both up and assuming similar structures, the measured pixel values translated to distance seems to match up. The detached haze layer is about 60 km thick, and the distance from the outer edge of the detached haze layer to the outer edge of the inner thicker haze layer (but still blue scattering) is about 150 km.

Click to view attachment

Lacking an absolute scale, I'll assume the haze structure is the same in the May 5th images and that the distance from the outer edge of the detached haze layer to the inner thicker haze layer is still about 150 km.

The detached haze layer is at 520 km
link (hope it works for you)

The curvature of the path might be due to refraction due to the atmosphere, which is affected by density (lower altitudes refract more)...
picture: http://gwest.gats-inc.com/sofie/refraction...occultation.jpg

Been doin' some readin (Thanks for the pointer, remcook!):

If I read the graphs and different plots correctly (see references below), the thickest part of the detached haze layer is located at 490 km - 570 km. With the thickest part centered at about 520 km.


References:
Lavvas et al. Icarus 201 (2009) 626-633. "The detached haze layer in Titan's mesosphere". doi: 10.1016/j.icarus.2009.01.004.
(Specifically Fig. 1 is a summary of the studies below)

Liang et al. The Astrophysical Journal 661 (2007) L199-L202. "Photolytically generated aerosols in the mesosphere and thermosphere of Titan."
(Fig 2 is a summary of UVIS theta-Scorpii occultation data)

Porco et al. Nature 434 (2005) 159-168. "Imaging of Cassini from the Cassini spacecraft." doi: 10.1038/nature03436.
(Fig 11 gives plot of I/F with altitude)

Fulchignoni et al. Nature 438 (2005) 785-791. "In situ measurements of the physical characteristics of Titan's environment." doi: 10.1038/nature04314.
(Fig. 2 is the HASI instrument plot from Huygens descent probe, giving the density measurement. Although not imaging, it provides density [=temperature] evidence for the existence of haze layers)

Corrected graphic with values assuming an inner edge of the detached haze layer at 490 km:
Click to view attachment

This is pretty cool -- more than I've seen yet from ISS on the occultation stuff, anyway. VIMS is in spectrum mode for these, and so once it shifts more than 1 pixel we lose the star.

- Jason

Here is the sequence of leveled and aligned (shifted) images using the CL1 VIO filter.

For each CL1-VIO subtracted image, I used the Photoshop integrated density measurement of alpha Eridani's image.
The integrated density was normalized to the integrated density in the initial image in the sequence.

Here is the plot (aligned with an image of Titan's haze)

Click to view attachment

-Mike

Mike, if you're doing what I think you are, you might be over-analyzing the raw data. These raws have individual contrast stretch applied to them which means relative star brightness from frame to frame cannot be compared because the star is point-like and could be clipped to max brightness (so you lose integrated brightness accuracy as it's saturated) and the camera 8-bit encoding is very probably nonlinear LUT. That, coupled with histogram stretch would mess with your frame subtraction.

I was worried about that:

I set 4 control points in dark middle-gray portions of the image for one of the images. (Off the top of my head, for the CB2-CL1 set the pixel values were: 25, 33, 61 and 207). These were in the haze layer and on the "twilight" portion of Titan's disk.
After alignment, each image was then adjusted so that the values of the control points were within 1 pixel value.
(It turns out the raw uncorrected values were all very close to begin with (within about 5 values of each other)

If the images were clipped during creation of the raw images, I think they may have been clipped consistently the same way.

You are totally right, there is no way the data I'm playing with could be used to derive extinction coefficients. But if the data was processed (clipped) consistently it should be possible to provide a relative detector response. (Qualitative, not quantitative)

Here is a graphic showing the placement of the control points for the CL1-VIO images:
(Actually, I think I used different ones originally, these show a pretty tight variation for most of the control points)

Click to view attachment

Here are the ranges for each of the points in the image sequence:

Control point 1: 81 (no changes throughout sequence)
Control point 2: 72-74
Control point 3: 116-125
Control point 4: 115-116

Here is an animated GIF of the rotated and aligned image of the CL1-VIO sequence of the alpha Eridani occultation:
Click to view attachment
(click to animate)

Cool! I think there is a bonus occultation by another star at slightly lower latitudes!

Look at this animated GIF. Archenar (alpha Eridani) is moving to the upper left in this rotationally coordinated image.
Dust artifacts are moving parallel to the lower left. (There is one in the haze gap going towards Archenar and another further towards the right edge of the frame.)

Click to view attachment

For three frames there is a dim star visible in the gap between the detached haze layer and the lower haze layers, indicated by a white circle. It tracks parallel to and with the same shifts as Archenar.
If I've worked Celestia right and reading from VP's Celestia screenshot, this could correspond to alpha Hydrus (magnitude 3).

Click to view attachment

Archenar is setting over the northern temperate zone at about Latitude 30N. The bonus star would be setting over the Equator.

Reworked the CL1-VIO filtered image. Here is a coordinated and rotated image showing the descent of Archenar through the haze column.
Click to view attachment
(click to animate)

(The spot moving from L to R is a camera artifact. Happily, this is also moving in a straight line.)

Corrected absolute altitudes of the CL1-VIO filter sequence:
Click to view attachment

These were redone based on the alignment above and using the value of 490 km for the inner edge of the detached haze layer.
The CL1-VIO filtered images probed the region betwen 427 - 336 km above Titan's surface.

Stacked images of the May 5th occultation of Archenar with the CB2 CL2 filter set:

Click to view attachment
(click to animate)

Here are the control points for the CB2 CL2 calibration of the images. (After leveling; before leveling the value were already very close)
Click to view attachment

Here is a coordinated and rotated image showing the descent of Archenar through the haze column throught the CB2-CL2 filter set:

Click to view attachment
(click to animate)

To calculate the distance from the CB2 CL2 images to the inner edge of the detached haze layer was kinda tricky.

The inner edge of the detached haze layer is not visible in the CB2 CL2 image sequence. (The outer edge is sorta visible by the steep falloff in pixel brightness).

To get the correct measurements, I took the W00056248 CL1 VIO image and coordinated to the W00056247 image in the CB2 CL2 image set.
(By my calculations, there should be less than a 1 pixel difference in the Archenar image.)
I then overlaid the two star images and made the limb sections parallel. (Titan's limb is "inflated" in CL1 VIO filtered images).
A mask was added to not interfere with the lower layer measurements so that the CL1 VIO detached haze layer is now visible in the CB2 CL image set.

The measurements were then made from the inner edge of the CL1 VIO detached haze layer.
(And they match up nicely!)

Click to view attachment

Here is the graphed relative transmission data of Archenar during the May 5th 2009 occultation of Titan's haze layers:

Click to view attachment

Graphed transmission data is normalized to the initial (outermost) integrated density response for each dataset.
It is not meaningful to compare the values between the two datasets - they can only be compared inside the particular filters dataset.

-Mike

Here is the same graph as above, but plotted on a log scale and with literature values.
The backround plots is a trace of Figure 1 (and references therin) from Lavvas et al. Icarus 201 (2009) 626-633. "The detached haze layer in Titan's mesosphere". doi: 10.1016/j.icarus.2009.01.004

Click to view attachment

And it all roughly fits!

(The weird bump in the CB2 CL2 data at about 365 km is not evident in integrated density measurments in the raw data. It may not be significant.)

-Mike

Error analysis of my derived occultation data.

Assuming a linear detector response.

Taking worst case leveled control point variation:
For CL1 VIO filter data:
distance CL1 VIO control point 3
42671 124
42008 122
41677 122
41014 120
40683 123
40352 130
39689 137

average 125.4285714
std dev 5.996030433

2x std dev = 11.8 (95% confidence)

Error propagation of subtracting one leveled image with consensus model built from leveled image: SQRT[(11.8^2)+(11.8^2)] = 16.7
Multiplication of a constant (area = 177 x 16.7 = 2956 error in integrated density measurement) {= pixel area x pixel value error}

Value used as 100% response for normalization is 5755. (thus, 5755 +/- 2956)

Division of the integrated density value with another integrated value density value gives normalization errors ranging from (+/-)71% (uppermost measurment) down to (+/-) 54% (lowest measurement nearest limb).


For CB2 CL2 filter data:
distance CB2 CL2 control point 4 (post level)
42641 208
42310 208
41979 208
41316 209
40984 207
40322 208
39991 209
39659 207

average 208
std dev 0.755928946

2x std dev = 1.4 (95% confidence)

Error propagation of subtracting one leveled image with consensus model built from leveled image: Sqrt[(1.4^2)+(1.4^2)] = 2
Multiplication of a constant (area = 177 x 2 = 354 error in integrated density measurement) {= pixel area x pixel value error}

Value used as 100% response for normalization is 5755. (thus, 823 +/- 354)

Division of the integrated density value with another integrated value density value gives normalization errors ranging from (+/-)61% (uppermost measurement) down to (+/-) 46% (lowest measurement near limb).

Them error bars be pretty wide.

-Mike

Image showing really really subtle dark details of Titan's haze layer in CL1-VIO filter image of the T55 flyby (May 21):

Click to view attachment

(The bright crescent is an artifact of the enhancement I used).