Venus cloud tracking -- May 29, 2006

Fig. 1

From earlier this month, this was the best vector map made from the 7/13/04 cloud data. This map used an adaptive flow element that changed size and used a set of 25 sub-elements to estimate the flow near a point.


Fig. 2

This is a speed analysis of the previous map, made by Scot using the 'gridded atmospheric data analysis tool' (http://grads.iges.org/cola.html).


Fig. 3

The first (baseline) image used in the previous, and subsequent, maps, is number 1092 from 7/13/04. Here it is shown in raw form, rescaled, and enlarged, preprocessed, and masked. The enlarged, preprocessed, and masked format is used in all new analyses. This is the sharpest image of the day, and was acquired at UT 16:19:09.347700.


Fig. 4

The second image used is number 2628, also shown in the 3 display formats. This image was acquired at UT 17:32:34.824172.


Maps 1

These images show 2 sets of paired maps made between images 1092 and 2628. These maps were made using the disk of the planet rather than a projection to latitude and longitude coordinates, so as to avoid the distortion and sampling artifacts inherent in such a projection. The left sequence of images shows the flow element geometry (domain and range) and typical candidate match vectors. The right sequence shows the vectors with the highest parametric correlation coefficient for each flow element at a finer grid spacing. Only vectors with a PCC of at least 0.5 are shown.

The top 12 images show flow elements for which the range size/area is 3x/9x the domain size/area. The bottom 12 images show flow elements for which the range size is 50 pixels larger than the domain. This is to limit the range to a region which might contain the actual flow, as it is anticipated to be about 10-20 pixels in magnitude (estimated by using FITSVector). Note that as the domain size increases, the flow becomes more regular and smooth, although it is less local.


Fig. 5
sharpimages2.txt
Notes052206.txt

A sequence of 5 sharp images near in time to 1092 were manually aligned using FITSAlign and then averaged. The composite image was then enhanced with an unsharp mask. This composite was then used as the baseline image in all subsequent maps. The sharpimages2.txt file shows a list of sharp images for time intervals during 7/13/04. The Notes052206.txt file shows details of the alignment results for these and additional composite images.


Fig. 6

A sequence of 6 sharp images near in time to 2628 were manually aligned using FITSAlign and then averaged. The composite image was then enhanced with an unsharp mask. This composite was then used as the comparison image in all subsequent maps.


Maps 2

Same sequence of maps as for Maps 1 above, but using the aligned, averaged, and enhanced composite images rather than single images.


Fig. 7

Using automated sharpness estimates allowed the selection of a sequence of unusually sharp images from the end of the observing night. In the screenshot, the 10 best and 10 worst images at the end of the observation are shown.


Fig. 8

A sequence of 10 sharp images near the end of the observation for 7/13/04 were manually aligned using FITSAlign and then averaged. The composite image was then enhanced with an unsharp mask. This composite was then used as the comparison image in all subsequent maps. The UT of the sharpest of these images was 18:25:11.676666.


Maps 3

Same sequence of maps as for Maps 1&2 above, but using the aligned, averaged, and enhanced composite images from the end of the observation. In this case, flow vectors are drawn at 1/2 the actual displacement, and the second sequence of images shows flow elements with ranges that are 100 pixels larger than their domains, to accomodate the additional expected displacements.


Fig. 9

This screenshot shows 2 frames from an iterated flow calculation of the first half of the observation for 7/13/04. This iterated calculation is similar to those described in Evans ("Glaciers") and in Wu. This sequence of iterations was performed for flow domains of 16x16 pixels. Although this is a preliminary result, it does show promise for making the overall flow more regular and smooth while using a smaller flow element to preserve locality.


Fig. 10

Same as above for flow domains of 32x32 pixels.


Fig. 11

Same as above for the entirety of the 7/13/04 observations, using 16x16 flow domains.


Fig. 12

32x32 flow domains.


Fig. 13

Iterations for the second half of the observations on 7/13/04, using 16x16 flow domains.


Fig. 14

32x32 flow domains.


Movie 1

A 3-frame movie of 7/13/04 constructed from the composite images.


Movie 2

A 2-frame movie of flows at 16x16 pixels.


Movie 3

A 2-frame movie of flows at 32x32 pixels.