In-Scan Image Elongation - 2d Image Diagnostics

In-Scan Image Elongation - 2d Image Diagnostics

R. Cutri, T. Jarrett, E. Kopan - IPAC

In the course of working on the 24hr System Health Monitor and photometric validation, we have encountered some examples of in-scan image smearing .

We have just begun to investigate this, but the implications are obvious for the effects on photometry and astrometry. In order to determine how long this smearing has been occuring, and its frequency between and within nights, Tom Jarrett has developed a 2d image shape parameter that will be run on calibration scans for a selection of nights already processed. Tom's code evaluates the in-scan and cross-scan second image moments from the psf-maps generate during pipeline processing. These psf-maps are constructed from frame data, so should be good diagnostics of the true scan-averaged frame image quality. The 2d shape diagnostic is the square root of the ratio of the cross-scan to in-scan second image moments (effectively the aspect ratio). Values of the ratio close to unity imply a symmetric image, values less than unity are elongated in the in-scan direction, values greater than 1.0 are elongated in the cross-scan direction. Position angle of the elongated will of course affect the ratio, but the examples below show its utility.

Figure 1 shows examples of the average J, H and Ks psf maps (left to right) for scans taken on 970521n and 970619n. The top row illustrates the psf's for scan 40 on 970521n, when the J and H image ratios are close to 1.0, but Ks has a value near 1.2. The center row is taken from scan 14 on 970619n; all three bands show symmetric images, and the image moment ratios are near 0.95. The bottom row shows the scan-averaged images for scan 66 also from 970619n, and the images are very assymmetric. Image moment ratios for this scan are between 0.6 and 0.7.

Figure 2 and Figure 3 show plots of the J, H and Ks image moment ratios for all calibration scans on 970521n and 970619n, respectively. The images were fairly well behaved on May 21, with some slight deviations from unity in the Ks image - a suggestion of slight misfocus since J and H are symmetric. The behavior on June 19 is dramatically different. The image quality in the first set of calibration observations is good. The later calibration observations show greatly elongated images. Scatter among the ratios in a given band for a given calibration observation set indicate that deviations of >0.1 from unity should be considered significant.

Median Nightly Moment Ratios

The image moment calculation has been automated and run on all nights that have psf-maps generated. The Table below summarizes the median and rms of the image moments in J, H and Ks for each of these nights. Also listed are the average and rms telescope temperatures as reported in the telemetry from the observatory.


Night Date   N  medrj  rms  medrh  rms  medrk  rms  Tavg  rms  
-------------------------------------------------------------
82    970521 58 0.974 0.046 1.006 0.049 1.065 0.075  9.9  1.0
83    970522 46 0.931 0.053 0.962 0.058 0.991 0.067 13.3  0.5
86    970525 54 0.953 0.024 0.976 0.022 1.020 0.030 10.1  1.5
91    970530 60 0.802 0.034 0.815 0.032 0.802 0.039 18.9  0.7
96    970605 60 0.791 0.032 0.801 0.041 0.791 0.045 18.6  0.8
100   970608 82 0.931 0.055 0.957 0.083 0.971 0.048  9.6  1.5
101   970609 60 0.893 0.057 0.926 0.058 0.933 0.086 12.9  0.6
102   970610 60 0.796 0.045 0.817 0.065 0.790 0.071 14.7  1.2
104   970612 12 0.849 0.064 0.867 0.039 0.829 0.085 16.9  0.3
105   970613 36 0.951 0.028 0.967 0.030 0.972 0.040 11.5  1.3
106   970614 33 0.871 0.050 0.922 0.049 0.922 0.055 12.0  2.1
107   970615 60 0.905 0.037 0.929 0.036 0.944 0.040 11.6  2.1
108   970616 57 0.903 0.026 0.935 0.030 0.953 0.034 12.3  1.6
109   970617 60 0.815 0.035 0.828 0.046 0.815 0.049 15.8  0.5
110   970618 62 0.697 0.069 0.699 0.065 0.674 0.082 19.2  0.3
111   970619 60 0.705 0.117 0.694 0.134 0.705 0.127 19.5  0.9
112   970620 60 0.750 0.040 0.750 0.046 0.724 0.064 18.3  0.6
113   970621 60 0.782 0.057 0.782 0.099 0.749 0.056 17.2  0.6
114   970622 55 0.791 0.030 0.795 0.047 0.775 0.041 17.3  1.1
122   970630 59 0.741 0.055 0.730 0.066 0.713 0.072 18.8  1.0
129   970707 60 0.774 0.047 0.766 0.061 0.750 0.057 19.1  1.9

The median ratios are plotted as a function of sequential night numbers in Figure 4 . In this figure, J moments are shown in blue, H in green and Ks in red. Plotted below the image moment ratios are the average telescope temperature.

Elongation is a Temperature Dependent Effect

Tom Chester has examined how nightly median image moment ratio varies with average nightly temperature. He has found a strong correlation between the two in the sense that higher average temperature corresponds to increased in-scan elongation. The bottom portion of Figure 4 also illustrates the strong correlation between image distortion and average telescope temperature.

Elongation in R1 Frames

There is evidence that when image elongation is most severe in the R1-R1 frames the images in R1 are also elongated. This is difficult to see since it requires a number of moderately bright stars to be present in individual frames. Fortunately, calibration scans covering dense fields were observed on 970619n and 970630n, nights with bad image elongation. A composite of R1 and R2-R1 J frames from two scans from the night of 970619n show elongated R1 frames. The image moment ratios derived from the R2 psf-maps for these two scans are both 0.64. It also qualitatively appears that the aspect ratio of the R1 star images are comparable to those in R2-R1. A composite of frames from 970630n shows data from two scans with image moment ratios of 0.72 and 0.73. Although it appears that the R1 images may be slightly elongated when the image is zoomed, it is not obvious.

These two examples illustrate the general results that for the very worst elongation, the star images in R1 frames also appear elongated. However, in frames with just slightly better quality R2-R1 images, no definite conclusion can be drawn about the possible elongation of R1 images. It will be difficult to adapt the tools for developing composite images from the R2-R1 frames to study the R1 frames since there are in general very few stars in the R1 images.

Summary

The strong correlation between image elongation and telescope temperature and the fact that R1 images are also elongated suggest that the distortion is an optical effect rather than a mechanical scan rate effect. This led to telescope focus tests that are described here. The conclusions drawn from those tests indicate that the image elongation results from a combination of focus and telescope mis-collimation.



Last update 11 July 1997