There is a problem w/ your write-up. Check that you have valied entries for \$CAID and \$Campn in your analysis.php file. If that checks out, then Contact Stansberry"; return ; } // get first matching task $row = mysql_fetch_array($result); $title = $row["title"]; $princ = $row["principal"]; $deputy= $row["deputy"]; $campn0 = $row["campn0"]; $aorkeys = $row["aorkeys"]; // get real name of principal, deputies $princ = ioc_get_person($princ); $princ = $princ[0]; $deps = explode(",",$deputy); foreach ($deps as $depty) { $depty = trim($depty); $depty = ioc_get_person($depty); $depty = $depty[0]; $depty = explode(",",$depty); $depty = $depty[0]; // last names only $deplist[] = $depty; } $deplist = implode(", ", $deplist); $caid = sprintf("%03d",$caid); $file = "mips-".$caid.$campn.".analysis.php"; // if more matches, append the AORKEYS from those $numrows = mysql_num_rows($result); if ($numrows > 1) { $aorkeys = " " . $numrows . " Task Executions:  ". $aorkeys; for ($i=0;$i < mysql_num_rows($result); $i++) { $row = mysql_fetch_array($result); $morekeys = $row["aorkeys"]; $aorkeys = $aorkeys .';  '.$morekeys; } } // END PHP. ?> <? echo "MIPS-$caid, Campaign $campn IOC/SV Analysis"; ?>

Principal:
Deputy:
Analyst:
AORKEYS:
Last Updated:

Task Outcome Summary

Abstract

The goal of MIPS-320 is to validate the operation of the MIPS 24 micron small field photometry mode in a rudamentary way. Because of the image orientation and WCS problems detailed in MIPS-100, placement of the star image on the array appears unlike the uplink design. However, by presuming an array flip in X (X = -X) and a 180 degree rotation in the Y WCS direction, we show that the image is placed as expected. We confirm that the measured stellar photometry apparently depends on exposure time (see MIPS-315). The scan mirror positioning is remarkably repeatable, placing the star within the same pixel during multiple cycles and different AORs.

Analysis

The design of the 24 micron small field photometry AOT is described in the SIRTF Observers Manual v3.0. The implementation of the AOT is slightly different from the description in the manual. For the purpose of this discussion, I define array directions as x, y with y positive toward the 160 um array and x positive in the direction away from the boresight. The (1,1) pixel (lower left) serves as the origin unless the center of the array is referenced. The initial placement of the target on the array is 1 arcmin left of the array center (-25 pixels in x). This placement is accomplished using a relative frame table entry (24um_small_FOV1) which is 25 pixels from the array center in the +theta-Y (-x) direction. The y offset of the initial pointing is accomplished entirely by the scan mirror offset from its optical zero position (2007) during the first DCE. The CSMM scanabs position for DCE0 is 1929, which produces an approximately 41.5" offset of the target in the +theta-Z (+y) direction. The scan mirror parameters for the 24 microns small field photometry AOT mode are as follows: scanpos1 = 2007, scanpos2 = 0 (2007), relpos1 = 1510, relpos2 = 3360, stepoffset = -86, stimcycle = 7. These parameters produce the following moves of the mirror and target (using the mirror emulation formulae described by C. Englebracht on his web page:
DCE Number
 CSMM SCANABS (Coarse DAC counts)
 Pixel Offset from Array Midline
 Theta-Z offset from Array Midline (arcsec)
 Offset from DCE 0 Position (arcsec)
0
 1929
 16.3
 41.5
 0
1
 2149.5
 -29.9
 -76.1
 -117.7
2
 1907.5
 20.9
 53.1
 11.5
3
 2128.0
 -25.4
 -64.6
 -106.4
4
 1886.0
 25.4
 64.6
 23.0
5
 2106.5
 -20.9
 -53.1
 -94.7
6
 1864.5
 29.9
 76.1
 34.5
7
 1929.0
 16.3
 41.5
 0
 In the case where only a single cycle is selected, eight DCEs are taken (the last at the position of the first), then the telescope is offset using PCS_SET_FRAME to -25.5 theta-Y which places the target 50.5 pixels +x of its previous position. The y offset remains the same as for the initial pointing, that is 41.5 arcsec (16.3 pixels) "up" = +y from the centerline of the array. The same set of mirror offsets are performed as in the first column, leading to the following target positions in pixel coordinates:

Exposure Number
 DCE Number
 Array x position
 Array y position
0
 0
 25
 80.8
0
 1
 25
 34.6
0
 2
 25
 85.4
0
 3
 25
 39.1
0
 4
 25
 89.9
0
 5
 25
 43.6
0
 6
 25
 94.4
0
 7
 25
 80.8
1
 0
 90
 80.8
1
 1
 90
 34.6
1
 2
 90
 85.4
1
 3
 90
 39.1
1
 4
 90
 89.9
1
 5
 90
 43.6
1
 6
 90
 94.4
1
 7
 90
 80.8

In the case of multiple cycles, the CSMM dithers are repeated several times in each column prior to the spacecraft offset. For the observed case of 3 photometry cycles, 22 DCEs are taken in each column of CSMM dithers. In order to test our ability to position an object on the array using this AOT, we observed a star which was bright enough to obtain good S/N for 3 s exposures, but would not saturate during a 30 s exposure. Four observations were obtained of the B8V star Nu Doradis using the 24um small field photometry AOT. We obtained single cycles of photometry with exposure times of 3, 10, and 30 seconds. Three cycles of 3 second exposures were also taken. In order to analyze these data, we processed the four AORs from this task through DAT V2.15 (mips_sloper and mips_caler), using flats and darks taken during the D2 campaign. Each individual DCE was then extracted into a separate fits file, which was then analyzed in IRAF. Image statistics were obtained for the entire array, as well as a 30 x 30 pixel region around the array center. Rough photometry (using imexam) was performed on the stellar image to obtain stellar centroids and background subtracted fluxes in DN/s. A comparison set of reductions for the single cycle of 3 second exposures only was made using the SSC offline pipeline S8.4 (which repairs the array x flip and WCS rotation noted in MIPS-100. Header keywords related to the scan mirror were extracted and compared versus expectations. Unfortunately, the array flip and WCS rotation prevented the initial analysis from utilizing undistorted images. Thus, this preliminary report will have to be revised when these data are available in undistorted form. Note that 8 lines of data were missing from the first delivery of the 6765056_0001_0001 DCE.

Results

>Since the array center offsets discovered in MIPS-100 had not been incorporated in the spacecraft frame table, we expected to see the entire pattern offset towards the bottom of the array. Here are the measured positions of the star during a single cycle (6765056) of 3 second exposures in the 24 um AOT after correcting the flip-flop. I used the algorithm Xpos = Xpos - ((xpos - 64.5) * 2) to simulate the x position of stellar images measured on unflipped images. The average offset in x from the array center is -23.31 +/- 0.13 pixels for column 1 and 27.61 +/- 0.47 pixels for column 2. Thus, the scan mirror tracks nicely in the array y direction for column 1, but is less well aligned for column 2 in the distorted images. Further analysis with undistorted DCEs is required to confirm that distortion on the right side of the array is responsible for this discrepancy (which assists the sampling of PSFs in the second columns. Given that the expected x offsets from the center are -25 and + 25.5 pixels, I confirm the finding of the FPS survey team that we are about 2 pixels offset from the nominal array center in the W direction. The average separation between the dither columns is 50.9 pixels, which is 0.4 pixels larger than the expected PCS_SET_FRAME offset. Again, further analysis should confirm whether distortion is the culprit. In the y direction (V, theta-Z), I find the following mean offset from the array center for a given CSMM position:
 
CSMM SCANABS
 Mean Y Position Offset from Array Center
 STDDEV Y Position Offset from Array Center
1929
 9.53
 0.05
2149.5
 -38.13
 0.05
1907.5
 14.22
 0.19
2128
 -33.42
 0.04
1886
 18.73
 0.04
2106.5
 -28.76
 0.07
1864.5
 23.32
 0.05

Positioning by the CSMM is generaly repeatable to 0.05 pixels (the value of 1907.5 is affected by CR hits). For the current AOT design, the y offsets are about 18 arcsec below the expected values, confirming the array center 18" -theta-Z offset found by the FPS team. The magnitude of the offsets between scan mirror positions is also listed with the expected offsets. The chops are repeatable to 0.5 arcsec at worst. Interestingly, the chops in the -y direction deviate from expectation (by 3 arcsec) more than the positive y chops. Further analysis is needed to determine whether this is a real effect.
CSMM Chop
 Mean Chop Magnitude (pixels)
 STDDEV Chop Magnitude (pixels)
 Mean Chop Magnitude (arcsec)
 STDDEV Chop Magnitude (arcsed)
 Offset from DCE 0 (arcsec)
1929 - 2149.5
 -47.65
 0.07
 -121.30
 0.17
 -121.3
2149.5 - 1907.5
 52.35
 0.19
 133.27
 0.47
 11.96
1907.5 - 2128
 -47.64
 0.18
 -121.28
 0.46
 -109.32
2128 - 1886
 52.25
 0.06
 132.78
 0.15
 23.46
1886 - 2106.5
 -47.49
 0.10
 -120.90
 0.26
 -97.44
2106.5 - 1864.5
 52.07
 0.11
 132.58
 0.28
 35.13
1864.5 - 1929
 -13.80
 0.03
 -35.13
 0.08
 0.0

 Comparison of reconstructed pointing (with the flip-flop corrected) with the actual stellar position reveals that the pointing reconstruction has an error of about 40 arcsec. A preliminary investigation suggests that the discrepancy is due to a bug in the MIRRORSYNC module of the pointing reconstruction. Apparently, the module is not correctly accounting for the offset from the optically null position of the scan mirror for DCE0. This error will be corrected in an upcoming build of the SSC software (ETA TBD).

The dark spots discussed in Almadena's reports are highly apparent in the 30 second exposures. Deeper images reveal at least 8 spots of varying size and depth, suggesting that masking of the images may eliminate an undesireable percentage of the array. Also, although the two darkest spots do not interfere with placement of the dither columns in photometry mode, other spots are directly interfering with the currently placed column. It is also true that the first dither column (-x) has one stellar position which falls too close to a bad pixel cluster (1864.5). The slight x correction to the array center position in the latest frame table update will help the situation, but the ultimate remedy may be to move the first column closer to the center by changing the 24um_small_FOV1 frame's definition. This could be performed at the same time as theta-Z offset is added to the frame definitions to center the pattern in the y direction on the array.

Image statistics were obtained for all images in this task. I ran IRAF imstat over the entire array and a 30 x 30 pixel subset in the array center which excluded the stellar images. The following table lists the results.

AORKEY
 Exposure Time/Cycles
 Median Array Slope (DN/s)
 STDDEV Array Slope (DN/s)
 Median Slope [50:80,50:80]  Subimage (DN/s)
 STDDEV  Subimage (DN/s)
6765056
 3/1
 556.24 +/- 1.11
 48
 558.42 +/- 1.90
 13
6765312
 10/1
 554.92 +/- 1.28
 37
 556.81 +/- 1.06
 9
6765568
 30/1
 559.95 +/- 0.59
 40
 561.86 +/- 0.40
 7
6765824
 3/3

555.59 +/- 1.94
 11
 
There is a statistically insignificant trend toward higher median slopes (DN/s) with increasing exposure times. Unfortunately, in these images which have not yet been cleaned of cosmic rays, it is not yet possible to assess the efficacy of increased exposure time on decreasing the noise of the background. There is a suggestion of this effect in the error of the median in the central subimage, which decreases from 1.9 at 1 cycle/3 s, to 1.09 (1 cycle/10s), and finally to 0.4 for 1 cycle/30s. Certainly, the images appear deeper at longer exposure times, with more faint sources becoming apparent in the longer integrations. However, the number of stellar images affected by strong CR hits also increasess. The interplay between exposure time and cosmic ray accumulation will be quantified in MIPS-245 during Campaign O.

 Stellar photometry was performed on the images in MIPS-320. The results are listed in the following table. Chad Engelbracht conducted an independent measure of the photometry in his MIPS-315 report, and finds consistent results.

AORKEY
 Exposure Time/Cycles
 Mean Stellar Flux (DN/s)
 STDDEV over Array (DN/s)
6765056
 3/1
 7193
 246
6765312
 10/2
 7320
 180
6765568
 30/1
 7698
 177
6765824
 3/3
 7197
 196

 As Chad reported, there is some evidence that the point source photometry of Nu Dor is dependent on exposure time. The mean flux of the star in a 5 pixel radius aperture is 300 DN/s higher for the 30 second exposures compared to the 10 s exposures. This is not an effect of saturation, since saturation would decrease the measured slope at the longer exposure time. However, this is only a 1.5 sigma result, and the number of usable 30 second exposures was small due to many sigma CR hits. A comparison of stellar photometry versus exposure time with a robust number of samples will be required to settle this question. If the effect is real, it may suggest a problem with linearity correction. Beyond this effect, we clearly see the variation of stellar photometry across the array reported by others at the 3% level. Although this effect is presumed due to distortion, reanalysis of undistorted data is required to confirm this cause.

Measurements from the first DCE in an exposure sequence were excluded from the image and photometric statistics reported above. We find evidence in the array median slopes that the DCE slopes are low by 3% for the 3 s exposures, and low by a bit more than 1% for the other exposure times. Visual examination of these frames shows that the "jailbar" effect is often enhanced, suggesting a problem with the darks. The SSC pipeline is set up to use different calibration files for the first DCE if necessary. Unfortunately, we have too few first DCEs in this task to thoroughly explore the phenomena. I suggest that additional datasets comparing RAW and SUR stellar photometry for many DCE0 exposure be obtained for the purpose of characterizing the behavior of the first DCE.

Finally, we processed AOR containing 1 cycle of 3 s exposures through the SSC offline 24 micron pipeline to compare versus the DAT processed exposures. Despite the fact that the SSC pipeline used Campaign E flats/darks rather than those from D2, the results were remarkably consistent. The SSC mean fluxes for the star were 0.8% higher than the DAT's, while the whole array median slopes were 0.6% lower. Consistency checks are planned to continue throughout IOC/SV. In both sets of reduced data, there is a slight gradient toward the bottom of the array in the sense that the bottom is fainter than the rest of the array by a few DN/s. This suggests that the linearity correction calibration file (identical currently in both pipelines) may not be correct. An upcoming task to evaluate 24um linearity may settle this question.

Conclusions

The 24 micron small field AOT is in good shape. All the images appear on the array near the expected positions once the frame table updates and array/WCS flip-flops are applied. Small discrepancies have been identified in the magnitude of the PCS_SET_FRAME offset and the scan mirror chops which move the image -y on the array. In addition, the CSMM direction appears slightly more well-aligned on the -x side of the array. The effects of distortion on these discrepancies has yet to be evaluated. An absolute pointing problem identified in the properly pointing reconstructed images appears to be due to a bug in the SSC MIRRORSYNC module. The relative frame table entries used by this AOT mode should be slightly adjusted to center the pattern in the y direction and move the first dither column away from a bad pixel cluster. Slope medians over the array and stellar photometry both suggest a slight dependence of slope on exposure time. As others have reported, stellar photometry varies by ~3% over the array, probably due to distortion. The slope and photometry values returned by the first DCE in an exposure sequence are systematically low.

Output and Deliverable Products

Deliverable products from this task will be a change request filed for the MIRRORSYNC module and a potential adjustment to the frame table entries 24um_small_FOV1 and FOV2. 

Actions Following Analysis

When correctly oriented, undistorted images are available for the entire dataset, the analysis will be updated to determine whether any action is needed on the small discrepancies in CSMM positioning.