Astronomy 350 Au2002

CCD Differential Photometry of CY Aquarii

Part 3: Measure the Images

Step 3 Updated 2002 Dec 1!!

This part is the most technical of the 5, and will likely take the most time. However, it is in fact rather repetitious once you get into it, so it picks up pace as you get into it. One caution here is not to give in to the temptation to try to cut corners. If you get off track it can take a while to get back.

In each image you will see the target star CY Aquarii as the brightest object near the center of the field, surrounded by a number of fainter stars. We will try to use these stars as "comparison stars" to do differential photometry on CY Aqr. So far as we know at the outset (and will learn more about them as we proceed), these stars are not likely to be variable, and so will measure the brightness of CY Aqr relative to them. We do this because some (maybe most) of the data were taken during conditions of variable atmospheric transparency, specifically thin cirrus clouds. By doing "differential photometry" we can remove the affects of thin clouds as over such relatively small fields of view, since the effect of thin clouds is to diminish the light of all stars on the image by the same amount. (Another way of saying this is that clouds should not have any structure on the scale of a few arcminutes, which experience suggests is true in general at the levels we are concerned with here).

This is why we will be doing our analysis without an absolute flux calibration. Further, since differential photometric monitoring is one of the YALO project's primary missions, only very rarely are standard stars observed. Differential photometry is a very powerful technique as we will see, as it lets us get information from variable targets even during observing conditions that do not permit absolute photometry.

In addition to CY Aqr and companion field stars, there are other things including detector blemishes that will appear in all frames, and bright pixels which mark the passage of a cosmic ray through the CCD during the exposure. These leave behind little bright spots that look different from stars, and which may be ignored unless they land right on a star.

Our goals for this part are to make the following measurements:

  1. Identify the variable star and two fainter stars to serve as comparison stars.

  2. Measure the typical seeing and the sky levels during the observations.

  3. Measure the brightness of each of the stars by adding up the counts that fall in circular apertures centered on each star, and using an annulus of pixels around the outside the "star" aperture to estimate the sky level around each star.

  4. Extract the time of the observation from the observing logs, and tabulate the times in terms of the Heliocentric Julian Date at the middle of each exposure.

The outcome of this part will be to compile a table of raw photometric measurements for each image of CY Aquarii and two nearby comparison stars. These measurements will form the basis of the differential photometry analysis in Part 4.

All measurements you take during this step must be recorded in your lab notebooks.

Step 1: Identify the Variable and Comparison Stars

Choose an image from among your data set (other than the very first image) and display it in XVista using the TV command. Using the finding chart you created in Part 1, verify that your images are of CY Aqr and vicinity. Identify the stars on your images and on the finding chart.

The image scale of the ANDICAM CCD detector is 0.6 arcseconds/pixel when the images are binned 2x2 as we have done for these observations.

  1. What is the orientation of your images as seen in the XVista image display? What is the approximate field of view in the North-South and East-West directions of your images?

  2. Make a hardcopy of the image using the XVista IMPOST command. This creates a PostScript file named Print this by typing "lpr -Pastrolab" in XVista. When you run IMPOST, use the SCALE= keyword to put an arcsecond scale on the axes of your hardcopy. On the printout of your image, mark with a pen the cardinal directions North, South, East, and West on the sky.

  3. Consult this GIF image of CY Aquarii taken from the initial observations done by the instructor. On it I have identified the variable ("Var") and two comparison stars I would like you to use, "C1" and "C2". Mark these on your CCD finder that you made above. You will refer to your finder for the rest of the exercise.

  4. Using the interactive cursor on the XVista TV display window, note the locations of the variable and the comparison stars in pixel coordinates (x,y). Make a table giving the locations of "Var", "C1", and "C2" as they appear on your images.

You will use these names for the comparison stars for all subsequent measurements. Make an annotated finding chart from your "reference" image in this step. You will include this image in your notebook as part of your final writeup. You should label the orientation of the image on the sky (showing which way is North and East is sufficient), the image scale, and neatly label the target star (CY Aqr), and the two comparison stars (C1 and C2). Also note the UTC date and time of the observation.

Step 2: Measure the Seeing and Sky Levels

Start by using your "reference" image from Step 1 above.

  1. Redisplay the image using the TV command. You may need to adjust the intensity levels to be able to see both the variable and the fainter comparison stars easily. I suggest using an inverse Black-&-White color map (give TV the cf=ibw keyword).

  2. Once you have the image displayed, use the HISTOGRAM command to plot a histogram of the pixel intensities in the image. You may need to adjust the X-axis limits by using the XMIN= and XMAX= keywords so that you can be able to see the peak of the histogram clearly. Using the cursor on the plotting window, make an eyeball estimate of the sky level by record the X-axis location of the peak of the pixel intensity distribution in your notes (this is the "mode" of the distribution of pixel intensities).

  3. Now compute the mode of the image intensities explicitly using the SKY command. This command computes the mode by fitting a parabola to the peak of the image intensity histogram to help average over the effects of noise on making the peak of the histogram appear "rough". It also provides you with an estimate of the uncertainty of the modal sky level. Record this number in your notes. How well does it compare with your by-eye estimate?

  4. The CCD gain has been measured to be 3.6 electrons/ADU, and the readout noise is measured to be 11 electrons (rms). Using the reported modal sky value above, what would you predict for the uncertainty assuming only Poisson and readout noise? How does the formal estimate compare to the uncertainty estimated from the width of the intensity histogram by the SKY program?

  5. Now issue the command
       tvstar scale=0.6
    and put the cursor onto the image of the variable star. If you hit the "C" key, it will plot the radial brightness distribution of the stellar image, showing pixel intensity as a function of radial distance from the center of the image. Since each pixel is at a different radius, the points should scatter along in radius in at apparently random intervals.

  6. By-eye, estimate the radius at which the intensity drops to halfway between the central brightness and the sky level at large radii. Twice this number is a rough estimate of the Full-Width at Half-Maximum (FWHM) of the stellar image. Record this number in your notes. The units are arcseconds since you told TVSTAR that the image scale was 0.6 arcseconds/pixel above.

  7. Still in the TVSTAR command, put the cursor on the star in the TV display and hit the "X" key. This will now fit a Gaussian profile to the radial distribution, and on the text window it will print the best-fit centroid of the star in (x,y), and its estimate of the Full-Width at Half-Maximum (FWHM) of the star image. Since images may be slightly elliptical, the FWHM is computed along the major and minor axes independently. The mean FWHM is computed from these values, reported by TVSTAR as "fwhm" (the other two are reported as "maj x min"). Record each of these numbers in your notes. We will adopt the mean FWHM as the "seeing" for the image. How well does the Gaussian fit value compare with your by-eye estimate from the radial plot in the previous step?

  8. Exit from TVSTAR by putting the cursor onto the TV display and hitting the "E" key (E for Exit).

Now, starting with the first image in your data set, do the following:

  1. Read the image into buffer 1. While in principle you should have enough memory on the workstation to load all of your images at once, this would just make it too easy to make mistakes. Do them one at a time, using only buffer 1, and things will go much faster and relatively mistake free (trust me).

  2. Display the image using the TV command.

  3. Compute the modal sky level and its uncertainty using the SKY command. Record this in your notes. There is no need to repeat the by-eye estimate like we did with the reference image.

  4. Using tvstar scale=0.6, measure the seeing for the variable star in arcseconds using the "X" key as above. If the two FWHMs are very different (i.e., the star is not very round), record this fact in your notes, along with the FWHM along each axis in addition to the "mean" FWHM. Remember to hit the "E" key to exit once you have made your measurements.

Now repeat this procedure for every 5th image in your data set (i.e., do this for images 1, 5, 10, 15, 20, and 25). This will give you an idea of the typical seeing and sky levels during the observations. (Doing it for all 25 galaxies is excessive).

Now you are ready to begin measuring the brightnesses of CY Aqr and two comparison stars on each of your 25 images.

Step 3: Estimate the size of the photometry and sky apertures to use.

From your table of seeing measurements from Step 2, compute the mean seeing for the data set.

  1. Adopt as the star radius 1.7 times the seeing. Convert this to units of pixels and round it up to the nearest pixel. Record this number in your lab notebook as the Star Aperture Radius.

    For example, if you measure the typical FWHM to be 2.63 pixels, the sky radius should be

        sky = 2.63 * 1.7
            = 4.47
    which you should round up to 5 pixels.

  2. Adopt as the Sky Annulus Inner Radius 11 pixels. This should be good for 99% of the images in this data set.

  3. Using the star aperture radius and sky annulus inner radius values, estimate the "outer sky radius" such that the area of the sky annulus is approximately equal to the area of the star aperture. Round up to the nearest integer number of pixels such that you have at least 2 pixels of width in the annulus. Record this number as the Sky Annulus Outer Radius in your lab notebook.

    For example, the star aperture has a radius of 5 pixels. The inner radius has been adopted to be 11 pixels, and the outer radius is then

           r_outer = sqrt(r_star2 + r_inner2)
    For r_star=5 and r_inner=11, this gives r_outer=12.1, which you would then round up to 13 pixels, so that you have an annulus width of 2 pixels. Thus,
        Star Aperture Radius = 5 pixels
        Sky Annulus R_inner  = 11 pixels
                    R_outer  = 13 pixels
    would be the parameters for the photometry apertures in this example. Your values, of course, may be different.

Step 4: Build a Template Star List

The outcome of Step 3 was to select the size of the object and sky apertures to be used for measuring the star and sky brightnesses on all of your images. It is essential to use a common aperture for all of your measurements.

Examining calibration images acquired with the ANDICAM over the last few months, the CCD gain and readout noise are:

gain = 3.6 electrons/ADU
readout noise = 11 electrons (rms)
These values are quite stable from night to night (actually from month to month), so we will adopt these for our noise model of the ANDICAM CCD detector.

In XVista, start with the first image:

  1. Read the image into buffer 1, and display it with TV.

  2. Have your finder with the variable and comparison stars C1, C2, et al. marked handy. Issue the command
       markstar new
    This starts the MARKSTAR command and activates the interactive TV display cursor.

  3. Put the cursor onto CY Aqr and hit "C" to mark it. MARKSTAR will draw a box around the star and report the measured centroid on the text screen.

  4. Now move the cursor onto comparison star C1 and mark it by hitting the "C" key.

  5. Now move the cursor onto comparison star C2 and mark it by hitting the "C" key.

  6. When you finished marking the comparison stars, hit the "E" key to Exit from MARKSTAR.

  7. Save your "star list" with the command
        save phot=cyaqr.pho
    This will create the file "cyaqr.pho" with the star list for the first image. This will be our "template" star list.

Step 5: Measure the sky-subtracted counts for CY Aqr and the two comparison stars

Now that you have created a template star list, you can measure the brightness of CY Aqr and the two comparison stars on all of your images.

In this step, you will use the XVista command aperstar which uses the star aperture and sky annulus to estimate the sky-subtracted counts in the stars. aperstar does all of the dirty work of measuring and subtracting the sky levels, and estimating the noise in the measurements.

Follow these steps:

  1. Load the template star list with the command:
       get phot=cyaqr.pho

  2. Read the first image in your data set into buffer 1.

  3. Display the image with TV.

  4. Measure the star centroids using the template list:
       markstar auto exit
    You should see squares plotted around your variable and comparison stars, along with some other info printed on the screen (e.g., the shift in rows and columns between the image and the template).

  5. Suppose you had determined earlier that your optimal star radius was 5 pixels, and that the sky annulus had inner and outer radii of 11 and 13 pixels, respectively. You would measure the star brightnesses with the APERSTAR command as follows:
       aperstar 1 star=5 sky=11,13 gain=3.6 ronoise=11
    You will see a printout like this
        524.60   870.69   4.068E+05 + / -  3.81E+02
        157.37   612.34   8.181E+03 + / -  1.83E+02
    The values in the table are as follows:
        Rcen  Ccen   Counts  + / -  Sigma
    Where the first two numbers are the measured centroid of the star in row and column coordinates, "Counts" is the sky-subtracted instrumental counts of the star in ADU, and "Sigma" is the estimated uncertainty in Counts in ADU, estimated using a standard CCD (Poisson + Readout Noise) noise model with the gain and readout noise given above. APERSTAR does all of the dirty work of computing the sky, counting the pixels in the star and sky apertures, and performing the formal noise estimates.

  6. Record the instrumental counts and its uncertainty for CY Aqr and the two comparison stars. You do not need to record the star centroids.

  7. Repeat this process until you have measured the instrumental counts and their uncertainties for CY Aqr and the two comparison stars for all of your images.

Now you can quit XVista, you're done with the images.

Step 6: Get the UTC and HJD times for your observations

Using the observing logs, compute the UTC time at mid-exposure for each of your images. Write these down in standard hh:mm:ss.s format.

In order to merge the data from different nights, we need to convert the UTC times at mid-exposure into Heliocentric Julian Dates. This puts the data all on the same timescale, removing effects due to the motion of the Earth around the Sun between observations taken on different nights.

For this step you will use the hjdcalc program. It works from the Unix command prompt as follows:

   hjdcalc ccyy-mm-dd hh:mm:ss RA Dec
   ccyy-mm-dd  =  UTC Date (e.g., 2000-09-15) from the observing log
   hh:mm:ss    =  UTC time at mid-exposure (UTC log time + 10 seconds)
   RA          =  Right Ascension of the object in hh:mm:ss format
   DEC         =  Declination of the object in dd:mm:ss format
Both RA and DEC are in J2000.0 equinox. You need the RA and DEC you got from your catalog search in Part 1; do not rely on the RA and DEC given in the observing log. The printout from hjdcalc will be something like this. The command
   hjdcalc 2000-09-25 04:53:25.3 22:37:48 01:30:49
Will print out
   Object RA=22:37:48.0 DEC=01:30:49.0 (J2000.0)
   UTC 2000-09-25 04:53:25.30
    JD 2451812.703765
   MJD 51812.203765
   HJD 2451812.709105  (deltaJD=0.005340)
The first two lines above just give you back your RA, DEC, and UTC Date/time. This lets you make sure it got them OK. The following three lines give
   JD  =  Julian Date at mid-exposure (i.e., at Earth)
  MJD  =  Modified Julian Date = JD - 2400000.5  (you can ignore this)
  HJD  =  Heliocentric JD at mid-exposure
deltaJD = difference between the JD and HJD in days
The only thing that changes for your 25 observations is the UTC time at mid-exposure; the RA, DEC, and UT calendar date (2000-xx-xx) are all the same.

Record the just the HJD for each of your observations to the nearest 0.00001 days (= nearest 0.8 seconds)

Step 7: Compile your results into a single table

Now bring all the pieces above together and compile a single "master" table of your measurements. This constitutes the preliminary photometric results that will then be analyzed in the next part.

Your table of results should be formatted as follows:

 Image  UTCmid  HJDmid  Var  ErrV  C1  Err1  C2  Err2
   Image  =  name of the image (e.g., ccd000925.0014)
  UTCmid  =  UTC time at mid-exposure
  HJDmid  =  Heliocentric Julian Date at mid-exposure
     Var  =  sky-subtracted signal of the Variable in ADU
    ErrV  =  uncertainty in Var in ADU 
      C1  =  sky-subtracted signal of comparison star 1 in ADU
    Err1  =  uncertainty in C1 in ADU 
      C2  =  sky-subtracted signal of comparison star 2 in ADU
    Err2  =  uncertainty in C2 in ADU 

Because you all have different computer resources, I am going to demand the following "uniform format" just to save me time compiling the master data from all of your measurements:

Flat ASCII text format with no TAB characters.

It's that simple. No Excel, no SigmaPlot, no Kaleidograph, no nuthin'. I want flat, vanilla, just plain and boring-as-it-gets ASCII text with spaces between the values. Use just enough spaces to keep the columns lined up and readable, but be sparing. Please don't make me have to do extra work to decipher your data files. Thank you.

Step 8: Submit your preliminary photometry results

Deadline: Thursday, 2002 Dec 5

Now that you have your preliminary photometry results ready, please submit them to me via email to Prof. Pogge ( on or before 5pm on Thursday, 2002 December 5.

Be sure to indicate your name in the email. I will confirm receipt of the email when I receive it.

This milestone has two purposes

  1. It provides a deadline for completing the measurment portion of this lab before you get into the final analysis stages that is 1 week before the final due date (Dec 12). This lets me check your data and catch any problems before it becomes too late.

  2. It establishes a "milestone" in the project that you need to complete 1 week before the final report is due. This lab is not one that can be left to the last minute.

You're done with the raw measurements! On to the final analysis steps...

Advance to Part 4: Differential Photometry of CY Aqr

Go back to Part 2: Retrieve your CCD Data

Return to the CY Aqr Lab Main Page

Updated: 2002 December 1 [rwp]

Copyright Richard W. Pogge, All Rights Reserved.