Best*. Photometry. Ever. IV: Promotion and Results

In the last installment, we discussed our efforts to characterize the holographic diffuser and convince Palomar to install it on WIRC.  Today… results!

With Palomar and Heather on board, and lab proof-of-concept photometry in hand, we went to NASA for funds to support Ming’s research.  We applied for an Origins of Solar Systems proposal with Ming as Science-PI and me as administrative-PI, because as a CEHW postdoc Ming did not have PI privileges at Penn State. Ming wrote a beautiful proposal showing all the great things one can do with the diffuser at Palomar, along with supporting spectroscopic observations from other observatories.  


In November, we learned that our proposal had been accepted!  We were thrilled.  Ming was now (mostly) supported for 3 years, and had a major grant to put on his faculty applications.  Ming’s postdoc was up, so we went to the astronomy department administration and argued that now that Ming has grant experience, we should bring him on as soft-money faculty.  The university agreed, and Ming is now a Research Associate at Penn State.  

The panel review of our proposal was remarkably good. I don’t mean “positive”, although it was that, too — I mean that sometimes reviewers really read the proposal and “get it”, and sometimes they don’t (for better or for worse).  This panel “got it” which is a testament to Ming’s proposal writing skills (and the team’s comments, especially Heather’s and Jason’s–MZ).  In particular, there was a very fair weakness noted by the panel:

The diffuser has been tested in the lab but not yet used in actual observations with a telescope. The actual performance on-sky is somewhat uncertain. 

Indeed.  Not even a month after we read those words, Ming was doing the test that would determine if the concerns identified by the panel would hound us, or evaporate.

Installing new hardware on a workhorse instrument like WIRC is not a trivial matter.  After much back and forth we determined that the diffusers we needed were at the very limit of what our vendor could produce — they were not used to such tiny angles.  We requested a tophat diffuser that would turn point sources into uniform disks — the ideal shape for photometry.  


Eugene Serabyn’s team at JPL used their Palomar testbed to test the diffuser when it arrived.  The results were disappointing — the diffuser was Gaussian, not tophat, and highly speckled, not uniform.  The vendor was very understanding and after some back and forth agreed to send along a better one at no charge– but in the meantime we moved ahead with the in-hand diffuser.  The speckles were actually not going to be a problem because natural seeing at the telescope would wash them out, and the relatively coarse pixels of WIRC (15 micron = 0.”25) would bin them, anyway (and we could always defocus a bit, if necessary!)  The Gaussian profile had broad wings that would necessitate a wider-than-optimal extraction aperture, and that would give us excess background noise from the sky.  But for a test run, this would do, especially since Ming’s test run was coming up and the new one wouldn’t arrive for a while.

Next, Palomar had to install the filter.  WIRC is a cryogenic infrared instrument, meaning that the components are very cold, and in fact they are in a vacuum chamber.  Changing anything means slowly letting air in and letting the system come up to room temperature, which hardware really doesn’t like to do.  Do this too many times, and eventually some piece of delicate equipment (maybe your $300,000 detector!) will fail.  So in general one wants to minimize the number of times this happens by queuing up all of your repair and upgrade work until the next scheduled maintenance.  The next time things were due, in went the test diffuser.

Ming tells the story:

We made an on-sky test observation with the diffuser and WIRC on December 21, 2013 (Ming Zhao & Joe O’Rourke, a grad student of Heather’s). We were barely able to observe that night due to high humidity and fog passing-by, and were interrupted by fog later on. So the condition was not quite photometric and we could only observe a test star. Nonetheless, we took 40s exposures continuously for ~3.5 hours, and the PSFs were stabilized quite well by the diffuser throughout the observation. The result, analyzed using standard aperture photometry, is shown in the figure. 

diffuser_test_run (2).png

The top panel shows the differential light curve of a Ks~10 magnitude star normalized against 8 nearby references. The raw RMS scatter is 1235 parts per million (ppm, or 1.235 millimagnitudes). The middle panel shows the scatter of 9-data point bins (8 min), and the bottom panel shows the scatter of 30-point bins (28 min). The light curve still has some time-correlated noise so the scatter doesn’t follow 1/sqrt(N) exactly. The 28min-binned RMS is 110 ppm in the bottom panel. And 100 ppm is roughly where our current noise floor is at. The bottom line is that we have reached an RMS of ~100 ppm with 30-min bins for a Ks=10 star, which is one of the best precision in the NIR from ground to our knowledge. 

(For reference, Kepler achieved 159 ppm on a Kp=10 star in 30 minutes — JTW)

The most important thing is that we should be able to reach this precision regularly, even under non-optimal conditions, and even with this sub-optimal diffuser. So we don’t have to bet on luck any more — a much more settling situation!  We hope we can get exciting scientific results in the upcoming observing runs this year, despite all the tough luck on weather and instrument we have had!

Also, the new, tophat, smooth diffuser has arrived and Heather has scheduled another lab test with Eugene’s team.  It will have a much more compact PSF with sharper boundaries, which means fewer pixels, which means less background, which means higher precision.  

So why does the diffuser work so very well?  Ming has put together a great animation showing the “before and after” effects of a diffuser on the distribution of light.  As you’ll recall from Episode I: The Problem, variable flux on a few pixels makes photometry very difficult.  What you want is perfectly stable poor focus or seeing. 

Click below for links to the results in movie form:  typical seeing, typical defocussed seeing, and the diffuser.  Note the change in pixel scale on the rightmost image:

Click me!

Screen Shot 2014-05-02 at 9.54.58 AM.png

Click me!


Next time:  A big step back, followed by a bigger step forward and… what does this mean for the possibilities of the best photometry from the ground?

Images: Penn State Research Associate Ming Zhao, outdoors; Caltech Assistant Professor Heather Knutson, mostly outdoors; photometry by Ming Zhao; animated gifs by Ming Zhao.

Leave a Reply

Your email address will not be published. Required fields are marked *