SNAP status report for Aug 28, 2003

Michael Richmond

I'm continuing to work on a "pipeline" to simulate photometric measurements made with the SNAP camera. The goal is to identify and characterize systematic errors. My pipeline will be much simpler than the full end-to-end simulator which the SNAP project will undoubtedly construct.

Recent developments:

• Added model for angle-of-incidence as a function of position in the focal plane, based on Alex Kim's document Effective Filters. The model is a simple linear fit to angle versus radial distance from center of optical axis.
• Added model for shift in passband as a function of angle of incidence. This is also taken from Kim's document.
• The snapshot routine now calculates "observed magnitudes" for stars on the fly, using
  • the stellar spectrum
  • the fiducial filter passband ...
  • ... which is shifted according to angle of incidence
• Output from the snapshot routine now includes over twenty quantities, so that residuals in photometry can be correlated with just about any variable

I used Alex Kim's model of the fiducial filter passbands, based on a B-like filter; filter N has wavelengths which are shifted by 1.16^N. Here they are: the even filters:

To check my work, I made a test: a snapshot of stars in the Northern SNAP Field. I compared the "observed magnitudes" to the "synthetic magnitudes"; the latter do not suffer from any angle-dependent shifts in passband. We expect that

• hot, blue stars should have brighter observed magnitudes if the passband shifts to the blue
• cool, red stars should have brighter observed magnitudes if the passband shifts to the red

I arbitrarily decided that the SNAP filters would all be designed to have a desired passbands for light incident at a nominal angle of 0.22 radians = 12.5 degrees, which occurs about halfway from the inner to outer annulus which defines the active focal plane. Now, according to our model of passband dependence on incidence, we expect

• angles smaller than the nominal will shift passband to red ... and so red stars should be brighter in inner regions of focal plane
• angles larger than the nominal will shift passband to blue and so blue stars should be brighter in outer regions of focal plane

As you can see, hot A stars have negative residuals -- are brighter than expected -- in the outer regions of the focal plane; cool K and M stars have negative residuals in the inner region of the focal plane. The code is evidently doing the right thing so far.

What's next?

First, I'll need to add some options to the simulator:

• individual zeropoints for each chip
• simple (?) variations in sensitivity across each chip

Then I'll start to generate results for noiseless observations. There will come a time when I want make graphs of many quantities versus other quantities, and graphs showing residuals as a function of position on the focal plane, position on a chip. I have looked for tools to generate such graphs, but haven't found any which seem to fit my immediate needs, so I'll probably end up writing some -- that will take several days.

In the more distant future, it will be necessary to add noise to the simulations:

• photon noise from the stars
• photon noise from the sky
• readout noise from the detectors

After that, a second round of simulations to verify that all the new features are correct.

The end goal is to determine how many observations are required to identify and characterize the different sources of photometric errors, and how best to arrange those observations.