Solar Observation Calibration
Last Update: JeffMangum - 27 September 2007
Solar Observation Calibration Questions
- Solar amplitude calibration requirements?
- Solar polarimetry?
- Wide field solar polarimetry?
Solar Observing Discussion
For a parallel discussion on general solar observing issues, see SolarObserving
Solar Observing Calibration Sequence
has written a draft document describing the Calibration of Solar Observations
for ALMA. This document was reviewed by the Calibration Group at their January 11, 2007
On September 13, 2007, Richard Hills noted that it might improve calibration sensitivity at submillimeter wavelengths if one could put the solar filter in on only, say, half of the antennas to start with. This means that on the outputs which correlate those antennas with the ones which have no filter one is only loosing the sqrt of the solar filter attenuation. That should make the s/n problem much easier. For a subsequent series of measurements one could then swap over and do the other half of the antennas, for example.
Solar Observation Phase Calibration
(Extracted from an email discussion between Mark Holdaway and Tim Bastian)
The timescale for phase calibration should be driven by the demands of the solar astronomers -- if it is 1-2s, then the whole cal part of the cycle (move time concurrent with attenuator removal time, cal obs, move back time concurrent with attenuator insertion) can take 3-5 seconds. SOLAR observers could look at the sun continuously for 25 seconds, then do a cal, losing 5 seconds. That sort of sampling should be OK, though certainly not optimal. Maybe they'll regirid their time series of images onto a canonical 5s grid.
Scheme For Solar Phase Calibration:
We require that the solar attenuators will be amplitude & phase stable. There should be specific specs on their stability, uniformity across the aperture, and repeatability of insertion orientation, and anything else someone thinks of.
We have enough SNR to do fast switching with 30 mJy quasars. The attenuators will reduce the incoming radiation by a factor of about 25, and the warm attenuator will add about 300K of noise to the system temperature -- at low freqs, that would be about 100K, so the noise is 4x higher. Therefore, if we can track the phase on a 30 mJy source without attenuators, we can track the phase on a 30 mJy * 4 * 25 = 3 Jy source WITH the attenuators.
IF the attenuators are phase and amplitude stable, then we can solve for each attenuator's amp and phase (or delay) by observing a bright quasar (ie, we'd pick the brightest in the sky, assuming that the attenuator amp and phase are not direction dependent). We observe 3C273 with attenuator and without attenuator, say spending 3 s with, 3 s without. Repeat if you need more sensitivity (remember that fast switching needs like < 1 s for enough sensitivity --- but we'll need more sensitivity for this than for fast switching (imaging is not SNR limited, so we want good gains -- on the other hand, we'll have sensitivity to burn for self-cal to touch up on any gains we don't get right).
IF the attenuator gains are NOT time dependent, then we observe 3C273 as a service observation once every 4 months or something, and apply them from a lookup table. IF they are variable, then we may need to observe the gains on a very bright quasar every time we do solar observing. A worst case scenario would have it that we can only do solar observing when the sun gets close to 3C273 or similar source.
With all of these if-statements satisfied, the solar observing would work like this:
- Observe a ~50mJy phase calibrator 2-3 degrees away, for about 1 second (+/-), without the attenuator in.
- Move to the sun while inserting the attenuator.
- The antenna move time and the attenuator insertion time are comparable, we we don't really lose any time.
- We observe the sun for 10 seconds.
- We then move back to the phase calibrator while taking out the attenuator. IE, regular old fast switching, with no reliance on the WVR system.
Fast Switching normally does a global optimization for sensitivity, trading off fast cycles times which minimize phase errors (and decorrelation) and long cycle, which minimize time lost off source. For the sun, the optimization will be different -- towards shorter cycles, perhaps, to get better phases.
OF course, we CAN self-cal on the sun. The sun is dynamic, so we need to self cal with 1-10 s integrations. It is complicated, so we'll have trouble getting a good model for self-cal'ing, but more antennas helps us there.
Finally, Darrel points out "We need to do it for single dish as well". Yup. And mosaicing will work if we meet the right specs on the uniformity over the aperture.