Earthshine blog

RH 10%
T 37 F
wind ~20 mph
Moon rising in thin clouds. Later v. nice data (past 2456035.1 for example).
Azimuth near 90.0 degrees.
Dawn started at 5:10 HST.

Note 1: File with actual rising is B frame at 2456035.0738666 – could possibly show halo of BS against the mountain, as the BS rises, despite clouds on horizon – air between mountain-side and telescope probably quite clear.
Note 2: Rose only a few minutes after ephemeris time so we almost caught it rising from sea. Other nights have been 20 m delayed due to slope of Hawaii.
Note 3: started tracking at altitude -4 degrees – so the KS allows this, at that azimuth at least.

Note 4: Changed SHUTTERCLOSINGTIME in the Andor ini file from 20 ms to 100 ms to see if this helps for the ‘dragging problem’. Later changed this back to 20 ms since I got the usual amount of ‘dragged’ frames.

On the night 2456003 we happen to have a nice sequence of the Moon rising through a clear atmosphere and we can determine the instrumental alfa – i.e. the power of the PSF due to instrumental effects only.

Normally alfa depends on atmospheric and instrumental effects but on a steady night it is possible to gather data for several airmasses and then extrapolate to airmass zero.
The different colors are for different filters (more or less matching the wavelength – blue to red). The data were obtained from the fits of the EFM method.

We see a little bit of scatter in VE2 (red) but otherwise a remarkable agreement in intercept and slope. It would seem that the instrumental alfa is very near 1.724 for all filters. Since we have a ‘base PSF’ (with a power of 1.6) which we raise to alfa’th power to fit observed PSFs we see that the instrumental ‘real alfa’ is near 2.76. The diffraction limit is at 3. We are not saying we have a system that is diffraction limited – but any values of alfa larger than 3 would be impossible!

If we assume that the instrumental alfa is always fixed or decreases (large alfa means narrow PSF, small alfa means wide PSF) with time then measuring the alfa_instr on all good nights would allow us to perform some sort of quality check on the data for that night. Also, keeping an eye out for the largest alfa_instr ever will help us set better upper limits.

Such diagnostic testing as the above should be made a routine part of a future pipeline for reducing data.

Note: This is our second visit to night 2456003 – we have also looked at the same as the above but based on FFM, here. The value for alfa_instr is nearer 1.76 for the FFM, suggesting that there is method-dependence. However, FFM is currently our most troublesome method, of the FM methods.