Calibrated magnitudes need to be fixed.

[mkelley]

Attached is a plot and table from the first full run of the pipeline. I'll need someone to run an independent check on from frames to help me figure out where to look for errors. big-table.txt

[mkelley]

Updated plot. Excursions tend to go towards fainter brightness estimates, or are all the brighter estimates bad?

[bodewits]

So why are there two populations in both the 2" and 3", split by about 2 mag?

On 4/2/17 8:56 AM, Michael Kelley wrote:

Updated plot. Excursions tend to go towards fainter brightness estimates. m-vs-rh https://cloud.githubusercontent.com/assets/1138643/24587312/1f0f3830-1782-11e7-812e-a71be73f61aa.png

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/mkelley/41P-LCO/issues/25#issuecomment-290984707, or mute the thread https://github.com/notifications/unsubscribe-auth/AZKjPGV5oC1ajI6xuPJKYlJPGnWnRWUrks5rr5r2gaJpZM4MwmQp.

[mkelley]

Unknown. We also don't know which set better matches the comet, if any.

[mkelley]

@bodewits reports there is a correlation with telescope aperture and suggested it may be pixel scale related. I found and fixed a bug calculating the aperture size in pixels which was used for the background correction (issue #31 ).

[mkelley]

That definitely improved things (by about 0.5 mag), but still an offset.

[bodewits]

There is definitely a correlation with FWHM thus seeing but it is inverse from what I'd expect, i.e. the comet is brighter when the seeing is larger:

[bodewits]

In addition, the background has been highly variable since rh < 1.06 which is when our campaign started. must be the effect of the three different exposure times: use only the longest exposure time.

[bodewits]

Experiment: i only consider mh for t_exp > 35 sec

The correlation with seeing is still there so this is not a selection effect:

[bodewits]

So finally, why is there so much scatter in last couple of days? Because the FWHM varies a lot:

Or a close up:

(we're currently only observing with McDonald. There is newer data coming in acquired with the 0.4meters.)

[mkelley]

Is this FWHM of the stars or the comet? The FWHM of the stars depends on integration time and delta due to non-sidereal tracking.

[bodewits]

listed as mean(FWHM), following NMatch

[mkelley]

This is the mean FWHM of the background sources, which are indeed smeared due to non-sidereal tracking.

[mkelley]

This means that the photometry of the background sources needs to be verified, since the calibration relies on it.

[bodewits]

I will verify how the calibration differes for the 10-30-60s exposures.

On 4/8/17 10:25, Michael Kelley wrote:

This means that the photometry of the background sources needs to be verified, since the calibration relies on it.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/mkelley/41P-LCO/issues/25#issuecomment-292721430, or mute the thread https://github.com/notifications/unsubscribe-auth/AZKjPPCZ-K9_EPNcygHeG4WQml8NZ1yAks5rt5jKgaJpZM4MwmQp.

[bodewits]

if that is so, we should see a difference between 10-30-60s exposures

[mkelley]

Here is a plot of nearly all the stars measured to date in the month of April. Δm is the calibration factor (m_panstarrs - m_instrumental), FWHM is the estimated FWHM of the object. The data are colored by file. What I see is that for individual files the Δm is essentially independent of the estimate FWHM. However, there are a few images that cluster together in an unusual Δm-FWHM area.

[mkelley]

This plot is comet apparent magnitude colored by telescope. The two 0.4 m cluster together:

[mkelley]

From Tony on April 12:

Hi Mike,

I’ve tried to track down the problem with the two levels of brightness.

It comes down to the calibration coefficients between the datasets from different sources.

The fluxes for observations in the same timeframe seem to agree pretty well. However, when you convert to magnitudes, the “elp” datasets end up being fainter than the “ogg” and “tnp” datasets. Since the fluxes agree, it seems to be caused by the zero point magnitudes that you get from the panstarrs calibration process being different by ~1.5 mags. I don’t know what the pipeline is doing, but it seems to be pretty systematic, so it looks like there is a problem with whatever coefficients relating to the different sources.

Alternatively, it might be that zero points are done right, but the fluxes are incorrect, but coincidentally agree between the different sources.

[mkelley]

Dr. Knight has been assigned some tasks to check the calibration.

[mmk8a]

I have solved the discrepancy in magnitudes: LCO's table of aperture photometry calculates photometry in pixels of a specified radius even though their documentation says it is arcsec. As a result, there are considerable differences in aperture sizes from telescope to telescope (pixel scales are 0.387-0.389 arcsec/pix on the 1-m telescopes and 1.16 arcsec/pix on the 0.4-m telescopes).

A rudimentary lightcurve is attached. 1-m telescopes are plotted as squares, 0.4-m telescopes are circles; color varies by telescope. I made this by measuring the comet's flux in a 3 arcsec radius apertures in all the frames I have on my machine (most data through the end of March plus one night last week), converted to magnitude, made an airmass correction of 0.1 mag/airmass, then took the average of the magnitudes for a given night. I then used the average "scmean(dm)" from Mike's calibrate.csv file to get an approximate zero point for the night and applied this to my instrumental magnitudes. There is some scatter in the plot, but this was expected given that the zero points are still tied to the pixel scale, and there may be a double accounting for airmass (they are probably inherently corrected for by scmean(dm)).

Despite the scatter, I think this is now understood, and with some tweaks to the pipeline we should start seeing sensible results. I have emailed Tim to make him aware of the pixel vs. arcsec issue.

[mkelley]

:100: And for the record, the dm already includes the airmass correction.

[mmk8a]

Taking out my magnitude correction didn't make a big difference (attached). Hopefully standardizing the aperture sizes in arcsec will do the trick.

[mkelley]

Revised photometry method in PR #37 , but still not right :(