legacysurvey / obsbot

Script to plan subsequent Mosaic and DECam observations based on current conditions

obsbot

Script to plan subsequent DECam observations based on current conditions

obsbot's Issues

mosbot: if pixcnt error seen, try another amp

avoid, eg, astrometry shifts

copilot: depth factors for whole night

Is there an easy way to look at the depth factors for the whole night? I wanted to make a list of all the frames with depth < 0.3 so that we can add these to the bad_expid.txt file.

Should assume relative differences in seeing (eg, g-band is usually 0.1" worse than r-band)

copilot MOSAIC: check fits data md5sums and complain loudly if repeats are found!

(for f in rawdata/mos3.649*.fits; do fitsgetext -i $f -e 1 -D -o - | md5sum | sed "s+-+$f+"; done) 2>err > fitsmd5sums

mosbot: tonight.sh and script files should complain loudly if expected datetime is very different than NOW

eg, if tonight.sh is re-run without re-running mosbot

mosaic focus plot: y limits

When the seeing is worse than 2”. all the points on the copilot focus plot end up off the top boundary. Could we please edit the copilot focus plot code so that it rescales appropriately based on the measured FWHM measurements?

copilot: if fewer than some minimum number of PS1 stars match, bail out (we've failed to find shift)

copilot should look at multiple CCDs

switch to Neff formula with 1.15 power

mosaic: transparency

The transparency values between copilot and mosstat are still discrepant.

decbot: add "blocks" of exposures that should be observed together

eg, g and r back-to-back.

decbot: estimate current sky color and use that rather than differences in nominal sky mag

copilot: make MISSING IMAGES mark more glaring

copilot: count & report total number of tiles

1. When copilot is run with the -- night option, can we have the total numbers of good tiles be listed (i.e., at least the pixcnt ones subtracted)? That would make it easy to write in the summary totals on the logs.

E(B-V) average over field, not just at boresight

For obsbot, we’d want a median E(B-V) over the field rather
than just at the boresite.

Specifically, the tiles file for DECam takes the median of extinction
values spaced every 0.1 deg that fall within 1.0 deg of the boresite.
To avoid spherical trig problems, I placed these positions around
a pole, and then rotate the pole to the boresite.

mosaic focus crash

tonight mosaic_focus.py caused a copilot crash on our focus frame. Running manually

python mosaic_focus.py /mosaic3/data3/observer/20160510/mos3.88174.fits

Gives the error message:

mosaic_focus.py:203: RuntimeWarning: divide by zero encountered in divide
wt = 1. / Ysig

when outside the PS1 footprint, give useful error message rather than exploding

Copilot: Use rotation measurement to get to same system as mosstat

Convert dRA,dDec,rotation in chip imX to dRA,dDec in im16 to match mosstats. (Also apply overall copilot-to-mosstat im16 difference).

copilot & mosbot: use MOS3_DATA environment variable as default rather than "rawdata"

mosbot: add command-line args to set the beginning pass number and exposure time

copilot: haven't seen data in N minutes line not showing up?

copilot: plot layout issues

Would it be possible to make copilot put the points on top of the labels, rather than the other way around? Currently the labels hide the points. Also, for the image exposure number labels - i wonder whether it might be better to only print the full number for the first and last points on the plot, and then just label using the last three digits for the rest of the points.

When the seeing is > 2.3” (roughly) no seeing value appears on the plot. Perhaps it is being written off the plot?

mosbot: warn about upcoming >10-degree slews

Update z-band zero-points for each amplifier

Arjun has measured the following z-band zero-points from photometric good-seeing
frames on the night of 9/10 Feb 2016 as follows:
im4: 26.406
im7: 26.609
im11: 26.556
im16: 26.499
Since it's not measured for the other 12 amps, just use the average of
these four at 26.518 .
These numbers already updated in the IDL mosstat.pro code.

Copilot: handle Mosaic3 images with "VR", "R Harris", etc filters

Crashed Nov 9/2016

mosbot should delete all old scripts from its scriptdir by default

Decbot: should include slew time in overhead calculation

mosbot: should use current sky brightness to set z-band saturation exposure time limit

For Mosaic3 images, is the exposure number supposed to start with a 3?

A raw mosaic 3 file, e.g. mos3107334.fits says that its EXPNUM is 3107334. Yet elsewhere, the exposure number is recorded as 107334. So, is 107334 the real exposure number, and mos3 the instrument designation, or is 3107334 the real exposure number and mos the instrument designation?

This is not just a hypothetical, because the exposure numbers are, in practice, being recorded with two different numbering conventions. For example, the bad_expid.txt file uses the latter convention.

mosbot: tonight.sh should skip exposures that were supposed to happen in the past!

copilot: by default, read ALL images from the $MOS3_DATA directory

copilot: Focus plot: summary text & seeing in title

Focus plot:

• could the summary plot that is put up list the best fit focus as the geometric mean of the X and Y fits and also list the seeing value derived? These should be at the top of the plot.

Problem with dRA,dDec layout

As per Arjun's report,

The observers have been noticing that sometimes the most recent dRA and dDEC values are not being printed at the bottom right corner of the copilot plot bottom panel. The problem is intermittent and may have something to do with the scaling of the y-axis of the plot and the location of the most recent (dRA,dDEC) point on the plot.

ALSO, consider flipping the sign of the printed dRA,dDec so that with DECam it is the correction that needs to be applied (ie, the values the observers need to read to the Telescope Operator)

copilot for Mosaic should output a Mosaic nightlystrategy command-line

Take g+r exposures back-to-back

Copilot plot during the night

this has the pass number in the second panel overlap the sky value. Could we put the sky value text of the last exposure just below the actual point? Also, is there some way of labeling the zero line with the “nominal” sky value? I suggest putting this in the middle of the plot just below the line.

• it would be great to label the points in the exposure time panel with the depth factor, perhaps by just adding a label above each point as it is plotted. That way if the depth factor is less than some number, we can ask the observers to log that in the bad_expid.txt file.
• The transparency values between copilot and mosstat are still discrepant. Any idea why? Or did i just miss / not read a number of e-mails on this topic ;)
• Having copilot print out the dRA and dDec offsets needed to correct the pointing at the start of the night would be very useful. It is hard to find these numbers in the huge amount of ascii output, so i usually just eyeball it off the plot. It would be nice to have this listed somewhere on the plot.

Copilot: nightly summary plot

• this still shows the last value of the data points in the various panels. This is fine, but not necessary in all except the bottom panel, which has the image numbers.
• could we also list the total number of good images taken = last image number - first image number + 1 - number of bad images,
  where number of bad images = number of images with PIXCNT errors + number of images with depth factor < 0.3 + number of images with seeing > 2” (David can check me on these limits)
  This number is useful, since it is what we try and enter in the summary page of the logs. In fact, if copilot can keep track of these numbers, perhaps it can also just update the bad_expid.txt file automatically, without requiring the observers to do that.
• could the copilot summary plot also compute the median seeing value, transparency, sky brightness for the night? Lori Allen actually wanted these numbers for each quartile of the night, but perhaps that is overkill to write down on the plots.