= Astrometry with the IPP =

The following page reviews my process of extracting astrometry information from Megacam images


----
=== Image Injection ===


{{{
ipp_serial_inject.pl --telescope CFHT --camera MEGACAM --workdir /data/ipp035.0/beaumont/reduce.sa_98/ --label sa.98 -- dbname beaumont
}}}


=== Pantasks Macro File === 
I use the following macro file (named input) to control injection, camera, and chip reduction with Pantasks.

{{{
macro setup

  $MODULES:0 = /home/kiawe/eugene/src/panstarrs/ipp/ippTasks
   module pantasks.pro

  # $LABEL = prpics.20081216
  # $POLLLIMIT = 100

  add.database beaumont
  load.hosts
end

macro load.hosts
  controller host add ipp023 -threads 4
  <etc etc>
end

macro check.books
  book list
  echo ""
end

macro register
 module register.pro
end

macro registerOn
 register.on
end

macro registerOff
  register.off
end

macro detrend
 module detrend.process.pro
 module detrend.stack.pro
 module detrend.norm.pro
 module detrend.resid.pro
 module detrend.reject.pro
end

macro chip
 module chip.pro
end

macro chipOn
 chip.on
end

macro chipOff
 chip.off
end

macro camera
 module camera.pro
end

macro cameraOn
 camera.on
end

macro cameraOff
 camera.off
end
}}}

Then, in pantasks:
{{{
(ipp035:~/data) beaumont% pantasks
inputno file /home/panstarrs/beaumont/.pantasksrc

Welcome to pantasks - parallel task scheduler


pantasks: input input
pantasks: setup
starting controller connection
Connected
HostID: 1
HostID: 2
HostID: 3
HostID: 4
HostID: 5
HostID: 6
HostID: 7
HostID: 8
HostID: 9
HostID: 10
HostID: 11
HostID: 12
HostID: 13
HostID: 14
pantasks: register
pantasks: chip
pantasks: camera
pantasks: run
pantasks: add.label sa.98
}}}

=== DVO Creation ===
The above commands produce the .smf files which can be injected into DVO. Adding images to a DVO database proceeds as follows:

{{{
<cd into working directory>
set x = `ls */*.smf` # selects all smf files
foreach a ($x)
  addstar -D CATDIR catdir.98 -D CAMERA megacam -update $a
end
}}}

Running addstar with -update adds new entries to the end of the DVO database. This saves time, since the catalog need not be repeatedly resorted. At the end of this process, running addstar with -resort re-arranges the database so that measurements of common objects appear consecutively

{{{
addstar -D CATDIR catdir.98 -D CAMERA megacam -resort
}}}

Next, I ran relphot to populate the secfilt DVO table. Note that this probably does not produce very good photometry measurements; this is just a workaround so that relastro can access the information it needs from the secfilt tables.

{{{
relphot -D CATDIR catdir.98 -D CAMERA megacam -averages -update -D GRID_TOOFEW 10
}}}

=== Relastro ===

Relastro has several different modes. I iterated between two of them. Update-objects updates values in the AVERAGE DVO tables by taking the weighted mean of each object's measurements. Update-chips compares, on a chip-by-chip basis, the measured stellar positions to the average positions. It uses this information to update the astrometric solution (in essence, it resolves for the polynomial to map focal plane coordinates to sky coordinates):


{{{
relastro -D CATDIR catdir.98 -D CAMERA megacam -region 0 360 -2 2
    -D RELASTRO_SIGMA_LIM .1 -update-objects +photflags -update
relastro -D CATDIR catdir.98 -D CAMERA megacam -region 0 360 -2 2
    -D RELASTRO_SIGMA_LIM .05 -update-chips +photflags -update -instmag -12 -8
}}}

I found that repeating these commands 2-3 times converged on a stable solution.


Some options for relastro are summarized here. These can either be provided on the command line or in your camera's dvo.config file:
 * RELASTRO_SIGMA_LIM
   * If set and nonzero, than measurements with photometry errors larger than this value are not considered in the analysis
 * RELASTRO_SRC_MEAS_TOOFEW 
   * If an object has this many detections or fewer, do not consider the object during the analysis. This is particularly effective at filtering out spurious measurements.
 * +photflags
   * During reduction, the IPP sets a number of quality flags. Those flags which correspond to problematic detections (e.g. saturation, cosmic rays, etc) can be excluded during relastro analysis by using this option. The specific bitmask defining which flags are problematic is listed in the Photcodes.dat DVO table as ASTROM_BAD_MASK
 * instmag
   * Set this to a range of instrumental magnitudes to consider. This can be used to filter out faint or saturated objects

=== Evaluating the Fit === 

There are several ways to ascertain whether relastro did a satisfactory job during fitting.

 * '''Textual output from relastro'''
 Relastro outputs lines like this to the terminal during --update-chips
{{{
image 3 : Nstars: 591
scatter limit: 0.826633 based on 403 detections; using 377 of 591 for fit
scatter limit: 0.290518 based on 403 detections; using 376 of 591 for fit
scatter limit: 0.288850 based on 403 detections; using 379 of 591 for fit
}}}

This says the following: For the third image (each chip is a separate "image"), there are 591 stars which pass all filtering criteria. By comparing these detections to each object's average position, the 3-sigma scatter is .826633 pixels. 377 detections have scatter less than this, and are being used to update the astrometry solution. This process repeats two more times, and the final 3-sigma scatter appears to be .289 pixels.

 * ''' DVO table values '''
The average (.cpt) dvo table has two fields -- RA_ERR and DEC_ERR -- which lists the positional uncertainty for each object (e.g. the error on the weighted mean of the measurements ~ rms/sqrtN). These values are updated each time relastro --update-objects is run. They are reported in arcseconds. These values can be extracted in DVO via the command

{{{
avextract ra dec dra ddec
}}}

The help information avextract --help fields seems to mislabel dra and ddec as being in degrees; I believe they are still arcseconds.