[News] Astrometric Calibration
Erik Deul
deul at strw.leidenuniv.nl
Thu Jul 8 11:11:31 CEST 2004
Hi AW-er,
Here is a note on possible adaptionof the astrometric calibration
procedure in the current pipeline.
Erik
--
_/_/_/_/ _/_/_/ Erik Deul
_/ _/_/_/ _/ _/ Sterrewacht Leiden/Leiden Observatory
_/_/_/ _/ _/ _/ _/ University of Leiden
_/ _/ _/ _/ _/ P.O. Box 9513, 2300 RA Leiden
_/ _/_/_/ _/ _/ The Netherlands
_/_/_/_/ _/ _/ _/_/_/_/ Email: Erik.Deul at strw.LeidenUniv.nl
_/ _/ Phone: (31) (0)71-5275827
http://www.strw.leidenuniv.nl/~deul
-------------- next part --------------
Astrometric Quality Control Parameters
======================================
Here is a proposal for the list of parameters that can be used as quality control
parameters. For each parameter the name, an example output and sensible limits
are given.
Rms and max ref stars residuals
-------------------------------
- Rms and Max values in arcsec of the residuals between the
calculated positrion after astrometric calibration application and
the reference catalog positions.
- Example values from astrom (testrun)
0.428727 and 1.290292
- Limitations: Rms < 0.5 arcsec
Max < 1.5 arcsec
stdev on polynomial parameters
------------------------------
Statistical measures on the solution parameters of the astrometric
calibration.
- Statistics
+ First some statistics on the number of fitting objects. These
values are availble for the single frame reference objects and
for the inter-frame overlap objects. Note example is without
overlap use.
+ Example
Number of reference stars: 275
Number of overlap stars: 0
+ Limitations: Nref > 100
- Position
+ Second, information about the derived offset parameters of the astrometric
solution.
+ Example
arcseconds radians plate center
frame xoff yoff std xoff yoff std ra dec err
1 -2.78 2.04 0.26 -1.3e-05 9.9e-06 1e-06 2.5945597 -0.3620920 1e-06
+ Limitations: std < 0.4 arcsec
- Polynomial parameters
+ Third, information about the first and high order terms of the astrometric
solution, Note, example is for non-overlap case
+ Example
(arcseconds) (radians)
X Y err X Y err
-0.2383 0.000261 +/- 0.00013 -0.002369 2.594e-06 +/- 1.29e-06 X*1*Cheb0(F)
0.001006 0.2381 +/- 3.61e-05 1e-05 0.002366 +/- 3.59e-07 1*Y*Cheb0(F)
-3.276e-05 -7.776e-05 +/- 3.37e-05 -3.256e-07 -7.728e-07 +/- 3.35e-07 X**2*1*Cheb0(F)
0.0001374 0.0001613 +/- 1.55e-05 1.366e-06 1.603e-06 +/- 1.54e-07 X*Y*Cheb0(F)
5.862e-06 6.608e-05 +/- 8.41e-06 5.826e-08 6.568e-07 +/- 8.35e-08 1*Y**2*Cheb0(F)
+ Limitations: err < 0.3 / Nref / Pdeg
For OmegaCAM the set of polynomial parameters (particularly the first order)
should be strict and well known after commissioning. So limits on the range
of X*1*Cheb0(F) and 1*Y*Cheb0(F) can be established.
Covariance matrix
-----------------
- Characterisation of the quality of the astrometric solution is presented by the
covariance matrix of the solution parameters. In astrom this covariance matrix is
normalised to 1000 to make a representable set of values visible.
- Example
Matrix:
(radians)**2 (")**2
1000 0 989 0 0 984
0 1000 0 842 0 0
989 0 1000 0 0 951
0 842 0 1000 0 0
0 0 0 0 0 0
984 0 951 0 0 1000
- Limitations: Diagonal elements all 1000, lines neighboring diagonal
all 0's.
Residuals
---------
- Individual residuals of objects and reference. This is a deepening of the
first quality control parameter which is only a global measure of these
individual residuals.
- Example
This is a full description of the first few objects from the RESIDUALS table.
# 1 DRa Right Ascension residual [Deg]
# 2 DDec Declination residual [Deg]
# 3 Ra Right Ascension object [Deg]
# 4 Dec Declination object [Deg]
# 5 Xpos1 X position first object [Deg]
# 6 Ypos1 X position second object [Deg]
# 7 Xpos2 Y position first object [Deg]
# 8 Ypos2 Y position second object [Deg]
# 9 F1 Field number first object [Deg]
# 10 F2 Field number second object [Deg]
# 11 C1 Camera number first object [Deg]
# 12 C2 Camera number second object [Deg]
4.67403e-06 -1.98503e-05 148.896252 -20.732737 -1 -1 1539.47 23.6963 -1 0 -1 0
6.13633e-05 4.90936e-05 148.937179 -20.730045 -1 -1 960.152 65.9206 -1 0 -1 0
-6.7815e-06 3.59786e-05 148.985052 -20.729637 -1 -1 284.43 72.3624 -1 0 -1 0
-0.000305546 3.32008e-05 148.985371 -20.729634 -1 -1 284.43 72.3624 -1 0 -1 0
0.000115075 -6.66176e-06 148.986496 -20.728337 -1 -1 262.184 91.4185 -1 0 -1 0
9.21497e-05 2.76691e-05 148.967871 -20.724937 -1 -1 525.897 143.251 -1 0 -1 0
9.30769e-06 -3.25175e-05 148.924557 -20.724084 -1 -1 1139.5 154.804 -1 0 -1 0
-3.95176e-05 -1.28883e-05 148.863991 -20.723503 -1 -1 1996.41 163.188 -1 0 -1 0
-4.20375e-05 -5.40213e-05 148.866960 -20.721840 -1 -1 1954.53 187.785 -1 0 -1 0
- Limitations: This information allows to make plots, similar to those in the OAC report.
Individual object statistics
----------------------------
- As part of the application of the astrometric solution each object will have
an associated of statistical accuracy parameter: the rms from statistical part
(does not incorporate measurement errors).
- Example
Only a subset of the available column information from the OBJECTS table is presented here.
# 1 Ra Right ascension object in world coordinates [Deg]
# 2 Dec Declination object in world coordinates [Deg]
# 3 RMS_RND Random positional error astrometric solution [Deg]
# 4 ERRA_IMAGE RMS position error along major axis [pixel]
148.872928 -20.733011 1.65109e-07 0.0186
148.956969 -20.734256 1.27054e-07 0.0255
148.896252 -20.732756 1.36571e-07 0.0124
148.951663 -20.732212 1.23105e-07 0.0218
148.868455 -20.731741 1.71571e-07 0.0172
- Limitations: < 0.3 arcsec
Estimated cost
--------------
The estimated costs to add these quality control parameters and limit checks to the
pipeline are, based on the fact that all input to these parameters are already
available from the LDAC tools, two weeks programming for the implementation
and two weeks testing for verification.
Astrometric calibration using overlap
=====================================
Introduction
------------
At the moment the OmegaCAM pipeline processes individual CCD separately in the astrometric
calibration section of the pipeline. Only at the stage where the dedithered image is created
the individual CCD's (parallel streams in the pipeline) are synchronized.
For the purpose of deriving an astrometric solution using overlaps, the parallel threads have
to come together at an earlier stage (just before the ldac astrom program is called) and
can disentagle after the derivation. Because for astrometric purposes only catalolgs are used
the overhead in disk I/O will be small. Overhead is created because the astroemtric solution
runs on one CPU, while generally the data is spread over 32 CPU's (that have local disk storage).
Software Implications
---------------------
The astrometric calibration procedure (AstrometricCatalog.do_astrometry) will have to have a
synchronisation point. This can be generated by separating the procedure into two consecutive
procedures that are `called' from the commanding make and by rearranging the top-level.
This way the astrometric calibration application (second procedural part) will be very similar
to the photometric calibration applciation.
Estimated costs
---------------
To implement and test the modification to the pipeline python scripts will costs one full month
for implementation and two weeks for full testing.
More information about the News
mailing list