UPDATE of AGASC1.4 to AGASC1.5

Recalibration of AGASC1.4 from ACA MAG FITS from ACA DATA to Produce
AGASC1.5

Paul J. Green, Tom Aldcroft, Anil Dosaj
23 May 2001

See /data/mp2/ACA_cal/apr01/ for data, code, and notes.
This file is /proj/asc/www/LOCAL/mp/html/agasc1p5.html

Occasionally, a poorly estimated AGASC magnitude has resulted in failure to acquire a star. A larger problem is that the padding on star magnitude errors in the OFLS Characteristics means that too many stars are being rejected from consideration, so that targets are either rejected, or have a number of guide stars returned from the OFLS that is less than optimal. Given that the Star Selection Algorithm (SSA) Working Group is planning to update the OFLS Star Selection Algorithm parameters to extract candidate guide stars from AGASC to a magnitude of 13 (rather than the previous value of 11.5), many more candidates would be considered potentially spoiled by these extracted stars, given the large MAG_ACA_ERR values of most stars.

Inclusion of Tycho2 data to a completeness limit of V=11.5 (rather than the previous V=10 limit of Tycho) will significantly lower the magnitude and position errors on an additional 1.5million stars in AGASC. The most significant improvement in ACA magnitude estimates results from the accurate tabulated colors available.

We also have the opportunity to recalibrate the AGASC magnitude estimates using a larger number of observed ACA magnitudes. Using a GRETA script, flight ACA data were extracted for about 8400 stars observed with the ACA from 1999-Sep to 2000-Oct. (Thanks to Jeff Shirer for creating and running the script). These data were then correlated by Tom Aldcroft with CXC Level 1 aspect products to determine AGASC catalog information for each star. Approximately 5700 stars were thus identified. These data showed that there was no mean offset between observed and AGASC1.4 catalog magnitudes. Neither was there any correlation in magnitude difference with either magnitude or time. There was however a noticable systematic residual (about +/- 0.2 mags) in magnitude difference versus (B-V) color. For details see Tom's January 2001 report.

The ACA star observations were obtained while in Normal Pointing Mode and while on Kalman filtering. We have further analyzed these data to recalibrate the estimates of MAG_ACA in AGASC1.5. Removal of repeat observations of the same star cull the above list down to 2737 unique stars with ACA observations matched to cataloged AGASC stars. 798 of those have only PPM data in AGASC1.4. PPM mags are less accurate thatn Tycho mags, and most PPM data will be replaced with data from Tycho2. However, we recalibrate the MAG to MAG_ACA conversion for Tycho stars, PPM stars and GSC1.1 stars below, as best we can with the available data.

Based on this new ACA data, I've done polynomial fits to predict ACA mags for all stars, to recalibrate the MAG_ACA values. We use the median ACA mags rather than the mean. For a primer on how MAG_ACA values are calculated, see MAG_ACA update pseudocode. Essentially, I fit (ACA - V) as a function of COLOR1. COLOR1 in Tycho is (BT-VT) where Johnson (B-V) = 0.850*(BT-VT)

New Tycho fit results are as follows:

ORDER       4th        3d           2d          1st          0th
--------------------------------------------------------------------
C0       0.456604    0.428638     0.375402     0.513502    -0.173217
C1      -1.26040    -0.774029    -0.364786    -0.889533
C2       1.69743     0.283002    -0.354679
C3      -1.61554    -0.267284
C4       0.402788    
X^2(red) 0.04014     0.040665     0.041344     0.045255    0.169381
--------------------------------------------------------------------
                     AGASC1.5

The difference between AGASC1.5 magnitudes and observed ACA magnitudes as a function of Johnson (B-V) color for this sample of 1939 is shown below. I've chosen the cubic polynomial fit for AGASC1.5. Overall, the residuals are not much worse than the 4th order fit, but the cubic actually looks better at the blue end.

Statistics are as follows for 1939 ACA-observed stars using the 3d order fit:

          OBSERVED  NEW CALC  TYCHO   COLOR1  NEW-OBSERVED
          ACA MAG   ACA MAG    MAG              MAG DIFF
 MED       9.125     9.131    9.364    1.002    -0.013
 MEAN      8.896     8.896    9.139    0.908     0.000
 SIGMA     0.888     0.869    0.938    0.466     0.202
 MIN       5.750     6.017    5.800   -0.161    -2.107
 MAX      10.688    10.860   10.986    2.399     1.988
 RANGE     4.938     4.843    5.186    2.560     4.095

Analysis of the distribution of AGASC1.5 magnitude differences for these stars shows that they are non-gaussian. The absolute value of magnitude difference was distributed as follows:

68%    < 0.085 mags
95%    < 0.3   mags
99.7%  < 1.5   mags

While the mean difference is 0 above, the 1-sigma dispersion in the mag difference is 0.20mag. This figure illustrates the inescapable dispersion involved in a transformation based only on B,V colors to a camera system with a much redder bandpass.

Finally, the new fit shows no signs of being magnitude dependent:

Important points:

1) A few percent of stars will be rather extreme outliers (a magnitude or more off) even when colors are known. This may be due to extinction, or unusual spectral features, or variability. The outliers in this dataset have Observed-Predicted extremes of -2.1 and +2.0 mag.

2) I made the fit and plot above using stars with cataloged (B-V) colors. Many stars in AGASC1.5 (nearly all that are fainter than 12th) have no cataloged colors, so a color of (B-V)=0.7 is assumed. The dispersion in a observed-predicted plot for such stars will be significantly larger. However, more than 90% of stars bright enough for guide star selection will be Tycho stars with measured colors. Inclusion of Tycho2 data should yield good color and mag data 90% complete to V=11.5

MAG_ACA CALIBRATION FOR PPM STARS

If any MAG values listed in AGASC1.5 are from the PPM catalog, they may have crude colors based on spectral type (see earlier AGASC documentation).

New PPM fit results are as follows:

ORDER       4th        3d           2d          1st          0th
--------------------------------------------------------------------
C0       0.381712    0.389887     0.478573     0.717156     0.238356
C1       1.10085     0.922156     0.221899    -0.677880
C2      -2.13413    -1.60675     -0.550461
C3       0.891056    0.407737
C4      -0.135640    
X^2(red) 0.164025    0.163930     0.167331     0.185935    0.287951
--------------------------------------------------------------------
                     AGASC1.5

4th order worsens the fit, and 3d order is an insignificant improvement, so we choose the 2d order fit. C0= 0.478573, C1= 0.221899, C2= -0.550461 Even so, the scatter is large, given that the color information is poor.

Below you can see how the PPM colors form a picket fence, since each spectral subtype listed in the PPM was assigned a corresponding color.

MAG_ACA CALIBRATION FOR GSC1.1 STARS

Finally, the poorest data, for the fainter stars, comes only from the HST GSC1.1, which has NO color information. Firstly, the MAG listed in GSC1.1 is in several different bandpasses. We attempt a conversion to V mag using an assumed typ[ical color of (B-V)=0.7 along with the conversion values provided (see earlier AGASC documentation). It is worthwhile to convert to V mag from the various GSC mags:

           ACA-MAG  ACA-V
 NUM        2691     2691
 MED      -0.128   -0.077
 MEAN     -0.148   -0.063
 SIGMA     1.194    1.153
 MIN      -4.962   -4.458
 MAX       4.072    3.967
 RANGE     9.034    8.425

Below you can see how the derived V mag appears to be virtually uncorrelated with the observed ACA mag.

We cannot examine and fit the difference between observed ACA mag and V mag as a functino of color, since we have no color information. However, taking the difference between the observed ACA mag and the GSC1.1 V mag shows a strong trend with magnitude.

We do NOT correct for this because it is SPURIOUS. The apparent trend is caused by the imposed faint limit of about 10.2 on the ACA mag, illustrated with the red line above. When Vmag is about 10, only the redder stars (those brighter in ACA than in V) were observed as guide stars in the ACA (all these had colors from either PPM or Tycho). If we actually used this to convert GSC V mags to MAG_ACA, we'd estimate ALL GSC mags to be sufficiently bright to use as guide stars, e.g., MAG_ACA=9.95 for GSC V=15. Even the mean derived offset using the whole sample is suspect. If we were to try limiting to V<8 to avoid the imposed ACA faint limit, then the BRIGHT limit comes into play, i.e., red stars with V<8 were NOT observed since they're too bright in ACA. So at V=8, the mean ACA mag will be FAINTER. To avoid the bright and faint selection limits as much as possible, we restrict our sample to 7<V<9, whereby we derived a simple offset of C0=-0.65, i.e.

MAG_ACA= V - 0.65

INCLUDING GALAXY CLASSIFICATIONS for AGASC1.5

UPDATING SPOILER CODES for AGASC1.5

RELEVANT STATISTICS for AGASC1.5

*********** REMAINDER AWAITS RECALIBRATION OF AGASC1.5 **********

Stellar Surface Density The average stellar surface density of unspoiled stars brighter than MAG_ACA=10.2 with color information ((ASPQ1=0, CLASS=0, C1_CATID.ne.0) is 9.5 stars per square degree in AGASC1.4 Near the galactic poles (b>80deg), where the stellar surface density is lowest, there are 4.1 stars per square degree. The desired figure of merit (FOM) of 5.1 per square degree over 95% of the sky is thus not quite achievable with these selection criteria from current catalogs, and may not ever be (i.e. we are already nearly complete). The current Chandra guide star selection includes stars without TYC colors or PPM SpType information, which boosts the surface density, but at this limiting ACA magnitude, such colors are available for 98% of stars.

  
 b1	b2	Nstars  Ndeg^2	stars/deg^2
		w/colors	w/colors
----  -----     -----   -----   --------------
 80	90	1285	313.3	4.10
 55	65	8405	1797	4.68	
 25	35	24951	3113	8.02	
-90	90	393623	41253	9.54

Changes in Selected Guide Stars from AGASC1.3 A key measure of the effects of this update is the change in selected stars between AGASC1.3 and AGASC1.4. Roughly 1/3 of the Cycle 1 observations will have different star sets chosen by the CXC MP Star Fidlight Evaluator (mp_sfe) for the new AGASC 1.4 mags. How do the stars chosen in these fields differ?

  
All Fields	AGASC1.3	AGASC1.4
		---------	---------
mean mag:	8.448		8.630
mean fom:	0.15308		0.31221
min fom:	0.13669		0.13721
max  fom:	0.32571		0.32755
  
Number of
guide stars:	AGASC1.3	AGASC1.4
		---------	---------
5		634  (94.9%)	 610  (91.3%)
4		 22  ( 3.3%)	  33  ( 4.9%)
3		  9  ( 1.3%)	  17  ( 2.5%)
3		  3  ( 0.4%)	   8  ( 1.2%)

FOMs and the number of suitable stars per field brighter than MAG_ACA=10.2 looks slightly worse in AGASC1.4 Since AGASC1.4 is simply more realistic, that's nature's fault, not ours. Here's the breakdown on the number of individual target fields whose stars change between AGASC1.3 and AGASC1.4: