This comparison involves matching northern and southern scans of six 2MASS tiles:
1123 to 1128. Three datasets are included - one northern set (101123), observations
from Feb 13, 1999; two southern datasets (301123), made on 22/12/1998 (designated
set A here) and on 24/2/99 (set B).
Sources were matched using the same criteria employed with tiles 900-907, with one
exception:
(RA, Dec) positions coincident to better than +/- 0.001 degrees; J < 18.5;
and the comparison is limited to PSF magnitudes with Delta(mag) = mag(S) - mag(N) < 0.9
magnitudes.
Magnitude/sigma (PSF) plots for each dataset are included for completeness,
but are generally of little interest.
There is no prior rejection of flagged sources. As a result, band-filled sources contribute
the second, smaller-sigma, sequence in those plots. Since these measurements
are largely fictitious detections, setting sigma to a standard value (eg -9.99) might
be more appropriate.
Magnitude residuals in J, H and K are plotted against position (RA, Dec) for both scans
of tile 1123. There is no significant trend with RA; there is a suggestion of some
systematic variation at the <0.02 mag level with declination (i.e. time).
Source counts for the northern field (all sources) and the paired dataset are
given in the postscript tables. Approximately 50% of the "sources" listed in these
tiles were detected only at H, with J, K magnitudes > 75 and sigma = 8.888
comparison of mean relations: South (A) - North
comparison of mean relations: South (B) - North
The above postscript files compare the mean Delta-mag for both
sets of scans of the six tiles in question. As before, the reference magnitude
scale is set by the northern observations.
The overall behaviour is similar, if somewhat less pronounced, to that found for tiles 900 to 907.
Each magnitude comparison shows a trend toward mean positive residuals at magnitudes
fainter than ~14.5. Inspection of the individual residual plots shows that there are
more large positive residuals than large negative residuals - as one might
expect, given the distribution of noise sources. The magnitude
cutoffs were set at +/-0.9 magnitudes to minimise the influence of the latter on the
mean. However, note that the distribution of residuals is skewed even at |delta mag| < 0.3 mag
for K > 14, J,H > 15. That accounts for the drift to mean positive residuals.
Rather than detector-linearity problems, this effect can be attributed to the fact
that sky-noise is linear, while the magnitude plane is logarithmic: hence,
for a source, flux f, in the presence of sky noise, rms=S,
The following postscript files plot magnitude residuals against colour. The datasets are limited to stars with 10 < K < 13.5 (Northern field), plotting delta-J and delta-H vs (J-H) [North] and delta-K vs (J-K)[North].
Allowing for the individual systematic offsets, tile-to-tile trends are
broadly consistent:
Positional residuals are plotted for each North/South comparison in the
following postscmript files.
Residuals are generally less than 2 x 10-4 degrees (0.7 arcsec)
peak-to-peak, with no systematic
trends evident with either RA or magnitude. Larger residuals are due to
spurious sources near bright stars.
There are quasi-periodic, systematic residuals
as a function of declination (scanning direction) in both RA and Dec.