Abstract

1
Extending the ICRF to higher radio frequencies
Astrometry at 24 and 43 GHz C.S. Jacobs (JPL),
P. Charlot (Obs. Bordeaux), D. Gordon (GSFC) ,
G.E. Lanyi (JPL), C. Ma (GSFC), C.J. Naudet
(JPL), O.J. Sovers (RSA/JPL), L.D. Zhang (JPL)
and the KQ VLBI Survey Collaboration
Abstract Astrometric observations of distant
active galactic nuclei (AGN) have been used to
construct quasi-intertial global reference
frames, most notably the International Celestial
Reference Frame (ICRF) which now forms the basis
for all astrometry including deep space
navigation. The ICRF frame was defined using X-
(8.4 GHz) and S-band (2.3 GHz) observations
collected over the past 20 years. There are
several motivations for extending this work to
higher radio frequencies, namely, to construct a
more stable frame based on more compact sources,
to provide calibrators for phase referencing, and
to support spacecraft navigation at higher
frequencies. As a first step toward these goals,
in 2002, we began a series of survey observations
using the Very Long Baseline Array (VLBA) of ten
radio telescopes at K-band (24 GHz) and Q-band
(43 GHz). Each session covers the full 24 hours
of right ascension and covers declinations down
to the VLBA's southern limit (approx. -30 deg).
Analysis of the first three sessions has produced
a full sky catalog of 108 sources with median
formal position uncertainties for K-band of 210
and 280 µas in RA cos(dec) and Dec, respectively.
For Q-band the result is 292 and 404 µas in Ra
cos(dec) and dec, respectively. Group delay
residuals were excellent at 20 psec WRMS. An
external comparison of K-band frame to the
S/X-band ICRF shows systematic errors at about 2
times larger than the formal precision. These
systematic differences include a zonal trends vs.
declination. We suspect that these errors are due
in part to the North-South geometric weakness of
the array and in part to mismodelling of
troposphere.
Results The first three sessions have been
combined to produce a global reference frame
consisting of right ascension and declinations.
Because these sources are at extreme
extra-galactic distances (typically redshift z
1) we have assumed that proper motions and
parallaxes are negligible.
Fig 5a 1.5 sources fixed
Fig. 5b 4 sources fixed
Fig 5. Zonal Error ?RA vs. Declination. The
figures above show K-band minus S/X-band
differences for RA vs. Declination. A 3-D
rotation (lt 150 µas per component) was removed to
account for unmodelled differences in the
conventional orientation of each frame. In
Figure 5a. the minimal 1.5 sources have been
constrained in order to fix the orienation of the
frame. A residual zonal trend can be discerned.
Removal of this trend reduces the RA cos(dec)
wrms scatter from 294 to 248 µas. Note that the
least squares fits plotted above account for the
correlations between source coordinates (i.e.
full covariance). In figure 5b we have
constrained four widely spaced K-band sources to
their S/X positions. This effectively constrains
the large scale zonal errors and reduces the
scatter down to 168 µas. Some hints of zonal
errors remain.
Fig 2. Distribution of 108 K-band Sources
Observed. There are several things to note 1)
The VLBA observed only down to about -30 deg
declination. 2) Declination precision
systematically Decreases as one moves south. This
may be seen from the color coding which indicates
the declination formal precision according to the
above given legend. 3) The dashed yellow line
indicates the galactic plane. 4) The dashed light
blue line indicates the ecliptic plane. 5) Q-band
results are similar but the precisions are
roughly 1.7 times larger due to fewer
measurements.
VLBA Array of Ten 25m antennas
Mauna Kea
OVRO
Brewster
N. Liberty
Hancock
Fig 6a 1.5 sources fixed Fig.
6b 4 sources fixed
Fig 6. RA inter-source correlations Fixing 1.5
vs. 4 source positions. The figures above show RA
inter-source correlations vs. the arclength
separating a given pair of sources. The color
coding indicates declination band
(orange,red,green,blue,purple,black
lowest,highest dec). Figure 6a has the
minimally necessary one and half source
constraint needed to fix the 3-D orientation.
Figure 6b fixes 4 sources which effectively
constrains the large scale zonal behavior and
greatly reduces inter-source correlations. As
made clear by fig. 6a., when only 1.5 sources are
fixed the RA parameters are not well separated.
In fig. 6b. fixing 4 widely spaced sources allows
a clean separation of RA parameters at the cost
of imposing outside (S/X-band) information on the
K-band solution. This takes away the K-band
solutions ability to stand as a completely
independent frame.
Fig 3a. and 3b. RA and Dec differences K-band
vs. S/X ICRF. In order to evaluate the accuracy
of our reference frame we compared our results to
a recent S/X solution (DDOR_2002) which includes
about 3 million group delay measurements acquired
from 1978 to 2002. This S/X frame is an
unpublished extension of the ICRF extension 1
frame and is consistent with that frame at the
100 µas level. The differences seen here are 293
and 574 µas wrms in Ra cos(dec) and Dec,
respectively. The differences are about 2 times
larger than the formal precisions. Thus these
differences are an indication of systematic
errors. Fig 4a. and 4b. RA and Dec differences
K-band vs. Q-band. The histograms below show
K-band vs. Q-band mean differences of 159 and 141
µas in RA and Dec, respectively. K and Q-band
differences give evidence for systematic errors
from sources such as plasma and/or
instrumentation. The scatter is roughly consitent
with formal errors.
Further study The results shown represent a
work in progress. In particular, more data is on
the way and various aspects of the modelling have
yet to be fully explored. Data We have been
granted time for four more VLBA sessions in 2003.
Data for the first of these sessions was acquired
in May and will soon be added to the global
analysis. Three more sessions are planned for
later in the year. Troposphere Preliminary
studies show that 3 hour troposphere estimation
intervals improve the results compared to 1 hour
intervals. Troposphere modelling is of interest
because experience gained over the years from
analyzing S/X-band data suggests that zonal
trends vs. declination may be due in part caused
by troposphere mismodelling. GPS calibrations
About half the VLBA sites have nearby GPS
antennas. Troposphere estimates from this GPS
data could be applied to calibrate the VLBI data.
GPS is also used for estimating global ionosphere
models which could be used to calibrate our data
sets. Source Structure Images of the sources
have been made (see Fey et al paper this
session). Code has been written to correct the
astrometric data for extended structures.
Preliminary results from applying these
corrections show very slight improvements (2
psec) in both the K and Q-band group delay
residuals. Estimated source positions changed
typically by less than 100 µas.
Kitt Peak
Pie Town
Ft. Davis
Los Alamos
St. Croix
Fig. 1 - Photos credit NRAO/NSF/AUI
(http//www.aoc.nrao.edu/vlba/html/vlbahome/thesit
es.html)
Observations The results presented here represent
progress to date based on the first three
sessions which took place on 15 May 02, 25 August
02 and 26 Dec 02. The 108 sources observed were
selected for their expected compactness and
strong flux (typically gt 0.4 Jy). Each source was
observed with 3-5 snapshots each which used the
entire array for 2 min at K-band and 2 min at
Q-band. This strategy was chosen to allow good uv
coverage for simultaneous imaging while still
permitting sub-mas global astrometry from group
delays measured over a 400 MHz spanned bandwidth.
Recorded bandwidth was 128 Mbps. As may be seen
in fig. 1 above the array has a greater East-West
extent compared to its North-South extent. As a
result the array produces more precise right
ascensions than declinations by roughly factor of
1.5. The array is able to measure sources down to
about -30 deg declination.
Acknowledgement The research described in this
paper was performed by a team from the KQ VLBI
survey collaboration with members from the Jet
Propulsion Laboratory of the California Institute
of Technology, Goddard Space Flight Center, U.S.
Naval Observatory, (all under under a contract
with the National Aeronautics and Space
Administration), and at the National Radio
Astronomical Observatory, and Bordeaux
Observatory.