2 Observations

MERLIN was used to determine the positions of the H₂O masers in the CSE of U Her with respect to 3 extragalactic reference sources. From the observations, which were performed on May 20 2001, we were able to determine accurate astrometric positions. The observations required a total of 8 hours on the target source and the calibrator and reference sources. As a result the beam size was $\approx$ $30 \times 25$ mas.

One of the main problems of using the phase referencing method to determine accurate positions at 22 GHz is the lack of bright reference sources at that frequency. For this project 3 reference sources were selected from the VLBI calibrator catalog (Beasley et al. 2002). These were J1628+214 (now J1630+231, $\approx$80 mJy, at $2.8^\circ$ from U Her), J1635+1831 ($\approx$50 mJy, at $2.4^\circ$) and J1619+2247 ($\approx$200 mJy, at $4.2^\circ$). They were selected to have a flat spectrum, making it possible to detect them at 22 GHz with a 2 min integration time. Our phase reference observation cycle consisted of 2 min per reference source and 4 min on the target source. The H₂O masers were observed in a 4 MHz band with 128 spectral channels, providing a velocity resolution of 0.42 km s^-1. The continuum reference sources were observed in a 16 MHz band with 16 channel, giving a 13.5 km s^-1 resolution. In the 2 min integration time on the reference source we typically reached a  ${SNR}\approx 5$.

Reference source J1628+214 was also used in vL00 to determine the positions of the 1667 MHz OH masers around U Her. We can thus determine the positions of the H₂O masers with respect to the OH masers. The positional accuracy of the reference sources with respect to the radio reference frame is $\approx$3 mas. However, as the positions were determined at 2.4 and 8.4 GHz the position at the different frequencies can be different at the sub-milliarcsecond level, as shown in e.g. Fey et al. (1997).

After initial flux calibration with the MERLIN software, the data were processed in AIPS. Because the reference sources at 22 GHz are weak, we performed both a normal and a reverse phase referencing scheme. In the reverse scheme, we were able to determine accurate positions of both J1619+2247 and J1628+214 with respect to the brightest H₂O maser feature of U Her. As a consistency check, we managed to get a good phase connection from J1628+214, the closest reference source, to the U Her H₂O masers with the normal phase referencing scheme. J1635+1831 was too faint for us to obtain a good phase connection.

The accuracy of the phase referencing model depends on the MERLIN correlator model. The two largest uncertainties in the model are the telescope positions (accurate to $\sim$5 cm) and tropospheric effects. Under normal circumstances these combine to produce an error in absolute position measurements of up to $\sim$10 mas for a target-reference source separation of $\sim$$3^\circ$ (e.g. Kovalevsky et al. 1997).