2 Observations and data presentation

Our H₂O observations were carried out on the RT-22 radio telescope of the Pushchino Radio Astronomy Observatory (Russia). The details of the observational techniques and equipment were given by Lekht et al. (1995). The problem of radiowave absorption in the terrestrial atmosphere at the frequency of the H₂O line was discussed by Lekht et al. (1999) and Rudnitskij et al. (1999).

Figure 2: Time dependence of the integrated flux of U Ori in the H2O line

Figure 3: Same as in Fig. 1, but in the three-dimensional presentation

Figures 1a-r present the profiles of the H₂O line of U Ori. To get a more vivid idea of the overall maser activity, the same profiles are presented in the three-dimensional form in Fig. 2. To construct them, we used the method described in our previous work on the H₂O maser star RT Vir (Lekht et al. 1999). The sensitivity at the 3$\sigma$ level is on the average 10 Jy. The radial-velocity resolution of the spectra is 0.101 km s^-1. We used a 96-channel filter bank spectrum analyser which allowed us to measure simultaneously a $V_{{\rm LSR}}$ interval of 9.7 km s^-1. Since July 1997 the number of channels was increased to 128, yielding a one-time velocity coverage of 12.9 km s^-1. Most H₂O spectra of U Ori were taken in the range from -32 km s^-1to -42 or to -46 km s^-1. Unfortunately, for technical reasons and tight observational programme, it was not always possible to observe the more blueshifted part of the spectrum at -42 to -46 km s^-1; therefore, our data on the very interesting, transient maser features in this interval are not complete, see next section.

Our observations cover a time interval from March 6, 1980, to September 29, 1999 (JD 2444305-2451451). The profiles of March and June, 1980, as well as those of June, August and September, 1999, when the peak flux density of U Ori in the H₂O line was below 10 Jy, are not shown.

Figure 3 shows the variation of the integrated flux for the entire timespan of our observations. Vertical bars in the upper part of Fig. 3 denote the epochs of visual maxima of U Ori. These epochs are based on the visual light curves from the associations of variable star observers, French (AFOEV) and American (AAVSO). The maxima of the H₂O maser emission are obviously correlated with the visual light maxima and follow them with some phase delay $\Delta\varphi$. In Fig. 4 the values of $\Delta\varphi$ (expressed in fractions of the stellar light period P) are plotted versus time.

The most conspicuous event, as it is visible from Figs. 2 and 3, was the flare of 1980-1981, when the integrated flux density reached $\sim900$ Jykm s^-1. Then there were two weaker bursts, and again a stronger burst of up to 450 Jy km s^-1 at the end of 1983. By the beginning of 1988 the maser activity calmed down, with bursts' amplitudes of 30-100 Jy km s^-1, and remains such until now (late 1999).

The periods of both visual-light and H₂O-maser variations are subject to fluctuations. Figure 5 presents the lengths of consecutive "radio periods'' of U Ori, i.e., time intervals between pairs of consecutive radio maxima in the H₂O "radio light curve''.

Figure 4: Phase delay (in units of the stellar variability period) between the visual maximum and the H2O maser maximum immediately following it

Figure 5: Time dependence of the "radio period'' (time interval between two consecutive maxima of the H2O maser emission)

Figure 6: Weighted mean radial velocity for the main group of the H2O maser features in U Ori

We have also found systematic radial-velocity variations of the H₂O maser emission in U Ori. Figure 6 shows the weighted mean radial velocity for the main group of the H₂O maser features. A long-term drift of the H₂O velocity centroid is obvious. In our early observations (1980-1981) the radial velocity was between -36.8 and -36.9 km s^-1. By the end of 1981, there was an abrupt change to $\sim -37.4$ km s^-1. Then, until 1995 the mean velocity had been drifting to more positive values. The extreme value was -36.5 km s^-1, and then a turnover toward negative velocities followed. The curve in Fig. 6 may resemble a section of a sine wave.

Finally, in Fig. 7 the averaged and normalised "radio light curve'' of the H₂O maser U Ori is presented. This curve looks quite smooth, in contrast to the analogous H₂O curve for the semiregular variable W Hya (Rudnitskij et al. 1999). For W Hya the curve is sharp-peaked, with a steep ascending branch and a slower, nearly exponential decline.

Figure 7: Averaged and normalised "light curve'' of the H2O maser U Ori