2 Solar radius variations observed at CERGA and Santiago

The most extended databases of homogeneous solar radius obtained from these programs and comprising continuous results over a lapse equivalent to or greater than a solar cycle, are those of Laclare at CERGA (Laclare et al. 1996) and Noël at Santiago (Noël 2001). However, there is quite a strong disagreement between the results of both series. While the CERGA measurements show a long term variation of the radius of $0\hbox{$.\!\!^{\prime\prime}$ }2$ amplitude in opposing phase with sunspot number (Laclare et al. 1999; Emilio et al. 2000; Pap et al. 2001), the measurements of Santiago show a significant radius variation of about $0\hbox{$.\!\!^{\prime\prime}$ }5$ amplitude, but in phase with solar magnetic activity (Noël 2001, 2002; Pap et al. 2001). This disagreement is illustrated in Fig. 1 which shows for the period 1990-1998, the smoothing curves of the variation in time of the observed solar radius and of sunspots numbers.

Figure 1: Variations in time of the solar radius according to the visual astrolabes of CERGA and Santiago, and variation of sunspot number. The curves for Santiago and sunspots are Vondrak fits to monthly means of solar radius, and to monthly sunspot numbers given by SIDC (Noël 2002; Vondrak 1963). The curve for CERGA was adapted from a harmonic fit to individual radius values given in the paper of Pap et al. (2001) (see also Emilio et al. 2000)

Bearing in mind that the solar astrolabes of CERGA and Santiago are essentially similar instruments and both have a similar wide bandpass around 538 nm, it is fair to recognize that this disagreement has been quite disconcerting. However, our view of its probable origin will be discussed in a report in preparation about the definitive results of our program of solar observations made over twelve years by a single observer with the Santiago astrolabe.

Due to its extension, internal consistency and pioneering character, the visual results of Laclare have been quite influential in the solar astrolabe community. It has been claimed that the first visual results that have been obtained at the other astrolabe stations were always different to those obtained simultanously by Laclare at CERGA. Nevertheless, during the course of time there is an evolution of the results that makes them finally converge to the results of CERGA (Chollet 1995; Sinceac et al. 1998). However, such evolution has never been observed in the results of Santiago. On the other hand, according to Chollet & Sinceac (1999),

"... the apparent solar radius variation is identical for all measures done with astrolabes (except Santiago), and in anticorrelation with the magnetic cycle of the Sun.''

We do not share this assertion since as far as we know, before 1999 the only long term data set with results produced simultaneously with those of Santiago was that of CERGA. The other astrolabe stations that could be compared with the results of Santiago did not have a sufficient extension in time to reveal a long term trend in their radius measurements. However, at the present time the results with the CCD astrolabe of the Tubitak National Observatory, Antalya, Turkey, available at the home page of this observatory, for the period 1999.5 to 2001.5 give a radius variation quite consistent with the results of Santiago (see Fig. 4). Concerning the CCD astrolabe of Rio de Janeiro, its results for 1997-2000 are now available at CDS (Jilinski et al. 1999; Puliaev et al. 2000; Penna et al. 2002). It is true that for this period the drift of the radius mesurements of this astrolabe is similar to the results of Laclare. However, it will be shown in this paper that there is a lack of homogeneity in the database from Rio de Janeiro. If a procedure of homogenization is applied, the solar radius observed from Rio de Janeiro shows a variation in time almost identical to the visual results of Santiago.

Figure 2: Daily mean values of solar radius measurements made during 1997-2000 with the CCD astrolabe of Rio de Janeiro. The smoothing curve is a least squares second order fit with a standard deviation of $\pm0\hbox{$.\!\!^{\prime\prime}$ }31$. Apparent discontinuities in the radius values can be seen around 1998.0 and 1999.0.