Aims. We investigate the flow field in the sunspot canopy using simultaneous Stokes vector spectropolarimetry of three sunspots (θ = 27• , 50• , 75• ) and their surroundings in visible (630.15 and 630.25 nm) and near infrared (1564.8 and 1565.2 nm) neutral iron lines. Methods. To calibrate the Doppler shifts, we compare an absolute velocity calibration using the telluric O 2 -line at 630.20 nm and a relative velocity calibration using the Doppler shift of Stokes V profiles in the umbra under the assumption that the umbra is at rest. Both methods yield the same result within the calibration uncertainties (∼150 m s −1 ). We study the radial dependence of Stokes V profiles in the directions of disk center and limb side. Results. Maps of Stokes V profile shifts, polarity, amplitude asymmetry, field strength and magnetic field azimuth provide strong evidence for the presence of a magnetic canopy and for the existence of a radial outflow in the canopy. Conclusions. Our findings indicate that the Evershed flow does not cease abruptly at the white-light spot boundary, but that at least a part of the penumbral Evershed flow continues into the magnetic canopy.Key words. Sun: photosphere -Sun: sunspots -Sun: magnetic fields
IntroductionFrom the pioneering work of Evershed (1909), we know that there is a flow field in the sunspot penumbra, which leads to a line shift and line asymmetry. For sunspots outside the disk center, it manifests in blue shifted spectral lines in the center side and red shifted spectral lines in the limb side. The penumbral fine structure has a close relation with the Evershed flow. It is generally accepted that the flow channels are more horizontal than the background field (Solanki 2003). Because spectropolarimetric data has lower spatial resolution than narrow band magnetograms, many questions regarding both the Evershed flow and the fine structure of the penumbra have remained (e.g. Solanki 2003; Bellot Rubio 2004). There are also controversial arguments about the continuation of the Evershed flow outside the sunspot.Brekke & Maltby (1963) studied horizontal variations of the Evershed flow. Investigating 2500 sunspot spectra, they reported that at the outer penumbral boundary "the velocity falls abruptly to zero". Wiehr et al. (1986) observed a sharp decrease of the Evershed effect and magnetic field at the visible boundary of sunspots. Using only Stokes I spectra, they argued that line-core shifts and line asymmetries are "strongly" limited to the continuum boundary of sunspots. Wiehr & Balthasar (1989), Schröter et al. (1989 and Title et al. (1993) confirmed the result of Brekke & Maltby (1963). Wiehr (1996) renewed his argument that the Stokes I profile asymmetries of Ni I 543.6 nm (g = 0.5) and Fe I 543.5 nm (g = 0) "disappear" within less than one arcsec from the penumbral border. Hirzberger & Kneer (2001) observed two sunspots at heliocentric angles of 31• and 20• ; they reported a sharp decrease of the Evershed flow (intensity profile asymmetry) at the penumbral boundary, using Sto...