Astron. Astrophys. 348, 621-626 (1999)

2. Observations and data analysis

The observations, data reduction, basic ideas on the identification and tracking algorithms, determination of image sharpness, and discussion of stray light effects have been thoroughly described in Paper I. Here we only give a brief summary of facts and methods that are common to both Paper I and this work, but present more detailed information about specific changes and problems related to the penumbral data analysis.

The sunspot NOAA 7519 was observed at heliographic position N05, E15 on 5 June 1993 with the Swedish Vacuum Solar Telescope (SVST, aperture 50 cm) at La Palma, Canary Islands (Simon et al. 1994). This slowly evolving spot reached maximum area on the date of observation. A Kodak Megaplus Model 1.4 CCD camera, in connection with a fast real-time frame selection system (working at approx 3.5 frames/s), was used to sample white light solar images at Å. The image scale was 0:00125 per pixel. Due to image rotation during the 11-hour time series, the sunspot was visible only for 4 hours 26 minutes, from 09:54 to 14:20 UT, during which we obtained 760 frames. After correcting for dark current, flat field, rotation, transparency, and exposure variations we reduced the field of view to pixels, or , centered on the spot. The frames were registered, corrected for instrumental profile, and destretched, to minimize seeing distortions. From these we selected 360 frames with rms granulation contrast higher than 7%, covering almost regularly the whole time period. The selected frames were then interpolated in time to obtain a time series with a constant lag of 44.5 s. Residual seeing-induced jitter which impeded the tracking of PGs in time was removed by (subsonic) filtering. The cutoff phase velocity of the filter was set at 3 km s^-1. For further analysis we selected three subfields of the sunspot that showed regular penumbrae.

To isolate PGs in the subfields we applied a geometrical mask to exclude light bridges, eliminated umbral dots by setting the minimum intenstity of bright features to 0.8 , and removed photospheric features located outside the P/Ph boundary. After this preparatory phase we tested several segmentation procedures to separate PGs from the rest of the penumbra. This task was more complicated than in the case of umbral dots due to the complex and fast-changing penumbral intensity structure.

We adopted a procedure similar to that described in Paper I: For each frame a differential image was computed by subtracting a smoothed ( boxcar) image from the original one. From this differential image we computed a binary mask, setting pixel values higher than a certain threshold to 1 and the rest to 0. The threshold was defined as a linear function of the image sharpness, computed for each frame with the Roberts gradient operator (see Paper I), such that its values were 0.06, 0.04, and 0.03 in the best, average, and worst frames, respectively. The original image was then multiplied by this mask, producing a segmented image.

From the images we produced a time series of variable-threshold segmented frames. A feature tracking algorithm, described in Paper I, was applied to this series to follow each PG in time. To be called a PG, each feature had to:

• (a) be larger than 1 pixel on each frame;

• (b) overlap in at least 1 pixel on subsequent frames;

• (c) have a maximum interruption of 3 frames;

• (d) have a lifetime of at least 3 minutes;

• (e) have a mean area over its lifetime of at least 9 pixels.

This procedure yielded a set of 1027 PGs. For each of these we tracked the locus and value of maximum intensity at each time step. Lifetimes were derived from the number of frames in which the objects were present. Time-averaged proper motion velocities were calculated using linear least-squares fits to the positions. Since the accuracy of velocity determinations increases with the number of frames in which the PG is present, we chose for analysis 649 PGs with lifetimes longer than 12 minutes. In this new set, the standard deviation is about 0.05 km s^-1.

In the last step of the data analysis we checked visually the trajectories of these 649 PGs to eliminate spurious objects and mistakes in feature tracking. We discarded all features that showed:

1. discontinuities in position or strong bends/breaks in the trajectory (caused by spurious coincidences of independent objects);

2. repeated back-and-forth motions (wrong tracking of poorly defined large features);

3. stationary objects at the P/U or P/Ph boundary (border effects).

After this visual consistency check we obtained a final sample of 469 PGs whose proper motions, intensities, and lifetimes were investigated.

© European Southern Observatory (ESO) 1999

Online publication: July 26, 1999
helpdesk.link@springer.de