logo kso logo uni graz


The old Kanzelhöhe H-α Imaging System (before 2005)

 

Instrumentation

The Kanzelhöhe Patrol Instrument housed several telescopes on a common mounting: A white light telescope with a projection system for the sunspot drawings, a coronagraph, the photoheliograph (Pettauer, 1990), a Magneto-Optical Filter operated in the Na-D lines (Cacciani et al., 1998) and finally the Ha Instrument (The layout is shown in figure 1). Attaching the new device required only minor changes to the system. A beam splitter allows to record the images simultaneously analogous on photographic film and digital using a PULNIX TM-1000 progressive scan camera with a 8-bit digital read-out. A parallel fibre optics links the camera to a DIPIX XPG-1000 frame grabber in a 486/66 PC running Windows 95.
 
 

Figure 1: The layout of the Kanzelhöhe Digital Full Disk Hα Patrol System. An interference filter (10 nm FWHM) preselects the wavelength and prevents the ZEISS Lyot filter (0.07 nm FWHM) from thermal stress. The beam splitter enables the simultaneous operation of the old analogous and the new digital system.

 
 

Image Aquisition and Processing

The software driving the frame grabber allows to save single frames or the "best" image of a preselected number of consecutive aquired frames (frame selection mode) to a local or remote HDD. These two modes can be repeated at selectable time intervals (for patrol operation). Servicing the frame grabber keeps the data aquisition PC busy, in particular in high cadence patrol mode. Therefore the further image processing is performed on a second PC using IDL. The raw images are copied to CDs for the basic archive.
 
 

Frame Selection

To select the "best" one of a number of consecutive aquired frames (as stated above) we are looking for the frame with the maximum variance in a subsection of the frame near disk center. The variance is directly correlated to the average contrast of the image and can be calculated using a fast internal routine of the frame grabber's on-board DSP.
 
 

Flat Fielding

The read-out noise of the system is very low. Only two pixels are different from zero in dark images, although one pixel column is corrupt and shows an arbitray shift in intensity. But investigation of correlation to the neighbouring columns and autocorrelation of the column showed the column to be consistent itself. So one can determine the shift for each image and correct it by subtraction. Further a large scale asymmetry can be observed. It can be removed by using a flat fielding method developed by K. Burlov-Vasiljev at KIS Freiburg, Germany. This method, which uses a single frame, splits the solar disk into concentric rings and fits polynomials into these. As the disk is supposed to be radial symmetric in large scales the polynomials give the image distortion similar to a Fourier decomposition. Combining the polynomials of the rings this method gives also the center-to-limb variation of the sun.
 
 

Image Centering and Orientation

Image center coordinates (x0, y0) are determined by looking for the middle between the maximum gradient positions along several pixel rows (for x0) and columns (for y0) and averaging the row and column data. The image is shifted near to center by an integer amount of pixels to avoid data distortions due to pixel averaging. Because of the equatorial mounting of the telescope we don't have a systematic image rotation during the day. Therefore the correction angle δP which denotes the inclination of the E-W-direction with respect to the frame axes is constant beside some small variations (less than ±0.2deg) due to small shortcomings of the telescope and the mounting. It can be determined and checked from time to time by recording a set of images with turned-off tracking system and computing the track of the disk centers. Heliographical positions can be given better than ±0.5deg.
 
 

Image Cosmetics

For quick-look purposes a composite image is created. It consists of a processed, CLV removed, structure and contrast enhanced disk image superimposed with a coronagram to enhance the visibility of the prominences. The coronagram is calculated by subtracting the disk from a raw frame and intensifying the residual image. From the time of observation we calculate the physical ephemeris (P, B0, L0) and rotate the disk to have Solar North up as generally used.

Figure 2: A processed composite image from the quick-look series as published online according to the standards for SOHO Ground Based Synoptic Observations. Asymmetries and CLV are removed, the region outside the disk is intensified to make prominences better visible. The disk is unsharp masked to enhance the structures. Annotations are only displayed in the quick-look series saved in GIF or JPEG format, the full sized data contain the relevant information in their FITS header.

 
 

Data Archive

The basic archive consists of patrol observations with a cadence of 1 image per minute if weather permits. This boosts the capability of the data set (Due to economical reasons the film recordings were limited to 1 image per 4 minutes). Data are saved in a raw image format on CDs. A software package is provided for the basic data handling. Low cadence data - extracted from the basic data - are processed to meet the standards of the SOHO Ground Based Synoptic Observations. The routine service started in Feb 2000.
 
 

References

Cacciani, A., Moretti, P. F., Messerotti, M., Hanslmeier, A., Otruba, W., & Pettauer, T. in Motions in the Solar Atmosphere, Hanslmeier, A., & Messerotti M. (eds.), Kluwer Academic Publishers, 271 , 1999

Otruba, W. in Third Advances in Solar Physics Euroconference: Magnetic Fiels and Oscillations, Schmieder, B., Hofmann, A. & Staude, J (eds.), ASP Conf. Series, 184, 314, 1999

Pettauer, T. Publ. Debrecen Heliophys. Obs. 7, 6, 1990

 
 

Image Types

The images (1008x1016 pixels, 1 byte) are saved in different formats. The filenames match the filenaming convention for SOHO synoptic data:

  • FITS: The header is styled to match the "proposed Keywords for SOHO". The images are zipped using gzip. (abt. 330kB)
  • JPEG: The images use 8 bit color depth (greyscale). (abt. 100kB)
Image Type Description Filetype
1 Unprocessed raw image. Known corrupt pixels of the CCD are corrected by interpolation. A set of the whole 7th-bit plane, which occurs occasionally due to some unknown bug in the datalink is corrected as well by checking the value of the pixel [0,0]. Further the disk image is shifted near to image center. (Only by an integer amount of pixels to avoid image degradation due to interpolation.) FITS
2 Center-to-Limb Variation correction table. The table gives an intensity profile of the quiet sun derived from the flat fielding process. FITS ASCII Table
3 Low contrast image. Image Type 1 is processed by removing large scale asymmetries using a flat fielding method. FITS
4 High contrast image. The image type 3 is normalized by dividing the image through a quiet sun image which is derived from the CLV profile. This makes the structures better visible. FITS
5 Low contrast image. Image type 3 is rotated to have Solar North up. The histogram is expanded to enhance the image contrast. Unsharp masking is applied to make faint structures better visible. JPEG
6 High contrast composite image. A coronagram is created by fading out the disk and intensifying the regions outside. It is superimposed with image type 4. The image is rotated to have Solar North up. The histogram is expanded to enhance the image contrast. Unsharp masking is applied to make faint structures better visible. JPEG
7 Unprocessed raw image. Format available only for observations before Feb 2000. FITS
8 Image type 7 centered and rotated to Solar North is up according to the SOHO synoptic image standards. No additional image processing is applied. Format available only for observations before Feb 2000. FITS
9 Like image type 8, but the images are half-sized, contrast- and limb-enhanced to make the prominences better visible. Format available only for observations before Feb 2000. JPEG
0 Nearly live image. The always latest observation is displayed completely unprocessed as observed for a quick updating. A text string in the image denotes the date and time of observation. Mainly intended for local purposes. JPEG
 
 

Kanzelhöhe Ha Archive Structures

FolderExplanationImage Format
/halpha /FITS /archive /YYYY ...unprocessed images 1
/halpha /FITS /archive /YYYY ...unprocessed images (until Jan 2000) 7
/clv /YYYY ...CLV correction table 2
/high /YYYY ...processed high contrast images 4
/low /YYYY ...processed low contrast images 3
/low /YYYY ...processed low contrast images (until Jan 2000) 8
/JPEG /high /YYYY ...processed high contrast images 6
/low /YYYY ...processed low contrast images 5
/low /YYYY ...processed low contrast images until jan 2000 9
/recent /YYYY icons of low and high contrast 5, 6