TUG SDIMM (İngilizce)

Solar - Differential Image Motion Monitor at the TÜBİTAK National Observatory 

Solar Differential Image Motion Monitor at the TÜBİTAK<br />
National Observatory


Daytime seeing qualities of observatory sites in Turkey are not known. Therefore, the TÜBİTAK National Observatory (TUG) and İstanbul University, Faculty of Science, Department of Astronomy and Space Sciencesdecided to construct a Solar-Differential Image Motion Monitor (S-DIMM) for daytime seeing observations at the TUG. Although main goal of the project is to estimate the daytime seeing quality of the TUG, with this project, our observatory will gain a permanent differential image motion monitor for daytime and nighttime seeing observations. This project was supported by the Turkish Scientific and Technical Research Council (TÜBİTAK), Project Number: TBAG-2217 (102T110). 

TÜBİTAK National Observatory (longitude 30° 20´ E, latitude 36° 49´ N) is located at the south of Turkey at a distance of about 50 km south-west of Antalya. It lies on the Bakırlıtepe which is one of the important peaks along the Taurus Mountain System at 2550 m above sea level. Observatory has four telescopes with apertures of 1.5m, 1.0m, 0.6m and 0.45m. This site is being tested as the likely site of future large aperture telescopes. 



The Earth's atmosphere is turbulent and variations in the index of refraction cause the plane wavefront from distant objects to be distorted. This distortion introduces amplitude variations, positional shifts and also image degradation. This causes two astronomical effects: 

Scintillation: Rapid intensity fluctuations on an image, which typically vary over scales of cm. Scintillation effect is generally very small for larger apertures. 

Seeing: Positional changes and image quality changes. The effect of seeing depends on aperture size of the telescope. Seeing effect can be ignored for small aperture telescopes and limits the resolution. Even if one has a big aperture telescope, resolution can be as high as seeing. Although there are several superb image processing techniques and instrumental developments to reduce the seeing effect, looking for an optical observatory site with good seeing is still an important matter.

The time variation scales of these effects are a few miliseconds and up (Martin 1987). Typical frequency of atmospheric disturbance is around 100 Hz in optical wavelengths. Image degradation in the atmosphere is broadly understood theoretically in the framework of the Kolmogorov turbulence model (Tatarskii 1961, Roddier 1981). In this model, Fried's Parameter (r0, Fried 1965) is sufficient to describe all the seeing effects. The spatial resolution of ground-based telescopes is limited to that of an equivalent diffraction-limited telescope of diameter r0.

Fried's parameter can be measured from the image motion observations by a small telescope. However, direct measurements of the image motion in the focal plane of a telescope suffer from confusion of turbulence induced motion with motion induced by telescope (Bally et al. 1996). The motion induced by telescope consists of vibrations from wind shaking and tracking errors.

In order to remove the vibrational effects of the telescope, Stock and Keller (1960) introduced the DIMM (Differential Image Motion Monitor) concept. This concept is based on differential motions of two images of a star in the focal plane of a telescope. These images (sub-images) are produced by, simply, a Hartmann Mask which has two small apertures. DIMM method practically makes the image motion measurements insensitive to the vibrational effects of the telescope. The variance of the differential image motion can be used to calculate the Fried's parameter. Modern implementation of the DIMM is described by Sarazin and Roddier (1990) and Vernin & Munoz-Tunon (1995). The reader can find a great representation of the DIMM concept in Tokovinin (2002).



A DIMM (Figure-1) basically consists of a telescope, a Hartmann mask with two small apertures, a wedge prism mounted on one of the apertures of the mask, a CCD camera and computer. Telescopes with apertures of more than 20 cm can be used for constructing a DIMM. The wedge prism (or the optical wedge) deviates the light beam coming from the celectial object, a bright star, by a small constant angle. Thus, two "sharp" images of the same star is created in the focal plane of the telescope. CCD camera detects the image and transfers it into the computer. As a matter of fact, wedge prism can be eliminated from DIMM. In this case, slightly defocused images can be used (Bally et al, 1996). Consequtive images taken by the CCD camera represent the differential motion of the images. Changes in longitudinal (i.e., parallel to apertures vector of the mask) and transverse (i.e., perpendicular to apertures vector of the mask) separations of the stellar images present the differential image motion. These separations can be measured for consequtive images. This is DIMM data. And, standard deviation of the differential image motion in both direction can be used to calculate the Fried's parameter. For a single seeing estimation, number of consequtive images should be larger than 300 for the sake of statistical confidence.

Basic components of a DIMM 

Figure-1 Basic components of a DIMM.


There are two crucial parameters with the DIMM: 

1- Aperture Separation/Diameter of the Hartmann Mask: With a DIMM, we can measure the differential image motions originated from the turbulence cells as small as the aperture diameter and as large as the aperture separation from center to center. Thus, aperture diameter should be as small as possible and aperture separation as large as possible. This means that we need a large b=d/D rate, here b is the aperture separation and D the aperture diameter. According to Sarazin and Roddier (1990), this rate should be larger than 2.5. 

2- Exposure Time of the CCD Camera: Since the typical frequency of the turbulent motion is about 100Hz, exposure time of the CCD camera should be shorter than 10ms, if possible. The exposure time can be adjusted by taking into account the magnitude of star and diameter of apertures in the Hartmann mask, but by keeping the b=d/D rate larger.



Well, we observe a bright star with a DIMM in order to calculate the Fried's parameter. But, if one wants to estimate the daytime seeing quality, there is only one object to observe: THE SUN. A solar filter attached in front of the telescope entrance transforms a DIMM to the S-DIMM (Solar Differential Image Motion Monitor). Since the sun is not a point source, limb of the solar disk is observed in seeing studies (Beckers 2001, Liu and Beckers 2000). The S-DIMM creates two images of the same limb in the focal plane of the telescope, and the relative positions of the limbs is the S-DIMM signal. Unlike the stellar images, the solar differential motions can only be measured in one direction. The solar apparent top or bottom limb is observed by S-DIMM. This selection makes the S-DIMM less sensitive to tracking errors.



A few S-DIMMs were developed until the S-DIMM of the TUG described here. The first one was built at the Yunnan Observatory (Liu and Beckers 2000). Another one S-DIMM was built for site testing studies of the ATST (Advanced Technology Solar Telescope) (Beckers 2001). The Instrumental parameters of the S-DIMM at the TUG are as follows: 


Telescope MEADE LX200 SCT (Aperture=30.48 cm)
Telescope Focal Length 304.8 cm
Diameter of the Apertures of the Mask 40 mm
Aperture Separation of the Mask 250 mm
Direction of Separation N-S
The Solar Filter Thousand Oaks Optical Glass Filter, Type 2+
Transmission of the
Solar Filter
Video CCD Camera Astrovid Stellacam EX
Pixel Scale 
on the image of 768x576 pixels
Strehl Ratio of the
S-DIMM/DIMM system
0.68 +/- 0.01 (from stellar observations)
Exposure Time 0.001 sec
Cadence 25 Hz
Data Recording Device SONY GV-D800E Digital Video Recorder
Recording Medium Digital 8 Video Cassette

As can be seen from the table given above, we removed the computer from the observations. Instead, we use a digital video recorder (Figure-2) in order to make the S-DIMM more portable. Moreover, with this device, data capacity of the S-DIMM became independent from the computer's disk, it is limited only to the number of digital 8 video cassettes we have got! In addition, b=d/D rate mentioned above is 6.25 for the S-DIMM at the TUG. Optical wedge separates the limbs for about 80".

Basic Components of the S-DIMM at the TUG 

Figure-2 Basic components of the S-DIMM at TUG.


The S-DIMM is placed on the terrace of the T40 telescope building at TUG. Thus, height of the telescope from the ground is about 7 m. During the daytime seeing observations, we measure the longitudinal image motions of the solar limb. We see an image including two limbs on the screen of the digital recorder. Then, we record the video sequences in every 15 min. Duration of each video record is about 20 sec. at a speed 25 fps. By this duration, we get about 500 frames for a single seeing estimation. Effective wavelength of the filter+camera combination is about 550 nm. Video records of the defocused solar center are also taken for the flat field correction.



Since any reduction procedure has not been defined in the literature for S-DIMM data, we developed a S-DIMM data reduction algorithm using IDL software. Steps for the reduction procedure for each video record of 20 sec. are as follows:

  1. Grabbing limb and flat field frames in FITS format from digital video records.
  2. Creating a superflat image and correcting frames for the flat field.
  3. Applying Median Smooth and Sobel Filters to each frame for the reduction of noise and edge enhancement respectively.
  4. Selecting a limb profile direction on the first frame. This position is constant for all frames grabbed from the video record.
  5. Finding the limb positions along the profile line using Gaussian fits.
  6. Repeating the last step for 8 different profile directions apart one pixel from each other: 4 directions to the right and 4 directions to the left of the first selected direction. By this way, we create a digital slit of 12" in width.
  7. Averaging the limb positions found in last two steps for each limb.
  8. Difference between the average positions of the first and second limb is the separation of the limbs in the first frame.
  9. Repeating steps 5-8 for other frames.
  10. Converting the separations to arcsecond.
  11. Calculating the standard deviation of the limb separations: s

Then the program calculates the Fried's parameter by using the following formula (Tokovinin 2002):

Here, lambda is wavelength, r0 is Fried's parameter and D is the diameter of the pupils on the Hartmann Mask. K is a constant:

And, the seeing is calculated from the following formula:

This calculation gives us a single seeing value for one data set. We record 35 data sets in a typical clear day. This gives us 35 seeing values for a day. 35 data sets include about 15000 frames. Reduction of this data takes about 6 hours Pentium IV CPU processing time. 


TUG S-DIMM observations was started in June, 2003. The observations continued until the end of September 2004. Then TUG S-DIMM was slightly modified and moved to a new placenamed Solar Hill at the site where a modern solar telescope can be constructed in future. We also improved the data acquisition and data reduction method (Figure-3). We replaced the digital tape recorder with the firewire (IEEE 1394) frame capturing device which allows us realtime image processing. A new software has been developed for frame capturing, data reduction and archiving the results. By this way, the seeing value is calculated immediately after every set of image capturing session. Observations at the new place were made between May and October 2005.

Improvement on the TUG S-DIMM data<br />

Figure-3 Improvement on the TUG S-DIMM data acquisition system.



The statistical results of the observations done from June 2003 to September 2004 are given in the following tables and the graphs. Since our observations cover a short period, the results can not present long-term characteristic of the site. These statistical results can be considered as a general representation of the day-time seeing conditions at the TUG site. More information can be found in a paper appeared in the journal Astronomy and Astrophysics (2004, Vol.422, p.1129) 

This project is one of the rare applications of the DIMM principle to the day-time seeing studies. In addition, we presented the first data reduction algorithm for this kind of studies. 

The experiences gained during this project was very useful in construction of a temporary DIMM in the TUG site. TUG-DIMM project has already been started and some pictures from this project can be seen at: 



Monthly distribution of the observations. 
Monthly distribution of the observations on the terrace of the T40 telescope building (Jun. 2003 - Sept. 2004).


Monthly median Fried parameters (r0) on the terrace of the T40 telescope building. q1 and q3denote first and third quartiles, respectively.

Month Day Duration
March 2 21 7.64 9.63 11.72
April 1 9 4.65 5.55 6.19
May 5 31 3.18 3.62 4.25
June 7 55 3.81 4.56 5.35
July 2 10 4.05 5.06 6.53
August 4 30 2.59 3.02 4.84
September 8 73 3.34 3.88 5.98
October 2 19 3.53 4.60 6.82
November 1 7 4.93 5.78 6.65
December 2 17 7.55 8.65 9.78


Seasonal median Fried parameters (r0) on the terrace of the T40 telescope building. q1 and q3are as in previous table.

Season Day Duration
Winter 2 17 7.55 8.65 9.78
Spring 8 61 3.97 5.55 7.41
Summer 13 95 3.18 3.34 5.74
Autumn 11 99 3.52 5.06 6.51
Yearly 34 272 3.69 5.19 7.01


Monthly median Fried parameters (r0) on the Solar Hill. q1 and q3 denote first and third quartiles, respectively.

Month Day Duration
May 1 4 3.76 5.16 6.20
June 4 36 2.73 3.31 4.06
July 1 7 3.68 5.93 6.51
August 5 39 2.95 3.58 5.56
September 3 10 3.33 4.58 7.01
October 3 18 2.09 2.67 3.51


Monthly distribution of the observations. 
Monthly distribution of the observations on the Solar Hill (May. 2005 - Oct. 2005).



Jun. 2003 - Sept. 2004 : T. Ozisik (TUG) and T. Ak (İstanbul University) 

May 2005 - October 2005 : M. Basal, T. Guver and F. Bostanci (İstanbul University) 



We wish to thank Dr. Alexander Yascovich from Russian Academy fo Sciences (IKI) for his efforts in obtaining an optical wedge. We would like also to thank to the staff of the TÜBİTAK National Observatory and the staff of the solar physics section at the İstanbul University Observatory for their comments and technical support. 



- Bally, J. et al., 1996, Publ. Astron. Soc. Aust., 13, 22.
- Beckers, J.M, 2001, Experimental Astronomy, 12, 1.
- Fried, D., 1965, J. Opt. Soc. Am., 55, 1427.
- Liu, Z., Beckers, J.M., 2000, Solar Physics, 198, 197.
- Martin, H.M., 1987, PASP, 99, 1360.
- Ozisik, T. and Ak, T., 2004, Astron. Astrophys.,422, 1129.
- Roddier, F., 1981, in Progress in Optics, ed. E. Wolf (Amsterdam: North-Holland), 281.
- Sarazin, M., Roddier, F., 1990, A&A, 227, 294.
- Stock, J., Keller, G., 1960, in Stars and Stellar Systems, Vol.1, Telescopes,
   eds. G.P. Kuiper & B.M. Middlehurst (Chicago:Univ. of Chicago Press), 138.
- Tatarskii, V.I., 1961, Wave Propagation in a Turbulent Medium (New York: Dover)
- Tokovinin, A., 2002, PASP, 114, 1156.
- Vernin, J., Munoz-Tunon, C., 1995, PASP, 107, 265.


TÜBİTAK National Observatory
Tansel AK
İstanbul University 

Last update: Dec 08, 2009


-A +A