Site testing for an optical observatory in Turkey

Z. Aslan¹*, C. Aydın¹, Z. Tunca², O. Demircan³**, E. Derman¹, O. Gölbaşı⁴, and A. Marşoğlu⁵

¹ Ankara University, Faculty of Sciences, Astronomy Department, 06100 Beţevler, Ankara, Turkey
² Ege University, Faculty of Sciences, Astronomy Department, İzmir. Turkey
³ Middle East Technical University, Physics Department. Ankara, Turkey
⁴ Boğaziçi University, Kandilli Rasathanesi, Çengelköy Istanbul, Turkey
⁵ İstanbul University, Faculty of Sciences, Astronomy Department, Beyazıt, Istanbul, Turkey
* Now at İnönü University, Faculty of Sciences and Arts, Malatya, Turkey
** Now at Ankara University, Faculty of Sciences, Astronomy Department, Ankara, Turkey

Summary. Site testing observations made between 1982 and 1986 are reported. Long-term meteorological records have indicated, and on-site observations have confirmed, that the southwest and southeast of Turkey contain good potential observatory sites. An inspection of the visual and infrared images received from the satellites Meteosat and NOAA-7 have indicated that the Antalya Bay region and the southeast of Turkey have statistically less cloud cover. The resolution of the images is too low to locate a mountain top above an inversion layer. The average height and frequency of occurrence of an inversion layer have been deduced from radiosonde data. Four candidate sites have been selected by reconnaissance in the potential regions. Atmospheric seeing observations by the polar star trail method and meteorological observations have been carried out simultaneously on two sites at a time. The comparison of the mountains has shown Bakırlıtepe to stand out as a good site. It has further been shown that Bakırlıtepe compares very favorably with the Roque de los Muchachos Observatory on La Palma, which is one of the world's best observatories.

Key words: observatory site - seeing - site testing

       In May 1978 a committee was set up by the Turkish Universities under the auspices of TUBITAK, Turkish Scientific and Technical Research Council, to make plans for a site survey. A commission was formed to study the long term (up to 50 years) meteorological records, to delineate possible regions with subsequent visits to the regions by several astronomers. In the meantime help was sought from the IAU. Letters were exchanged with Prof. M. F. Walker of Lick Observatory, and with Prof. F.G. Smith, then director of the Royal Greenwich Observatory. Mr. B. McInnes of Edinburgh University was invited to Turkey in October 1981. He, together with Turkish astronomers, visited several sites in the southwest of Turkey selected on the basis of the existing meteorological data.

       Long-term meteorological observations published by the State Meteorological Office have been studied for the number of sunshine hours, cloud cover, temperature distribution, relative humidity, precipitation, and wind speed. The earth-quake maps were also examined. The south of Turkey (Fig. 1) below about 38' latitude, with sunshine hours per year greater than 2700, was found to be the best region for potential optical observatory sites.

       Visual and infrared images received daily from Meteosat and NOAA-7 by the State Meteorological Office and covering an interval of an entire year between April 1982 and April 1983 were examined. Inspection of the images showed that statistically the Antalya Bay region and the Southeast of Turkey (see Fig. 1) had the least cloud cover, confirming the meteorological records. No quantitative results could be obtained because of the poor resolution of the images; in particular, it was not possible to locate a mountain top above an inversion layer.

       The radiosonde data obtained by the Meteorological Office at three stations in or near the regions of interest, namely İzmir, Isparta, and Diyarbakır (see Fig.1), were examined to get information concerning the height and frequency of occurrence of the inversion layer. Initially this was done for the year 1981, but it was later extended to cover the five-year interval from 1980 to 1984. No seasonal change was noticed in the average height of the night inversion layer. Table 1 and Fig. 2 summarize the results. It may be noted that İzmir is a coastal station, Diyarbakır an inland one, while Isparta is at an intermediate distance from the coast and has an intermediate climate. The sharp change of slope in Fig. 2 for Isparta and Diyarbakır is due to the higher frequency of

Fig. 1. Map of Turkey showing the positions of sites tested (triangles), radio sonde stations (filled circles), and lines of sunshine hours per year

Fig. 2. Percentage of nights with inversion layer formed below a given height

surface inversions. It can be seen from Fig. 2 that curves level off after 4000 m, which means that the difference between coastal and inland sites virtually disappears about 4000 m above sea level. This is consistent with the results of Barletti et al. (1977), who found that the levels of thermal turbulence above 4000 m altitude over different sites and latitudes were about the same.

    Figure 2 and Table 1 show that most of the inversions occurring over inland sites are due to surface cooling. most of which occurs below I 500 m, and that the formation of "high" inversion layers by subsidence of the airmass with the bottom of the temperature inversion well above the ground is not frequent. Therefore, as far as the effect of the inversion layer is concerned, not much will be gained by going to altitudes over 2000 m. Furthermore, inland mountains over 3000 m have rather severe winter conditions, such as heavy snow fall and strong wind. On the other hand, coastal mountains with altitudes over 2500 m will be above the inversion layer most of the time.

A total of 17 candidate sites, 13 in the southwest and 4 in the southeast, were reduced to four sites by surface reconnaissance. In doing so, conditions such as the shape of the mountain, availability of water and electricity, nearness to a university and an airport etc.. were considered, and sites accessible by existing roads or by a land vehicle at least up to the base of the mountain were selected, because no funds were available for road building.

    The sites selected for on-site observations are given in Table 2 (see also Fig. 1). Kurdu has a wide summit area covered with fir trees. There is a sentry building for forest fires. The other sites are bare and rocky. Bozdağ is a peak on a short ridge running approximately E.-W. Nemrutdağı is on a similar ridge with loose

Table 1. Average heights (above see level) and frequency of occurrence of inversion layers (IL) from 5-years radiosonde data (height refers to top of IL)

^a Temperature minimum coincident with the surface, i.e. height refers to top of surface convection zone

^b Temperature minimum occuring on the average about 200 to 400 m below the top of the layer

^a Single season only

^b KR, BZ, BK, and NM stand for Kurdu, Bozdağ, Bakırlıtepe, and Nemrutdağı, respectively

stones and soil. Bakırlıtepe is one of the two peaks about 1 km apart in N-S direction. The station was set up on the slightly lower peak at 2450 m.

Meteorological observations and astronomical seeing measurements were carried out simultaneously on two sites at a time by a team of two observers who remained on the site continuously for two weeks on Kurdu and one week on the others. On Bakırlıtepe and Bozdağ, an aluminium cabin was set up, whose elements were carried to the summit manually piece by piece. A tent was used on Nemrutdağı, while on Kurdu the observers stayed in the sentry building for forest fires.

    Time intervals of the observations, which are limited to "summer" months because of the lack of protection for the observers during winter conditions, are given in Table 2.

    On each site measurements of percentage cloud cover, relative humidity, temperature, wind speed and direction, and astronomical seeing were made, whenever possible, hourly during the night. However, no quantitative extinction or sky brightness measurements could be made. The thermometers, the thermograph, and the hygrograph were housed in an standard box at a standard height above the ground. The wind gauge was mounted 4 m high, and the observed wind speed was reduced to a height a height of 10 meters. The number of nights assessed on each site are given in Table 3.

Night time meteorological observations are summarized in Figs. 3-5 and Table 4. Their details and daytime observations have been given elsewhere (Aslan et al., 1988). Figure 3 gives monthly numbers of photometric, spectroscopic and unusable hours. The data have been normalized to a 30-night month and to full coverage in each case. A dark hour, i.e. an hour within astronomical darkness, was assessed as photometric if the following conditions were fulfilled: obscuration above 10 elevation less than 20%; relative humidity less than 90%; average wind speed

Fig. 3. Monthly numbers of photometric ■, spectroscopic ■. and unusable ■ hours

Fig. 4. Month by month average wind speed (dark hours)

Fig. 5. Month by month average relative humidity (dark hours)

(at 10 m above ground) less than 20 ms^-1; astronomical seeing not more than 5". A spectroscopic hour was one in which the humidity and wind speed conditions were as above, excepting that the obscuration limit was relaxed to 50 % and no limit was imposed on the seeing. All other dark hours were classified as unusable. It may be noted that these definitions of photometric and spectroscopic hours are not identical with those of McInnes (1981), as we have no extinction measurements.

    The monthly mean wind speed during dark hours is plotted in Fig. 4. Figure 5 is a similar plot for the relative humidity. For two sites, Kurdu and Bakırlıtepe, relative humidity data are available for the whole year: At Kurdu humidity during winter months was recorded at the site itself, while a small station was set up at Bakırlıtepe about 600 m below the actual site. In Fig. 5, S refers to observations obtained at this station, which should be considered as upper limits to the humidity at the summit. Note that the inland mountain Nemrutdağı is the least humid, as expected.

    The mean temperature during dark hours and the mean amplitude of the daily temperature variation are tabulated in Table 4. Figure 6 shows the percentage of nights with a temperature drop ∆T smaller than a given value, where ∆T refers to cooling during dark hours.

Fig.6. Percentage of nights with a temperature drop smaller than ∆T (dark hours, May to October)

Astronomical seeing measurements were made by the polar star trail method (Harlan and Walker, 1965; Walker, 1984). In this method. the pole star is photographed using a telescope giving a scale of 15".5 mm^-1. During the exposure, Polaris is allowed to drift across the film by diurnal motion. For proper trail density, Kodak Panatomic-X film is used. Harlan and Walker obtained a good correlation between the Polaris trails obtained in this way and the image diameter observed simultaneously on the slit of the 120-inch reflector's coude spectrograph at the Lick Observatory. According to Walker, the image diameter measured on the slit of the 120-inch telescope corresponds approximately to that at FWHM. There are many sites around the world now with seeing determined using the same method (Walker, 198d). making comparison of sites possible down to 0".7.

    The two Polaris Trail Telescopes used in this work were borrowed from the Royal Greenwich Observatory. They were set up on rigid mountings, with the objective approximately 2.5 m above ground level. Blurring and movement of the Polaris trail on films exposed for 15 min were assessed against standard trails, which were calibrated and supplied to us by Prof. M. F. Walker at Lick Observatory. Specimen trails were sent to Prof. Walker to check on the match to the standards and on our assessments.

    Figure 7 shows the frequency of nights with average seeing smaller than a given value, while Fig. 8 displays the cumulative frequency distribution of the individual seeing measurements. As already mentioned, the Polaris trails were obtained every hour whenever weather conditions permitted. Thus Fig. 8 gives the equivalent of the percentage of hours with seeing smaller than a given value.

    When different seasons within each site were compared separately, it was found that the frequency distributions of the temperature drop ∆T and of the seeing were strongly correlated. One sees in Fig. 7 or 8, however, that the seeing on Bozdağ is not as good as would be expected from the ∆T distribution in Fig. 6. However, from the wind data obtained at the summit and at about 50 m below the summit, the average seeing was found to depend on the wind direction, the air flow from the east being disturbed the most (Aslan et al., 1988).

    The best seeing, e.g. seeing better than 0".9, occurred most frequently on Nemrutdağı, but its average seeing is the worst, as clearly seen in Figs.7 and 8. This persists even when the comparison is made month by month. A strong correlation was found between seeing and wind speed at Nemrutdağı. The telescope mechanical stability was not measured, but the mounting used is very stable and should permit observations without any detectable effects of telescope vibration in wind speeds up to at least 55 km h^-1 (Harlan and Walker, 1965). In fact no correlation between seeing and wind speeds up to 80 km h- t was found on other sites. Also, it was observed on Nemrutdağı that during strong wind, which was dominantly from the north, the dust content of the air increased. Strong wind brought dust which in turn deteriorated the atmospheric seeing and increased the extinction through scattering, resulting in a fainter Polaris trail, as well (Aslan et al., 1988).

The four sites have been assigned ordinal numbers from 1 to 4 according to various parameters discussed above and are shown in Table 5, the smaller number referring to the better condition. It can be seen from the last column that Bakırlıtepe is dominantly the best site. It may perhaps be remarked that not all the parameters in Table 5 are independent, no reasonable weighting, however, will change the result.

    There is one important aspect of this comparison that is not readily apparent in Table 5. The observations on Nemrutdağı, which is the only inland mountain, were made for one season only from 7 July to 7 September 1985, while for other sites the observations used in the comparison were obtained from May to October in two seasons (see Table 2). A reference to Fig. 3 will show that the number of usable hours is higher during the summer months. A study of the long term meteorological records show that this general behavior should apply to Nemrutdağı also. Thus

Fig.7. Percentage of nights with average seeing smaller than a given value (May to October)

Fig.8. Percentage of dark hours with seeing smaller than a given value (May to October)

Table 5. Superiority order of the sites according to various parameters

Table 6. Comparison of atmospheric seeing on Bakırlıtepe and Roque de los Muchachos (RMO) ("Summer" refers to the interval from May to October)

^a Pole star trail method (not reduced to the same zenith distance)

^a At 21 h local time (see text)

we believe that there has been no bias against Nemrutdağı. What makes this inland mountain inferior to Bakırlıtepe near the Mediterranean is not the number of clear dark hours, but rather the strong wind: A relatively large number of dark hours (and nights) were rejected as nonphotometric just because of the high wind, irrespective of seeing. Tests on other sites in the Southeast need to be made to find out whether this is a local effect. Excepting the relatively severe winter conditions and heavy snowfall in the region, it is not unlikely that an extremely good inland site will be found. At present, however, such mountains, if any, are not easily accessible.

    Bakırlıtepe as a potential observatory site has been compared in Table 6 and 7 with the Roque de los Muchachos Observatory (RMO) on La Palma in the Canary Islands. The data for the RMO have been taken from McInnes (1981), Ardeberg (1983). and Murdin (1985). Although full night observations during winter months are not available for Bakırlıtepe, meteorological observations at 9 p. m. local time (which is a dark hour) are available for two winters. These observations were obtained from a station set up approximately 600 m below the top of the mountain. The number of nights that were classified as photometric at 9 p. m. are given in the last line of Table 7. Statistically, the number of nights that are observed to be clear at 9 p. m. local time is a good measure of the number of completely clear nights (Aslan et al.. 1988). It should be noted, however, that the definitions of a photometric night for the entries in Table 7 are not exactly equivalent. For Bakırlıtepe a night with five or more consecutive photometric hours was defined as photometric. It can be seen from Tables 6 and 7 that Bakırlıtepe compares very favorably with the Roque de los Muchachos Observatory.

Acknowledgements. This work has been supported by TUBITAK. Turkish Scientific and Technical Research Council (proje No. TBAG-607 D), and by the Astronomy Departments of Ankara University. Ege University, Istanbul University, Physiçs Department of Middle East Technical University, and Kandilli Observatory. We are indebted to the directors of the Royal Greenwich Observatory for the loan of two Polaris Trail Telescopes. We thank Mr. B. McInnes of Edinburgh University for visiting several sites in Turkey and for advice. We are grateful to Prof. M. F. Walker of Lick Observatory for providing the standard Polaris trails, for checking our assessments of seeing and for his ready advice. A great deal of computer reduction was carried out by Mr. Z. Müyesseroğlu and Mr. F. Ekmekçi helped with the drawing of the figures.

References

Ardeberg. A.: 1983, ESO Conf. Workshop Proc. Vo. 18, 73

Aslan. Z., Aydın, C., Tunca. Z., Demircan, O., Derman. E.,

Gölbaşı. O., Marşoğlu, A.: 1988. Doğa. Turkish J. Phys. Astrophys (in press)

Barletti, R., Ceppatelli, G., Paterno, L., Righini. A.. Speroni, N.: 1977, Astron., Astrophys. 54, 649

Harlan. E.A., Walker, M. F.: 1965. Publ. Astron. Soc. Pacific 77,

Mclnnes, B.: 1981, Q. J. Roy., Astron. Soc. 22. 266

Murding. P.G.: 1985, Telescopes, Instruments, Research and Services, Roy. Greenwich Obs., p. 48

Walker, M. F.: 1984, Site Testing on Figure Large Telescopes, ESO Conf. Workshop Proc. No. 18