2 Seeing measurements with the DIMM

A simplified and portable version of the standard ESO DIMM described in Sarazin & Roddier1990 was used at Maidanak. It is based on a standard Celestron-11 telescope with a 279 mm primary mirror and a focal length of 2800 mm (Fig. 2). The pupil is covered by a mask with two circular 80 mm diameter apertures 200 mm apart. The mask is fixed above the Schmidt corrector plate. A prism with a wedge angle of 195 $^{\prime\prime}$ is added to one of the holes. The light detector is an ST5 CCD array with $10\times 10~\mu$m square pixels. The instrument produces two stellar images on the detector, 121 pixels apart at the nominal focus (the plate scale is 0.73 $^{\prime\prime}$ per pixel).

Figure 2: Optical scheme of the Differential Image Motion Monitor (DIMM): M1, M2 are the parabolic primary and hyperbolic secondary mirrors, W1 and W2 are the entrance apertures. Two stellar images are separated on the detector by a wedge prism on W2

Routine measurements were carried out on top of a 6 m high pillar placed on the slope of the main summit at the distance 80 m south of the main 1.5 m telescope of the Maidanak Observatory (Fig. 1). Standard DIMM operation consists in pointing a star brighter than +2.5 magnitude, within $30^\circ$ from zenith ($\pm 2$ h hour angle) that allows to reduce observations to the zenith with a sufficient accuracy. Equivalent long exposure seeing is computed from the rms differential image motion in the direction of the baseline (longitudinal) and perpendicular to the baseline (transverse) estimated each 4 minutes from interlaced series of 40 individual frames of $\tau_{1}=10$ and $\tau_{2}=20$ ms. This low data rate is due to the limited speed of transfer in the serial link between the detector controller and the computer. As a consequence, the statistical error on the image motion variance estimate is the dominating noise term, of the order of $\pm 1/\sqrt{39}$ or $\pm 16$%. This random error is fortunately averaged out in long term statistics, but well apparent on individual measurements as shown in Fig. 3, presenting samples of seeing records for a good and bad nights, respectively.

Figure 3: Maidanak seeing measured by DIMM in arcseconds at zenith and 0.5 $\mu$m: examples of a good and a bad nights

The finite duration of exposure time was taken into account by using the relation Sarazin1997a:

$$\sigma^{2}_{\tau=0}=\sigma^{2}_{\tau_{1}} \frac{\sigma^{2}_{\tau_{1}}}{\sigma^{2}_{\tau_{2}}} \; \cdot$$ (1)

Because of the low wind speed, the correction factor $\sigma^{2}_{\tau_{1}}/ \sigma^{2}_{\tau_{2}}$ had typically small values of 1.1 - 1.4 over the entire period of observations at Maidanak.

The full series of seeing measurements obtained at Mt. Maidanak from August 1996 to November 1999 are shown in Fig. 4 and summarized in Table 1. The total number of nights covered over the entire 1180 days period was 483, with an average duration of 8 hours. The corresponding histogram of the seeing estimates at Maidanak, including all DIMM measurements from 1996 to 1999, is shown in Fig. 5.

The median and mean values of the seeing for the entire period of observation were 0.69 $^{\prime\prime}$ and 0.76 $^{\prime\prime}$, respectively. The seasonal trends are shown in Table 2 where monthly medians are computed after merging different years. The winter period, which also provides a smaller percentage of clear nights, has the most turbulent atmosphere.

Figure 4: Whole span of seeing measurements obtained at Maidanak with the DIMM for the period August 1996 to October 1999

Figure 5: Statistical distribution of Maidanak seeing from the measurements shown in Fig. 4

 

 

Table 1: Monthly statistics of the seeing at Maidanak for the period August 1996 to October 1999

Date	N	N	Seeing (arcsec)
	nights	data	25%	50%	75%
1996
Aug.	21	1802	0.52	0.67	0.85
Sep.	23	2094	0.55	0.71	0.90
Oct.	24	2589	0.52	0.72	0.96
Nov.	20	2104	0.48	0.60	0.75
Dec.	17	1953	0.55	0.69	0.89
1997
Jan.	14	1320	0.62	0.98	1.31
Feb.	18	2388	0.58	0.74	0.94
Apr.	9	836	0.63	0.80	1.04
Aug.	16	1827	0.53	0.68	0.87
Sep.	19	3877	0.54	0.68	0.84
1998
Jun.	4	150	0.59	0.81	1.02
Jul.	30	2633	0.53	0.68	0.89
Aug.	21	2301	0.57	0.72	0.92
Sep.	28	3213	0.53	0.67	0.84
Oct.	27	3753	0.56	0.72	0.93
Nov.	19	2672	0.50	0.63	0.79
Dec.	15	1492	0.59	0.73	0.93
1999
Jan.	15	1299	0.61	0.78	0.96
Feb.	6	665	0.72	0.90	1.12
Mar.	7	684	0.52	0.64	0.79
Apr.	11	1057	0.61	0.79	1.00
May	21	2015	0.53	0.66	0.86
Jun.	25	2083	0.55	0.69	0.90
Jul.	24	1785	0.56	0.70	0.87
Aug.	10	1075	0.59	0.75	0.96
Sep.	19	2602	0.54	0.69	0.90
Oct.	20	2795	0.50	0.62	0.77
All	483	53064	0.55	0.69	0.90

   

Table 2: Seasonal statistics of the median seeing at Maidanak for the period August 1996 to October 1999 sorted by month

Month	N nights	N data	$\beta_{\rm atm}$, arcsec
Jan.	29	2619	0.84
Feb.	24	3053	0.77
Mar.	7	684	0.64
Apr.	20	1893	0.80
May	21	2015	0.66
Jun.	29	2233	0.70
Jul.	54	4418	0.69
Aug.	68	7005	0.70
Sep.	89	11786	0.68
Oct.	71	9137	0.68
Nov.	39	4776	0.62
Dec.	32	3445	0.71
All	483	53064	0.69

A comparison of the present results with the previous estimations of seeing at Mt. Maidanak shows that past data sets are unfortunately not entirely consistent. However, the reasons for the discrepancy with some Polar star motion measurements has been understood and these data can probably be used after re-calibration to analyze the long term stability of the site Ilyasov et al.1999.

Four international astronomical observatories (including Maidanak) have systematic seeing measurements taken with cross-calibrated DIMMs. In Fig. 6 we present the cumulative distributions of seeing at Paranal and La Silla measured over the period [1989-1995] Sarazin2000, Roque de los Muchachos Observatory (ORM) at La Palma over the period [1994-1998] Munoz-Tuñon et al.1999,Wilson2000 and at Maidanak [1996-1999]. The median value of the seeing at ORM measured by ING and DA/IAC DIMMs is 0.69 $^{\prime\prime}$, whereas at La Silla and Paranal Observatories it is 0.87 $^{\prime\prime}$ and 0.66 $^{\prime\prime}$, respectively. We can conclude that the atmosphere above Mt. Maidanak is on a par with the conditions at the world's leading observatories. Regarding seasonal variations, Fig. 7 shows that Paranal and Maidanak, belonging to different hemispheres, are slightly anti-correlated. However, the seasonal seeing modulation is only about $\pm$10%.

Figure 6: Statistical distribution of the seeing at Maidanak [1996-1999]: comparison with ESO observatories of La Silla and Paranal [1989-1995] and Roque de los Muchachos Observatory at La Palma [1996-1999]

Figure 7: Monthly median seeing at Maidanak in [1996-1999] (full line) and at Paranal (squares on dotted line) with the minimum and maximum monthly averages (dotted lines) during the period [1993-1999]