5 Dark time sky brightness at Paranal

Due to the fact that our data set collects observations performed under a wide range of conditions, in order to estimate the zenith sky brightness during dark time it is necessary to apply some selection. For this purpose we have adopted the following criteria: photometric conditions, airmass $X\leq$ 1.4, galactic latitude |b|> 10$^\circ$, time distance from the closest twilight $\Delta t_{{\rm twi}}>$ 1 hour and no moon (FLI = 0 or $h_{\rm M}\leq-$18$^\circ$). Unfortunately, as we have mentioned in Sect. 4.1, very few observations have been carried out at $\vert\beta\vert>$ 45$^\circ$ and hence we could not put a very stringent constraint on the ecliptic latitude, contrary to what is usually done (see for example Benn & Ellison 1998). To limit the contribution of the zodiacal light, we could only restrict the range of helio-ecliptic longitude ( $\vert\lambda -\lambda _\odot \vert\geq$ 90$^\circ$). The results one obtains from this selection are summarized in Table 4 and Fig. 7, where we have plotted the estimates of the sky brightness at zenith as a function of time. Once one has accounted for the zodiacal light bias (see below), the values are consistent with those reported for other dark sites; in particular, they are very similar to those presented by Mattila et al. (1996) for La Silla, which were also obtained during a sunspot maximum (February 1978). As pointed out by several authors, the dark time values show quite a strong dispersion, which is typically of the order of 0.2 mag rms. Peak to peak variations in the V band are as large as 0.8 mag, while this excursion reaches 1.5 mag in the I band.

Figure 7: Zenith corrected sky brightness measured at Paranal during dark time (thick dots) from April 1st, 2000 to September 30th, 2001. The selection criteria are: |b|> 10$^\circ$, $\vert\lambda -\lambda _\odot \vert\geq$ 90$^\circ$, $\Delta t_{{\rm twi}}>$ 1 hour, FLI = 0 or $h_{\rm m}\leq -$18$^\circ$. Thin dots indicate all observations (corrected to zenith). The horizontal dotted lines are positioned at the average values of the selected points. The histograms trace the distribution of selected measurements (solid line) and all measurements (dotted line), while the vertical dashed lines are placed at the average sky brightness during dark time.

 

 

Table 4: Zenith corrected average sky brightness during dark time at Paranal. Values are expressed in mag arcsec^-2. Columns 3 to 7 show the rms deviation, minimum and maximum brightness, number of data points and expected average contribution from the zodiacal light, respectively.

Filter	Sky Br.	$\sigma$	Min	Max	N	$\Delta m_{{\rm ZL}}$
U	22.28	0.22	21.89	22.61	39	0.18
B	22.64	0.18	22.19	23.02	180	0.28
V	21.61	0.20	20.99	22.10	296	0.18
R	20.87	0.19	20.38	21.45	463	0.16
I	19.71	0.25	19.08	20.53	580	0.07

In Sect. 4.1 we have shown that the estimates presented in Table 4 are surely influenced by zodiacal light effects of low ecliptic latitudes. To give an idea of the amplitude of this bias, in the last column of Table 4 we have reported the correction  $\Delta m_{{\rm ZL}}$ one would have to apply to the average values to compensate for this contribution. This has been computed as the average correction derived from the data of Levasseur-Regourd & Dumont (1980), assuming typical values for the dark time sky brightness: as one can see, $\Delta m_{{\rm ZL}}$ is as large as $\sim$0.3 mag in the B passband.

The sky brightness dependency on the ecliptic latitude is clearly displayed in Fig. 8, where we have plotted the deviations from the average sky brightness (cf. Table 4) for B, V and R passbands, after applying the correction $\Delta m_{{\rm ZL}}$. We have excluded the I band because it is heavily dominated by airglow variations, which completely mask any dependency from the position in the helio-ecliptic coordinate system; the U data were also not included due to the small sample. For comparison, in the same figure we have over imposed the behaviour expected on the basis of Levasseur-Regourd & Dumont (1980) data, which have been linearly interpolated to each of the positions $(\lambda-\lambda_\odot,~\beta)$ in the data set. As one can see, there is a rough agreement, the overall spread being quite large. This is visible also in a similar plot produced by Benn & Ellison (1998, their Fig. 10) and it is probably due to the night-to-night fluctuations in the airglow contribution.

Figure 8: B, V and R dark time sky brightness variations as a function of ecliptic latitude. The solid lines trace the behaviour expected from Levasseur-Regourd & Dumont (1980) data for the different passbands.