New Astronomy Reviews 42 (1998) 451–454 The temporal behaviour of seeing a b ˜ ˜´ J. Vernin , C. Munoz-Tunon a ´ UMR 6525 Astrophysique, Universite de Nice –Sophia Antipolis, 06108 Nice Cedex 2, France b ´ Instituto de Astrof isica de Canarias, 38200 La Laguna, Tenerife, Spain Abstract Using a large amount of data gathered during previous seeing campaign at ORM, we analyse the temporal evolution of seeing in order to find out whether predictions could be made over a short time interval of a few hours. The first results are presented.  1998 Elsevier Science B.V. All rights reserved. 1. Introduction sents the true seeing? The answer requires a knowl- edge of the statistical distribution of the seeing. Seeing is known to vary over a large range of time Being a positive variable, its distribution is more scales: minutes, hours, days, seasons and years. This is due to the physical process which gives rise to atmospheric instabilities and to the generation of optical turbulent layers. Coulman et al. (1995) proposed a phenomenological description of the apparition of the already observed thin laminae, triggered by gravity waves. These waves are gener- ated by non-linear interactions between the air flow at various altitudes and the orography. So it is not surprising that the seeing, which is the superposition of several laminae, might be of a random nature. For seeing prediction, it is tempting to find statistical laws which describe its temporal behav- iour. To manage the operational ‘‘flexible schedul- ing’’ of astronomical observations one needs to restrict the study of the time scale range from say one hour to one day. To our knowledge only three ˜ ˜´ authors (Racine, 1996; Munoz-Tunon et al., 1997; Sarazin, 1997) have worked on this problem. We will review briefly their conclusions here. Fig. 1. Probability density function of the seeing corresponding to one month’s observations at the ORM. The dotted line represents 2. First-order statistics the result of the deconvolution of the measured psf assuming Of the mean and the median, which better repre- Gaussian noise. 1387-6473 / 98 / $ – see front matter  1998 Elsevier Science B.V. All rights reserved. PII: S1387-6473(98)00051-7