In this paper we investigate the correlation between the atmospheric perturbations at Paranal Observatory and the Chilean coast tides, which are mostly modulated by the 14-day syzygy solar-lunar tidal cycle. To this aim, we downloaded 15 years (2003-2017) of cloud coverage data from the AQUA satellite, in a matrix that includes also Armazones, the site of the European Extremely Large Telescope. By applying the Fast Fourier Transform to these data we detected a periodicity peak of about 14 days. We studied the tide cycle at Chanaral De Las Animas, on the ocean coast, for the year 2017, and we correlated it with the atmospheric perturbations at Paranal and the lunar phases. We found a significant correlation (96%) between the phenomena of short duration and intensity (1-3 days) and the tidal cycle at Chanaral. We then show that an atmospheric perturbation occurs at Paranal in concomitance with the low tide, which anticipates the full (or the new) moon by 3-4 days. This result allows to improve current weather forecasting models for astronomical observatories by introducing a lunar variable.