Forecasting Precipitable Water Vapour and Cirrus Cloud Cover For Astronomical Observatories: Satellite image processing guided by synoptic model dissemination data

摘要

Astronomical observatories are extremely dependent on sky transparency. As the expensive new very large telescopes enter into operation, flexible observing modes are being introduced, which allow each 'observing block' to be shceduled at the most appropriate time. In such modes, it makes sense to develop tools for forecastign ambient conditions. We present here the operational water vapour and cirrus cloud forecast model developed for ESO observatories in Northern Chile. Because the observtories are located well above the inversion layer, nebulosity is mainly due to high altitude cirrus clouds which have to be detected as early as possible. For that purpose, a model has been developed using 3-hourly GOES-8 satellite images in the water vapour channel combined with the 00UT and 12UT analyses and forecasts fromt he European Centre for Medium-range Forecasting (ECMWF) over a 45x30 degree area including the observatories, the Andes to the East as well as a large area of the Pacific Ocean to the West. In the water vapour channel, radiation emitted by mositure in the middle troposphere is aborbed and re-emitted by high altitude cirrus clouds at low temperatures thus producing high (above 100percent) upper tropospheric humidity (UTH) values. Empirically, UTH thresholds were determiend for deciding on the presence of transparent or opaque cirrus. The observatory "sky" at analysis time is defined by a circular area of about 32 km in radius centred over the site. The motion algorithm, whcih takes into account both the translation and the development of the pressure systems as forecasted by ECMWF, defines a movement vector that is used to determine the image areas for the forecast periods. The image pixels in the forecast areas are classified in terms of water vapour and cloudiness and attributed a weight proportional to their likelihood to cross the observatory sky in teh course of a 3-hour long forecast period. We discuss the performance of the model after one year of operation, including comparisons with local sky transparency monitored from the groun din the visible part of the spectrum.