Modern optical and infrared astronomical sites are getting used to a flexible way of operation, namely queue modes, allowing astronomical observations in the most appropriate weather conditions for each specific observing scientific program. The forecast of weather conditions is then a mandatory issue to plan in advance the observations queue for each night in order to exploit efficiently the astronomical facilities with the largest high quality data output for scientific exploitation. The precipitable water vapour is the parameter accounting for the infrared (IR) quality of an astronomical site. The temporal fluctuation of this parameter drastically affects the quality of the IR data recorded at ground telescopes. An optical/IR telescope needs the forecasting of the precipitable water vapour for a proper queue scheduling of IR observations. The Roque de los Muchachos Observatory (ORM) on the island of La Palma (Spain) presents an abrupt topography which difficult the forecasting at this astronomical site. We discuss the performance of a mesoscale numerical weather model (WRF, Weather Research and Forecasting) applied to ORM region including the comparison with local precipitable water vapour estimations from GPS (Global Positioning System).