Weather forecasting models are used to show atmospheric conditions by computing changes in meteorological parameters on a 3D atmospheric grid model. Physical laws are used to determine behaviour: fluid mechanics, variation in water phase, turbulence, radiation, and atmospheric interaction with space, the continents and the oceans. The initial conditions are determined by assimilating variational data including a large volume and wide variety of in situ observations obtained from remote detection systems. The weather forecasts available on data.shom.fr only contain "wind at 10m" and "atmospheric pressure at sea level" parameters. A land/sea mask is then applied to exclude non-maritime forecasts. These forecasts are taken from 2 different types of models: ARPEGE for world and European scale input (0.5° to 0.1° resolution) and AROME for mainland France (1.3 km resolution). ARPEGE is a global hydrostatic spectral model, with variable horizontal resolution (centred on France), vertical finite element modelling and hybrid vertical coordinates. ARPEGE is an integral part of the Arpège-IFS software package, designed, developed and maintained by Météo-France in cooperation with ECMWF (European Centre for Medium-Range Weather Forecasts - http://www.ecmwf.int/). AROME is a non-hydrostatic spectral model for weather forecasts in mainland France, with finite difference modelling for vertical input and hybrid coordinates. AROME was developed by Météo-France thanks to close national (CNRS) and international (CEPMMT, Aladin, Hirlam) cooperation programmes on the basis of the Méso-NH research model and the dynamic core of the Aladin model.