Wildfires cause significant losses in both material value and human lives and are becoming a frequent problem, also due to climate change. Preventing their occurrence or reducing their effect is a crucial issue. This work is part of the development of a framework that uses optimisation models for the prepositioning of resources, and for resource movement during the suppression phase. These problems involve several sources of uncertainty, one of which is the behaviour of the wind (direction and speed). It is then of the utmost importance to obtain statistical information about wind speed and direction. The aim of this work is to model historical wind data of a weather station located in a case study region in the North of Portugal (Luzim) and, for that purpose, several statistical techniques were used (e.g. descriptive statistics, correlation analysis, goodness of fit tests) in order to better understand it. It was decided to use the empirical distributions of wind speed and wind direction instead of fitting theoretical distributions, like the Weilbull Distribution used in the literature, as there is a large quantity of data available. For this, the data was split into three wildfire risk seasons: low (November, December, January and February), medium (March, April, May and October) and high (June, July, August and September). Using the data from the high risk season, the values of wind speed (5 values) and of wind direction (8 values) and their respective estimated probabilities were used in order to create scenarios that were later incorporated into the optimisation models.