In the rapidly expanding market of wind energy, sites in cold climates are becoming popular because of their high wind potential. However, several challenges are associated with energy production in those sites. Such challenges not only include low temperature and high winds, but also multiple icing events. To cope with the negative impacts of icing, wind turbine manufacturers nowadays include ice protection systems (IPS) to their products. It has been observed that the performances of those systems can be greatly influenced by the ambient conditions. In many occasions, under harsh conditions, turbines IPS may not able to prevent ice formation on the blades nor to remove ice that is already formed. Multiple numerical analysis were made on the IPS performance but none of them relies on actual field data. In this paper, the relationship between the turbine stoppage time, wind speed and ambient temperature was studied on a wind farm of over 80 turbines in eastern Canada. The test site is subject to harsh weather and long recovery time and can therefore provide interesting data on icing events. A simplified 1-D heat transfer model of the blades during recovery time has been developed and then compared to the field data. The model was used to generate IPS performance envelopes to delimit, on a temperature vs. wind speed chart, the conditions where the IPS are expected to operate successfully. This analysis showed the importance of having an IPS adapted to the conditions encountered in a given site. To help wind turbine manufacturers design their IPS, a set of distributions of the most relevant meteorological parameters is presented. This set of data has been retrieved using a custom made meteorological conditions monitoring station (MCMS) installed on one of the wind turbines' nacelle. Wind turbine designers, developers and operators can greatly benefit from the data presented in this paper.