SPATIAL VARIATION OF EXTREME VALUES FOR FREEZING RAIN

It is still a challenge today to get the statistical approach accepted by some transmission line engineers, namely for freezing rain icing l oads. They see heavy icing storms as rare events and they still believ e that in this case a sound engineering judgment is better than poor s tatistics. When they are convinced that statistics are good, such as f or the flood problem, they do not hesitate to accept the probabilistic approach. Developing extreme values distributions of freezing-rain ic ing is however a difficult task. Because of the relatively small dimen sion of major freezing rain storms, freezing-rain icing is not a conti nuous ''variate'' at a particular site. One way of improving the stati stics is to take simple icing measurements on a fine grid stations net work. After only 17 years of measurements with Passive Ice Meters, goo d fit of extreme values is possible if we use ''mesh'' extreme values instead of ''station'' ones. Before pooling extreme values into a ''me sh'' or ''region'' distribution, some attention must be given to the h omogeneity of the data within a given area. This has been done for fou r different regions of the province of Quebec with very conclusive res ults. A mesh of about 50 km seems adequate with a temporal resolution of 12 h. Since transmission lines are spatial constructions, this find ing can easily satisfy the need of the design engineers in defining ic ing loads. And, if some caution is given to identify special topograph ical features, a better knowledge of the spatial variation of icing ma kes also possible better line routing.