De-aggregated hazard of freezing rain events

Abstract The objective of the work presented here is to improve estimates of atmospheric icing hazards, specifically for radial ice accumulation on electrical transmission lines, by clustering freezing rainstorms into different groups and using de-aggregate statistical analysis. In this paper, freezing rain storms were clustered based on anomaly maps constructed using NCEP reanalysis data of 1000 hPa to 500 hPa geopotential heights or SLP. The procedure is demonstrated with data from Montreal, as well as a superstation of Montreal, Quebec City, and Ottawa together. The physical meaning of the different clusters was also analyzed and discussed in terms of wind speed, total precipitation or ice accumulation, air mass positions, and compared with Rauber's archetypical patterns. The results presented indicate that the clusters have statistical and physical significance. Clustering based on either SLP or 1000 hPa to 500 hPa height anomalies produces clusters with different precipitation characteristics which can be explained on a physical basis. Clustering Montreal storms based on 1000 hPa to 500 hPa height anomalies produced one cluster containing the five worst ice storms that occurred in Montreal during that time period. This cluster was characterized by an expansive zone of 1000 hPa to 500 hPa height positive anomalies over Montreal that extend far north. It also shows significant SLP anomalies and cold air damming. The de-aggregated hazard analysis is able to demonstrate that the relative contribution of each cluster changes for more extreme events. The de-aggregated analysis produces optimal statistical fits for clusters and introduces robustness in the estimation of the return period over a range threshold. Finally, a small bias is suggested in the single population analysis where the return periods are consistently lower over a range of quantiles and thresholds.