A quality control approach to better characterize the spatial distribution of snow depth over New Brunswick, Canada

Abstract

AbstractThe aim of this work is to develop, from a high‐resolution climate analysis, a quality control standard methodology applied to manual snow cover (HS) series managed by the Snow Survey Database in New Brunswick (Canada). The database collected snow depth data biweekly starting at the end of January until the end of April. A 30‐year (1981–2010) analysis of 60 weather stations belonging to two independent meteorological networks was performed. A quality control of the climatic series was performed to evaluate the homogeneity. Three snow depth climatic areas were defined by means of two geostatistical methods (Kriging and Cluster analyses) applied on monthly snow depth, precipitation and temperature data series. Then, for each cluster, the climatological thresholds that characterize a snow fall event during the cold months were detected. Subsequently, a quality control on the daily snow depth series recorded during the January to April period was performed. For each daily series, outlier values were identified by checking both the sudden day‐to‐day changes and extreme thresholds (95th percentile). The quality control was then carried out to the manual series and the observed doubtful events were compared with the snow depth values recorded in the nearby stations. The results show that for the daily snow depth series, the highest number of suspect events was recorded during the months of March and April, and the analysis also shows that there are rain‐on‐snow events. As for the manual records, questionable snow depth errors randomly distributed in the series were highlighted. Finally, in order to improve the spatial distribution of stations located in the Canadian territory, the results give evidence that, thanks to the high‐resolution climatic analysis, the proposed approach provides all the benchmarks required to conduct a quality control of snow depth series in absence of other auxiliary variables.