Baseline cloudiness trends in Canada 1953–2002

Abstract

Baseline cloudiness trends in Canada for the 1953 to 2002 time period were assessed using raw hourly observations of total cloud cover from eighty‐five airport stations with at least forty years of data. Trends in the count of the number of hours when more than half of the sky dome was observed to be covered by clouds (Mainly Cloudy conditions) increased significantly by several hundreds of hours during the fifty‐year period at numerous stations along the Canada‐U.S border and the few stations located in the Mackenzie Basin. Trends in average annual overall, daytime, night‐time, and seasonal total cloud cover were also computed. The general spatial pattern was similar as in the case of Mainly Cloudy conditions, however, the exact areal extent and significance of the trends would vary from one case to another. In particular, the area of negative trends in British Columbia and the Prairies, gained in significance and extent in the case of daytime cloudiness, while it reversed to weak, mostly positive, trends in the case of night‐time cloudiness. This could correspond to the unprecedented increases in both daily temperature maximum and minimum in these locations during the second half of twentieth century. Significant positive seasonal trends were seen at some stations in southern Canada during spring (coasts) and summer (centre of the country) and the Mackenzie Basin during summer and fall; while significant negative trends were seen at stations in northern Alberta and British Columbia during winter and north‐eastern Canada during summer. The quality of the observations was given careful consideration and potential issues that could affect the continuity of the records of observed total cloud amount are discussed at length. The complete monitoring system was examined: training, affiliation and workload of the observers; automation of the observations in the 1990s by the introduction of the Automated Weather Observing Systems (AWOS); numerous changes in data archiving procedures especially prevalent in the 1990s; underestimation of night‐time cloudiness on nights with insufficient illumination, etc. It was determined that observations, although to a certain degree subjective and not particularly accurate for cloud amounts near 5/10, should be reasonably consistent in the case of human observers (no systematic biases); underestimation of night‐time cloudiness, although relevant if the actual amount is required for certain purposes, probably is not critical in studies concerning cloud trends. Automation, possibly along with the changes in data processing in the 1990s, unequivocally emerged as the least studied subject, however, it is likely to have a considerable impact on the continuity of cloud observations at some stations; in this study about 25% of the stations use AWOS in some way.