Observed Precipitation Trends Inferred from Canada's Homogenized Monthly Precipitation Dataset

This study first developed a comprehensive semiautomatic data homogenization procedure to produce gap-infilled and homogenized monthly precipitation data series for 425 long-term/critical stations in Canada, which were then used to assess Canadian historical precipitation trends. Data gaps in the 425 series were infilled by advanced spatial inter-polation of a much larger dataset. The homogenization procedure repeatedly used multiple homogeneity tests without and with reference series to identify changepoints/inhomogeneities, the results from which were finalized by manual analysis using metadata and visual inspection of the multiphase regression fits. As a result, 298 out of the 425 data series were found to be inhomogeneous. These series were homogenized using quantile matching adjustments. The homogenized dataset shows better spatial consistency of trends than does the raw dataset. The improved gridding and regional mean trend esti-mation methods also provide more realistic trend estimates. With these improvements, Canadian historical precipitation trends were found to be dominantly positive and significant, except in central-south Canada where the trends are generally insignificant and small with mixed directions. For annual precipitation, the largest increases are seen in southeastern Canada and along the Pacific coast; however, the largest relative increases (in percent of the 1961-90 mean) are seen in northern Canada. The largest trend difference between northern and southern Canada is seen in winter, in which significant increases in the north were matched with significant decreases in the south.

Automated summary

This study developed a semiautomatic data homogenization procedure to assess Canadian historical precipitation trends. The results showed that the trends were predominantly positive and significant, except in central-south Canada where the trends were generally insignificant and small.