Monday, 15 January 2007
Assessment of discontinuities due to joining precipitation observations in Canada
Exhibit Hall C (Henry B. Gonzalez Convention Center)
Long-term homogeneous precipitation records are required for climate trend studies. About 460 evenly distributed stations were carefully selected across the country for the preparation of the second version of the Adjusted Historical Canadian Climate Database (still under development). Daily rain gauge and snow ruler data were adjusted separately for all known issues related to changes in rain gauge type, snow density and trace measurements. In the most recent work, joining precipitation observations from nearby stations is examined. In Canada, when the station is relocated even to close vicinity, a new station identification number is often assigned to the new location and the two observations segments can be merged into one. Also when a station is closing, the data can be joined to a neighbour (within 20 km) with similar characteristics in order to produce a longer time series. The new site can be located at a different elevation and/or with a different exposure. Therefore, joining precipitation observations can create artificial discontinuities affecting the trend. This work presents a preliminary method to determine if the join of the observations has generated an artificial step. Adjustment procedure for monthly total rain and snow will also be discussed.