Representativeness of the Precipitation Observing Network for Monitoring Precipitation Change and Variability in Canada

While there are thousands of precipitation stations in the Canadian climate archive, it has been challenging to estimate regional and national averages of precipitation for the purpose of monitoring climate change and variability, because of evolution of the monitoring network and generally sparse network density, in particular in the North. Changes in the observing network have resulted in segmentation and inhomogeneities in precipitation records. We used monthly precipitation from the Canadian Regional Climate Model (CanRCM4) large ensemble simulations as a proxy of observations with complete spatial and temporal coverage. By comparing results from the complete-coverage dataset with versions masked by observational coverage, we examined the representativeness of two long-term precipitation datasets for the estimation of annual mean precipitation at regional and national levels for the purpose of monitoring precipitation change and variability in Canada. We also analysed the implications of changes in the network and the possible added value of data processing, such as in-filling through spatially interpolating station records. We find that at the best coverage of approximately 450 precipitation stations in the Adjusted and/or Homogenized Canadian Climate Data dataset, station coverage is, in general, adequate for the purpose of monitoring long term precipitation trends. However, this capability is severely compromised if station density changes (reduces) with time or if there is a substantial number of missing values over time. The addition of station records in regions already better represented (i.e. regions with more population) does not provide significant improvement. In-filling over space through spatial interpolation does add value, provided that there is sufficient information in the station network. Our analysis demonstrates the importance of maintaining a consistent long-term network with sufficient station density for the purpose of monitoring climate change and variability.