Teleconnections between reconstructed long Canadian streamflow series and climate modes of variability over the 1851-2011 period

Abstract:

The observation of dry and wet decades at the regional scale is, in different parts of the world, explained by teleconnections between regional and global climate modes of variability. This decadal variability raises interesting issues, such as the long-term planning of hydropower generation. Indeed, quantifying these teleconnections requires analyzing long series, which are rare and often lacking in remote areas such as Northern Canada, where some of the largest hydropower systems in the world are found. In northern Canada, no meteorological station close to any large dam has more than 50 years of data. These few decades of observations are not sufficient to adequately study multi-decadal hydrological variability and teleconnections. To cope with this issue, several reconstructions methodologies have been proposed, using for example natural proxies (e.g. tree-ring). Here, we applied a reconstruction methodology based on an historical atmospheric reanalysis on several Canadian catchments, to produce daily streamflow series. These series were then used to study teleconnections at the catchment-scale over the 1851-2011 period.

Firstly, a climatic reconstruction method (ANATEM) was applied, resulting in an ensemble of daily climatic series for each studied catchment. These ensembles were obtained by resampling the observed catchment climatic series based on both season and similarity of geopotential height fields. The NOAA 20^th Century Reanalysis ensemble has been used to estimate similarities between daily geopotential fields over the 1851-2011 period. Secondly, a hydrological model was used to transform the climatic ensembles into streamflow ensembles. The studied catchments are located in two different - and distant - Canadian regions (British-Columbia and Québec provinces), allowing quantifying potential teleconnections with different climatic modes of variability, such as the NAO, ENSO or the PDO for each region. Thus, correlations between the variability of the reconstructed streamflow series and the different climate indices have been estimated at the decadal scale, for each studied catchment. Finally, the high temporal resolution of the applied reconstruction method (daily time-step) allowed studying the variability and teleconnection of seasonal streamflows over the past 160 years.