Michel Slivitzky

Internal Variability of the Canadian RCM’s Hydrological Variables at the Basin Scale in Quebec and Labrador

AbstractThe effect of a regional climate model’s (RCM’s) internal variability (IV) on climate statistics of annual series of hydrological variables is investigated at the scale of 21 eastern Canada watersheds in Quebec and Labrador. The analysis is carried out on 30-yr pairs of simulations (twins), performed with the Canadian Regional Climate Model (CRCM) for present (reanalysis and global climate model driven) and future (global climate model driven) climates. The twins differ only by the starting date of the regional simulation—a standard procedure used to trigger internal variability in RCMs. Two different domain sizes are considered: one comparable to domains used for RCM simulations over Europe and the other comparable to domains used for North America. Results for the larger North American domain indicate that mean relative differences between twin pairs of 30-yr climates reach ±5% when spectral nudging is used. Larger differences are found for extreme annual events, reaching about ±10% for 10% and ...

Sensitivity of runoff and projected changes in runoff over Quebec to the update interval of lateral boundary conditions in the Canadian RCM

An analysis was carried out of the sensitivity of runoff simulations from version 4.2 of the Canadian Regional Climate Model (CRCM) to the frequency of lateral boundary condition (LBC) forcing at 6 and 12-hour intervals. The motivation for this study was that some climate model output may only be available at a 12-hour interval and it is important to know if CRCM runs with these outputs are comparable to runs made with 6-hourly forcing. The LBC sensitivity was assessed over two different regional domains (North America and Quebec) for annual runoff simulated over 21 river basins located in the Quebec/Labrador peninsula. The sensitivity results were compared with the CRCMs internal variability and natural climate variability to reach conclusions about the relative importance of LBC update frequency. The results show that LBC frequency can have a significant influence on mean annual runoff over the investigated basins when the simulation domain is relatively small, as in the case of the Quebec, but not for the larger North American (AMNO) domain runs. The entire ensemble of five members of the Canadian Coupled Global Climate Model (CGCM3) can therefore be safely used to generate dynamically downscaled projections over the basins, even though three of the members were archived at a 12-hourly interval. Climate projections for the 2041-2070 horizon (with SRES-A2), from a five-member ensemble of CRCM 45-km runs performed over the AMNO domain (driven by each of the five CGCM3 members), project an increase of annual runoff over all investigated river basins with the largest changes towards the north. This ensemble also provides an estimate of uncertainty of projected basin runoff change related to natural variability, but there remains a need to consider additional projections (more RCMs, more driving GCMs) to produce a more complete assessment of uncertainty.

Climatic Changes during the 20th Century on the Laurentian Great Lakes and their Impacts on Hydrologic Regime

A number of authors, looking at the time series of precipitation and temperature for the Laurentian Great Lakes, have shown that these series experienced shifts during the last 90 years. However analyses of Net Basin Supply (NBS) for each of the five individual lakes always had difficulties in pinpointing some of these changes which could have been caused by changes in the climate forcing variables or by errors in the calculation of NBS.

Investigation of the hydrologic cycle simulated by the Canadian Regional Climate Model over the Québec/Labrador territory

This paper discusses the surface hydrology of a two-year simulation conducted over the Quebec/Labrador territory with the Canadian Regional Climate Model (CRCM). The CRCM was driven by NCEP (National Center for Environmental Protection) atmospheric objective analyses and was run at a 30 km horizontal grid-point spacing. Model results are compared with available observations for the period from June 1992 to June 1994. A spatial analysis of the CRCM annual runoff shows that the model is in good agreement with basin observations although the simulated precipitation and surface evaporation are too important. The main features of seasonal runoff variation are captured by the CRCM with differences mainly attributed to the oversimplification of the surface processes in the single-layer surface scheme. We also find that spring peak runoffs are quite well synchronized with the observations but with some overestimation when a lumped hydrologic model is fed by CRCM meteorological outputs. Despite the fact that the hydrologic model was not recalibrated with the CRCM data, this first experience shows the potential use of the CRCM to feed a hydrologic model. In summary, even though our analysis shows the usefulness of the CRCM simulated surface hydrology information, results from this experience indicate the need to refine the surface scheme to represent more accurately surface hydrology processes.