Evaluation of snow cover simulated by the CRCM5 for the CROQ database

Snow cover accumulation can have significant repercussions on a number of sectors including the built environment, transportation, energy, tourism, etc. In order to assess the impact of climate change, a good understanding of past climate is required. In Québec, especially north of the St. Lawrence Valley, it is difficult to characterize past climate because meteorological observations are sparse in time and space, and available only over short periods of time.

The CROQ project (Climate Reconstruction at Ouranos for Québec) was developed to produce a database of historical climate data, including snow cover, across the province, based on simulations of the Canadian Regional Climate Model (CRCM5), adopted by Ouranos and currently used to produce regional climate simulations.

• Assess the ability of the Canadian Regional Climate Model (CRCM5) to reproduce snow cover over Québec at a spatial resolution of 15 km for historical simulations produced as part of the CROQ project;

• Evaluate a historical snow cover data set in order to better identify the uncertainties of the observations;

• Provide a detailed study that will be useful for users.

• Validate the robustness of the CRCMC5 by statistical and comparative studies between different simulations produced at Ouranos;

• Compile available historical data;

• Evaluate the spatial distribution of CRCM5 snow cover across the province in comparison to observations;

• Evaluate interannual variation by comparison to observations for each year of the study (1980-2014) at six locations representing the range of topography and climatology found in the region.

A compilation of different available historical databases reveals a disparity in the spatial distribution of snow cover in Québec, especially in the north. This demonstrates the difficulty of producing an accurate picture of snow cover, which is exacerbated by the large amount of missing data. In this study, we evaluate snow cover simulated by the Canadian Regional Climate Model (CRCM5) at a 15 km resolution over a domain covering Québec. The simulated spatial distribution is contained within the envelope of possibilities indicated by the set of observations. Furthermore, in areas where there is better consensus among the various observations, the CRCM5 produces good estimates for both snow season length and maximum snow cover.

Comparison with data from six Environment and Climate Change Canada (ECCC) weather stations across eastern Canada (Montréal, Kuujjjuak, Kuujjuarapik, St John’s, Wabush Lake and Geraldton) shows that the CRCM5 is consistent with the observed annual series of maximum snow cover (in water equivalent) but with a slight overestimation. With regard to annual days of snow cover, the CRCM5 generally provides a good reproduction of the observed annual series at each of the six stations although a slight overestimation is observed.

Figure 1. Annual days of snow cover (minimum of 4 mm snow water equivalent) at the Kuujjuarapik station (Northern Québec). The CRCM5 simulation (black) is compared to MERRA-2 reanalysis (green) and Blended-5 data (blue), using the closest grid cell, as well as to observations at the ECCC station (purple).

Figure 1 presents the example of Kuujjuarapik, where the simulated variable is compared to data obtained from MERRA-2 reanalysis (Molod¹ et al. 2015), Blended-5 (a combination of 5 databases; Mudryk² et al. 2015) and observations at the ECCC station. The distribution between the first half of winter (August 1-January 31) and the second half of winter (February 1-July 31) shows that the model tends to overestimate the number of days of snow cover during the first half of winter, while in the second half the number generally agrees with the observations. These results, validated with available data, could serve as a data set of fine-resolution snow cover projections with homogeneous coverage over Québec.

¹ Molod, A., L. Takacs, M. Suarez and J. Bacmeister (2015) Development of the GEOS-5 atmospheric general circulation model: evolution from MERRA to MERRA2. Geoscientific Model Development, 1339–1356, doi:10.5194/gmd-8-1339-2015.

² Mudryk, L., C. Derksen, P. Kushner and R. Brown (2015) Characterization of northern hemisphere snow water equivalent datasets, 1981–2010. Journal of Climate, 28, 8037–8051. doi:10.1175/JCLI-D-15-0229.1.

• UQAM ESCER

• ECCC/CPS Climate Processes Section