Learn about CanDCS-M6, the most recent dataset to come to ClimateData.ca and why it replaced CanDCS-U6.
The Canadian Downscaled Climate Scenarios-Multivariate dataset for CMIP6 (CanDCS-M6 or M6 for short) is a Canada-wide statistically downscaled dataset developed using the CMIP6 global climate model (GCM) ensemble and the emissions scenarios known as Shared Socioeconomic Pathways (SSPs). Outputs from each climate model were statistically downscaled using an improved downscaling method and target dataset.
Statistically downscaled datasets use high resolution observations to adjust climate model data to better capture small-scale variations in the climate1,2. By making use of statistical relationships between coarse-resolution outputs from GCMs and observed climate data, statistical downscaling can adjust climate projections to a spatial resolution that is more relevant for local climate change adaptation planning1.
Almost all temperature and precipitation-based data products (e.g., maps, time series plots, and download and analyze pages) on ClimateData.ca are now derived from the M6 dataset, which represents an advancement in statistically downscaled climate data for local adaptation purposes in Canada. The M6 dataset released on ClimateData.ca in the fall of 2024 replaced its predecessor, CANDCS-U6 (U6) in November 2024. Beyond November 2024, users will be able to access the U6 dataset on PAVICS.
Both M6 and U6 datasets include daily minimum and maximum temperature and precipitation data, which have been used to derive the 30+ climate indices available on ClimateData.ca. The main differences between the original U6 and improved M6 datasets are the statistical downscaling techniques and target datasets used. The M6 dataset also includes an additional emissions scenario, SSP3-7.0.
M6 is a multivariate dataset that was derived using a statistical downscaling technique called n-dimensional Multivariate Bias Correction (MBCn), which prioritizes preserving the relationships between climate variables. This technique increases confidence particularly for indices such as freeze-thaw cycles and snowfall, which are derived from both temperature and precipitation.
The previously available downscaled dataset, U6, is a univariate dataset, meaning that downscaling is performed for each climate variable independently, in this case using the BCCAQv2 downscaling method. This method is highly effective for downscaling a single variable at a time, such as daily maximum temperature.
A target dataset is a set of high-resolution historical data used to adjust climate model simulations to better reflect local climate conditions.
U6 uses NRCANmet V1 as the target dataset3,4. This dataset is produced by Natural Resources Canada, and is available at a gridded spatial resolution of approximately 6 x 10 km (it is also commonly referred to as “ANUSPLIN”).
M6 uses an improved target dataset, PCIC-Blend, which has the same spatial resolution as NRCANmet. This dataset was developed by the Pacific Climate Impacts Consortium (PCIC) by combining pre-existing datasets that each have their own advantages in certain parts of the country. PCIC-Blend combines PNWNAmet over western Canada (daily maximum and minimum temperature, precipitation) with NRCANmet V2 (daily maximum and minimum temperature) and NRCANmet-Adjusted (precipitation) over central and eastern Canada3. The three datasets were smoothly blended over a transition region east of the Rocky Mountains from the Arctic Ocean to the US border. While the NRCANmet datasets provide a good description of these variables over central and eastern Canada, PNWNAmet performs better in the west when compared to station observations.
Learn more about the M6 dataset by reading the article, Multivariate Canadian Downscaled Climate Scenarios for CMIP6 (CanDCS- M6).
M6 includes the three emissions scenarios in U6 (SSP1-2.6, SSP2-4.5, SSP5-8.5) plus a fourth scenario, SSP3-7.0. Results are available for the same 26 GCMs in the U6 ensemble, except for SSP3-7.0, where only 24 of the 26 models were available.
In short, M6 data are better for calculating climate indices that use more than one variable, such as the Standardized Precipitation Evapotranspiration Index (SPEI). Additionally, the M6 data provide more accurate precipitation data for Western Canada.
Since the downscaling method used to create M6 allows for the consideration of multiple variables simultaneously, it better preserves the relationships between variables. For this reason, M6 projections of indices that are derived from multiple variables (e.g., freeze-thaw cycles and snowfall) and compound events5 are more certain than those derived from U6. For single-variable indices (e.g., Hottest Day), M6 performs as well as U6.
The M6 dataset also capitalizes on the improved representation of precipitation across Canada in the PCIC-Blend target dataset5. Figure 1 illustrates the general differences in precipitation between the two datasets for three different time periods: historical (1971-2000), mid-century (2041-2070) and late-century (2071-2100). These changes are almost entirely due to the change in target dataset (Figure 2), particularly in Western Canada, where it is generally wetter.
All maps, time series plots, and custom analyses on ClimateData.ca have been replaced with M6 data. Users can access the U6 dataset on PAVICS.
The M6 dataset is based on the same set of GCMs used for U6. Aside from addressing some known issues related to precipitation in Western Canada, these new data do not change the story for how climate change will impact communities in Canada. Instead, this new dataset will allow the development of more robust compound climate indices (combinations of temperature and precipitation) which may be useful for adaptation planning (e.g., hot and dry or hot and wet conditions).
For additional support and information regarding the use of M6 or U6 data, contact the Climate Services Support Desk.