Statistically downscaled climate scenarios and indices from CMIP6 global climate models

On this page

• Overview
• Downscaling methodology and data
• CanDCS-U6
• CanDCS-M6
• Climate indices
• Data updates and corrections
• Updates
• Corrections
• Mean temperature
• Seasonal climate indices
• Ensembles
• File formats
• Daily
• Indices
• Monthly
• Ensembles and seasons
• Dataset licence
• References

Overview

Environment and Climate Change Canada provides Canadians with climate change information for impact assessment, adaptation planning and developing mitigation policies. There is increasing demand in Canada and around the world for high-resolution projections from global climate models (GCMs).

To meet this need, ECCC’s Climate Research Division (CRD) and the Pacific Climate Impacts Consortium (PCIC) developed statistically downscaled climate scenarios in 2015. These were based on climate model simulations from the Coupled Model Intercomparison Project phase 5 (CMIP5).

These datasets are widely used and are cited in Canada’s Changing Climate Report (2019).^Reference1 They are available through ECCC’s Canadian Climate Data and Scenarios website and PCIC’s data portal.

ECCC and PCIC have updated the CMIP5-based downscaled scenarios by creating two new sets based on the Coupled Model Intercomparison Project phase 6 (CMIP6). CMIP6 projections use updated global climate models and new emissions scenarios known as “Shared Socioeconomic Pathways” (SSPs).^Reference9

Statistical downscaling was applied to simulations from 26 CMIP6 GCMs (Table 2),using three emission scenarios: SSP1-2.6, SSP2-4.5, and SSP5-8.5.

The first set of new scenarios uses the same method (BCCAQv2)^{Reference3Reference13} and target dataset (NRCANmet)^Reference8 as the CMIP5 version. The second set uses a new multivariate method (MBCn) and a new blended target dataset (NRCANmet and PNWNAmet).

The dataset includes daily maximum and minimum temperatures and daily precipitation. These are available across Canada on a 10 km grid for historical (1950-2014) and future periods (2015-2100) under each of the three emission scenarios.

In 2024, ECCC produced additional products, including mean temperature and ensemble datasets. The same year, PCIC added a new emission scenario, SSP3-7.0, to the CanDCS-M6 dataset. This scenario is available for a smaller subset of models (see Updates section for details).

Downscaling methodology and data

Each step in the climate modelling and downscaling process involves uncertainty. Users are encouraged to review the basic principles of climate modelling and downscaling before using these datasets.^Reference4

The statistically downscaled datasets were created using output from 26 CMIP6 GCMs. These models are available through the Earth System Grid Federation (ESGF) Data Nodes. The dataset includes 10 initial condition runs from the CanESM5 model, produced by the Canadian Centre for Climate Modelling and Analysis^Reference11, and one run from each of the 25 other CMIP6 models (Table 2).

CanDCS-U6

Each dataset was statistically downscaled using a hybrid downscaling algorithm called the Bias Correction / Constructed Analogues with Quantile mapping, or BCCAQ, It combines the BCCA method (Bias Correction / Constructed Analogues)^Reference7 with quantile mapping.

BCCAQv2 uses Quantile Delta Mapping (QDM)^Reference3, a form of quantile mapping that preserves changes, setting BCCAQv2 apart from the earlier versions. Details are available in Werner and Cannon (2016).^Reference13 The BCCAQv2 method is available in the R package “ClimDown”, which is accessible through the Comprehensive R Archive Network (CRAN) repository.^Reference5

Using the BCCAQv2 method, PCIC downscaled daily minimum and maximum temperature (°C), and precipitation (mm/day) from 26 CMIP6 GCMs.

Downscaling requires a target dataset for calibrating models. A target dataset is based on observational data and is used to train the historical simulations of a model. The goal of calibration is to make the model’s historical simulation resemble the statistical characteristics of the target as closely as possible.

Once a modelled historical simulation is calibrated for the historical reference period, in this case 1951-2010, the BCCAQv2 method is then applied to all future projections. This process is repeated separately for each variable and model and hence is termed a “univariate” downscaling scheme.

The target dataset for the BCCAQv2 method was the NRCANmetv1 dataset. Natural Resources Canada (NRCan) produced this historical Canadian dataset on a ~10 km grid. The Australian National University Splines (ANUSPLIN) software package, which uses thin-plate smoothing splines to interpolate data into a gridded format was used to create the target dataset.^Reference8 Because of this, the dataset may also be known as ANUSPLIN observations.^Reference10

CanDCS-M6

A univariate downscaling method downscales each variable independently. While this approach is simpler, it can omit or poorly represent important events that involve both temperature and precipitation.

Some climate phenomena like drought, wildfires, and snow, depend on both temperature and precipitation together. Accurately simulating these events requires an approach that can capture the relationships between these variables. A multivariate downscaling approach can do this. Although this method is more computationally expensive, it can maintain inter-variable dependencies and allows for analyses of multi-variable phenomena.

One such method is N-dimensional Multivariate Bias Correction, or MBCn.^Reference2 MBCn uses a modified version of quantile delta mapping (QDM). It is both a bias correction and multivariate downscaling method. This version is able to work with multiple variables while preserving the statistical properties between them.

The CanDCS-M6 dataset uses the same variables, models, and grid as the CanDCS-U6 but with a new target dataset. This is because NRCANmetv1 has a known dry bias on Canada’s west coast. To address this, PCIC blended several observational gridded datasets into one new dataset.

The new blended target (hereafter referred to as PCIC-Blend) combines daily values of all three variables from the PCIC meteorology for Northwest North America dataset (PNWNAmet)^Reference12, with daily precipitation from the Adjusted Precipitation Dataset for Canada (NRCANmet Adj)^Reference6, and daily temperature from version 2 of the NRCANmet dataset (NRCANmet v2).^{Reference8 Reference10}

The blending is done by applying a weighted average over a specific region. This region follows a diagonal line along the eastern side Rocky Mountains. West of the line, the PNWNAmet dataset represents western Canada. East of the line, the two NRCANmet datasets represent central and eastern Canada.

The PCIC-Blend dataset improves how precipitation is represented in western Canada. Since this new dataset includes the PNWNAmet, the target not only covers Canada but also an extension south into the western United States down to 41°N. This allows the CanDCS-M6 dataset to cover the entire Columbia River basin, a key watershed for studying hydrologic impacts.

Climate indices

For each downscaled GCM, 31 climate indices were calculated (Table 3). These include 27 Climdex indices, which were developed by the Expert Team on Climate Change Detection and Indices (ETCCDI)^Reference14 and 4 indices that are modified versions of the Climdex ones. Depending on the index, the results are available either annual or monthly.

Data updates and corrections

Updates

In September 2024, several updates were made to the CanDCS6 datasets.

PCIC added SSP3-7.0 to the CanDCS-M6 dataset, which, before this, only included SSPs 1-2.6, 2-4.5, and 5-8.5. The new SSP3-7.0 dataset only includes the first run (r1i1p2f1) from the CanESM5 model, while other CanDCS6 datasets have 10 runs (r1-10). Also, three models from the original 26 were unavailable for SSP3-7.0 and are therefore excluded from the dataset (HadGEM3-GC31-LL, KACE-1-0-G, and KIOST-ESM). However, in October 2025, the KACE-1-0-G model was added to the SSP3-7.0 dataset. All related datasets were updated at that time to include this change.

ECCC expanded the CanDCS6 datasets by adding mean temperature and ensembles. Mean and percentile ensembles are available for all CanDCS6 datasets, including climate indices.

Corrections

During the creation of the CanDCS-M6 dataset, PCIC identified sporadic anomalously high daily maximum temperature values in three of the 26 GCMs. PCIC determined that the cause was errors in the land surface and atmospheric components shared by the models. To prevent these errors from affecting the downscaled data and avoid replacing the values with missing values, PCIC created a method for outlier detection and correction. This method replaces errors with values robustly interpolated from surrounding grid cells.

The three models with errors were the two Met Office Hadley Centre (MOHC) models, HadGEM3-GC31-LL and UKESM1-0-LL, and KACE-1-0-G from the Korean Meteorological Administration (KMA). PCIC re-ran simulations for these models using the CanDCS-U6 method and corrected the original data. The updated CanDCS-U6 datasets replaced the old versions on August 25, 2023.

The Climdex and agroclimatic indices based on the CanDCS-U6 dataset were also recalculated and updated at the same time. Note: 15 agroclimatic indices from the U6 dataset were updated with corrected data for the three models on September 23, 2024, see updates and errata.

PCIC later corrected parts of the CanDCS-M6 precipitation dataset and a small portion of the CanDCS-M6 CanESM5 dataset (specifically, CanESM5, r3i1p2f1, SSP1-2.6, for the years 2060-2100). Both corrections used the same method for detecting and replacing anomalous values. The updated datasets were published on CCDS on September 25, 2024. Unlike earlier corrections to maximum temperature, these updates did not affect the related Climdex indices.

On October 2, 2025, corrupted data was found in some of the CanDCS-U6 agroclimatic indices from the MPI-ESM1-2-HR model under SSP2-4.5. See updates and errata for more details.

Mean temperature

CanDSC6 datasets include maximum and minimum temperature and total precipitation. Using these data, ECCC calculated mean temperature datasets for all models, scenarios, and time periods. Daily datasets were produced with Python (v3.10.4) and the Xarray package (v2022.3.0). Monthly datasets were created using Python along with Climate Data Operators (CDO, v2.0.3, function ‘monmean’) and NetCDF Operators (NCO, v5.0.5). All processing was done in a Bourne Again SHell (Bash) environment.

Seasonal climate indices

Of the 31 Climdex indices produced by PCIC, 12 are available monthly instead of annually. For these indices, ECCC created seasonal and annual versions by calculating seasonal or annual maximums, means, or minimums depending on the index (Table 4). These calculations were done using CDO (v2.0.3) in a Bash environment.

Ensembles

ECCC calculated an ensemble mean and five ensemble percentiles (5^th, 25^th, 50^th, 75^th, 95^th) across the CanDCS6 models. These ensembles are available for all datasets, variables, indices, SSPs, and time periods. To avoid bias towards any single model, ensembles use only one run from each model, including the CanESM5 model (r1i1p2f1).

The ensembles were created using CDO (v2.0.3), Python (v3.10.4), and NCO (v5.0.5) in a Bash environment. CDO’s functions ‘ensmean’ and ‘enspctl’ were used to calculate the ensemble statistics. However, before calculating daily, seasonal, and annual ensembles, extra preprocessing was required.

Seasonal and annual versions of model data were produced with CDO (temperature: 'seasmean', 'yearmean'; precipitation: 'seasum', 'yearsum'). For models using different calendar types, the daily datasets were converted to Proleptic Gregorian using the Xarray package (v2022.3.0) in Python. This calendar was chosen because most models use it. The conversion adds days with missing values where needed, so no values are artificially created by interpolation through this method.

File formats

Daily

The downscaled simulations are organized into separate directories. Each directory contains one simulation for a specific GCM, variable, realization, and shared socioeconomic pathway (SSP). For CanDCS-M6 datasets, the directory name also includes the “MBCn” tag.

For example, CanDCS-U6 daily precipitation data from the first run (r1i1p1f1) of the ACCESS-CM2 model under the SSP1-2.6 scenario is stored in the following directory:

ACCESS-CM2_pr_r1i1p1f1_ssp126{_MBCn}_5_year_files

As the daily files are large (~57 Gb per file), datasets have been split into 5-year subsets. Each 5-year file follows a consistent naming format that identifies the simulation and time period it covers.

For example, the first file in the directory listed above is:

pr_day_BCCAQv2+ANUSPLIN300_ACCESS-CM2_historical+ssp126_r1i1p1f1_gn_19500101-19551231.nc

Components of the filename format include:

• Variable: "pr"
• Timestep: "day"
• Downscaling method and target: "BCCAQv2+ANUSPLIN300" (or “MBCn+PCIC-Blend”)
• GCM: "ACCESS-CM2"
• Past and future pathway: "historical+ssp126"
• Realization: "r1i1p1f1"
• Grid Label: "gn"
• Time interval: "19500101-19551231"

Each directory includes one 6-year file (1950-1955, ~2.3 Gb) and 29 files of 5-years each (~1.9 Gb each). Together, these files make up the full 151-year dataset for that GCM, variable, realization, and socioeconomic pathway.

The grid label shows whether the model datasets use the model’s native grid ("gn") or have been regridded ("gr") by the modelling centre. If a specific GCM is available for more than one grid resolution, this is shown by a number after “gr” (e.g. "gr2").

Indices

Files containing climate indices use the same naming format. For example, the first file in the Climdex directory of ACCESS-CM2 is the annual (ann) CDD index and is named as follows:

ACCESS-CM2_ssp126_r1i1p1f1/cddETCCDI_ann_BCCAQv2+ANUSPLIN300_ACCESS-CM2_historical+ssp126_ r1i1p1f1_1950-2100.nc

Components of the filename format for the climate indices include:

• ETCCDI climate index: "cddETCCDI"
• Time resolution: “ann”
• Downscaling method and target: "BCCAQv2+ANUSPLIN300" (or “MBCn+PCIC-Blend”)
• Time interval: “1950-2100”

Monthly

Monthly GCM files for differ slightly between CanDCS-U6 and CanDCS-M6. The monthly sum of pr for the same GCM and SSP is the following:

pr_monthly_total_BCCAQv2+ANUSPLIN300_ACCESS-CM2_historical+ssp126_r1i1p1f1_1950-2100.nc

ACCESS-CM2/pr_day_MBCn+PCIC-Blend_ACCESS-CM2_historical+ssp126_r1i1p1f1_195001-210012_monthsum.nc

Components of the filename format include:

• Variable : "pr"
• Monthly calculation:
• U6 : “monthly_total” (pr) or “monthly_average” (tasmin, tasmax, and tas)
• M6 : “monthsum” (pr) or “monthmean” (tasmin, tasmax, and tas)
• Downscaling method and target: "BCCAQv2+ANUSPLIN300" (or “MBCn+PCIC-Blend”)
• GCM: "ACCESS-CM2"
• Past and future pathway: "historical+ssp126"
• Realization: "r1i1p1f1"
• Time interval: “195001-210012”

Ensembles and seasons

Ensemble and seasonal datasets use a filename format similar to the original files. Instead of a GCM, the ensemble type is used, while time resolution depends on the dataset. For example, the first file in the seasonal “tas” directory is as follows:

seasonal{monthly}/tas/tas_annual_avg_BCCAQv2+ANUSPLIN300_ensemblemean_historical+ssp126_1950-2100.nc

Components of the filename format for the ensembles include:

• Time resolution: "annual" (or “monthly_avg{sum}” (sum only for pr)) or season: “DJF”, “MAM”, “JJA”, “SON”)
• Operation: “avg” (or “sum” for pr)
• Downscaling method and target: "BCCAQv2+ANUSPLIN300" (or “MBCn+PCIC-Blend”)
• Ensemble type: “ensemblemean” (or percentile: “enspctl5”, “enspctl25”, “enspctl50”, “enspctl75”, “enspctl95”)
• Time interval: “1950-2100”

The first file within the daily “pr” directory for SSP1-2.6 is as follows:

daily/ssp126/pr/pr_day_BCCAQv2+ANUSPLIN300_ensemblemean_historical+ssp126_19500101-19551231.nc

Components of the filename format include:

• Variable : "pr"
• Timestep: "day"
• Downscaling method and target: "BCCAQv2+ANUSPLIN300" (or “MBCn+PCIC-Blend”)
• Ensemble type: “ensemblemean” or percentile (“enspctl5”, “enspctl25”, “enspctl50”, “enspctl75”, “enspctl95”)
• Past and future pathway: "historical+ssp126"
• Realization: "r1i1p1f1"
• Grid Label: "gn"
• Time interval: “195001-210012”

Dataset licence

Open Government Licence - Canada (http://open.canada.ca/en/open-government-licence-canada)

Individual model datasets and all related derived products, including the multi-model ensembles, are subject to the terms of use of the source organization.