CMIP6 statistically downscaled agroclimatic indices technical documentation

On this page

• Overview
• Data and processing
• CanDCS-U6
• CanDCS-M6
• Indices
• Global climate model vs. downscaled data
• Use limitation
• Contact information
• References

Overview

The Canadian Climate Data and Scenarios (CCDS) website provides climate indices at high spatial resolution. These indices are derived using statistical downscaling and based on projections from 26 global climate models (GCMs). The models for the dataset come from phase 6 of the Coupled Model Intercomparison Project (CMIP6).

EEnvironment and Climate Change Canada’s (ECCC) CMIP6 statistically downscaled agroclimatic indices are an updated version of the CMIP5 dataset. They use downscaled projections (CanDCS-U6) from the Pacific Climate Impacts Consortium (PCIC). The indices span a historical time period (1951-2014) and three future emissions scenarios called “Shared Socioeconomic Pathways” (SSPs), SSP1-2.6, SSP2-4.5, and SSP5-8.5 (2015-2100). The data are available on a 10 by 10 km grid.

The CanDCS-M6 version of the 49 agroclimatic indices, produced by PCIC, is now available. This includes SSP3-7.0 indices for most GCMs. PCIC also introduced 18 new indices to the original 49 for a total of 67 indices. These are available for both CanDCS-U6 and CanDCS-M6. See the definition table for a list of indices.

For more details on the methods, please see Li et al. (2018).^Reference1

Data and processing

CanDCS-U6

Daily minimum and maximum temperature, along with precipitation data from GCMs, were statistically downscaled using the BCCAQv2 approach. This stands for Bias Correction / Constructed Analogues with Quantile mapping version 2. A total of 26 downscaled GCM outputs were used to derive the indices.

BCCAQ is a hybrid downscaling algorithm that combines downscaling by BCCA (Bias Correction / Constructed Analogues)^Reference2 with quantile mapping. More information is available in Werner and Cannon (2016).^Reference3 BCCAQv2 differs from earlier versions by using Quantile Delta Mapping (QDM)^Reference4, a method that preserves changes over time.

For more details on the CanDCS-U6 and BCCAQv2 methods, please see the technical notes.

Note: The BCCAQv2 method requires each historical simulation to be paired with each future projection (SSP). Since each dataset is downscaled separately, this process produces small differences in the historical data for each SSP. However, these differences are usually minor.

Downscaling variables independently can sometimes cause minor physical inconsistencies. For example, the minimum temperature may exceed the maximum temperature on certain days. However, studies show that correcting these cases (by swapping values) usually has a negligible effect on the resulting climate indices.^Reference1

CanDCS-M6

A univariate downscaling method downscales each climate measure on its own without assessing any combined effects. While this approach is simpler, it can omit or poorly represent important events that involve both temperature and precipitation.

In contrast, a multivariate downscaling approach handles multiple climate measures together. Although this method is more computationally expensive, it can maintain relationships between climate measures and allows analysis of events that involve multiple variables.

One such method is, N-dimensional Multi variate Bias Correction, or MBCn.^Reference5 MBCn uses a modified version of quantile delta mapping (QDM). It is both a bias correction and downscaling method. This version is able to work with multiple climate measures while preserving the statistical properties between them.

For more details on MBCn and CanDCS-M6, please see the technical notes.

Indices

There are several sets of climate indices available. A well-known example is the Expert Team on Climate Change Detection and Indices (ETCCDI). The ETCCDI indices were developed to study past and future changes in climate extremes.^{Reference6Reference7} However, they may not always show local or regional climate impacts.^Reference1

To support a wide range of users in Canada, CCDS offers custom set of indices. ECCC developed these indices with input from Canada’s climate adaptation community. A full list of these indices along with descriptions and equations is available in the definition list.

Global climate model vs. downscaled data

Future projections from statistically downscaled data are not always more reliable than those from the base climate models. In many cases, especially for indices based on fixed thresholds, downscaled data often vary less because of the removal of model biases. However, this is not true for all indices.

Downscaling GCMs to fine spatial scale needed for impact assessment adds detail and captures more temporal variability, making the data closer to real observations. Since these changes depend on the GCM, the spread of results from downscaled data can be wider than those from the original model. However, this wider spread does not mean downscaling adds more uncertainty. Instead, it reflects the natural high variability found at finer scales.

Use limitation

Open Government Licence - Canada

Individual model datasets and all related derived products are subject to the terms of use of the source organization.

Contact information

Email: f.ccds.info-info.dscc.f@ec.gc.ca