Welcome to the first post of our new series “Ask a Climate Expert,” where we address real-life questions that were sent to the Canadian Centre for Climate Services’ Support Desk. We’ll be sharing our answers here with added details, in an easy-to-understand format that paints a compelling picture for everyone, not just climate scientists.
Today’s post is all about understanding baseline periods.
When working with climate data, you may have encountered the term “baseline periods”, sometimes also called “reference periods”. We recently received a thoughtful inquiry on the topic, pointing out the variety of baseline periods in use, such as 1971-2000 and 1981-2010, and questioning which one to choose for a project.
So, let’s get straight into it. What are baseline periods and why are they important?
Simply put, a baseline period is a span of years that scientists use as a reference to compare climate change over time. This time frame is usually 30 years because this is considered long enough to account for natural climate variability, yet short enough to be relevant to societies and ecosystems. Moreover, the World Meteorological Organization (WMO) has designated 30 years as the standard baseline length.
Baseline periods are used with both observed climate data and data from climate models. Here, we’re going to focus on their use with climate model data. Just as a reminder, the output from a climate model is know as a simulation, which includes a component over the historical period (modelled historical data) and a component which extends out into the future (the future simulation or projection). The historical component of the model experiment uses observed atmospheric concentrations of the various greenhouse gases (GHGs), while the in future component makes use of an emissions scenario to supply the future atmospheric composition.
The selection of your baseline period can depend on various factors. One key consideration is the source of your climate projections – whether you’re using CMIP5 or CMIP6.
ClimateData.ca provides projections from two phases of the Coupled Model Intercomparison Project, CMIP5 and, most recently, CMIP6.
One of the differences between these two climate model ensembles is the duration of the historical and future components of the simulations. In the earlier CMIP5 models, the historical component of the simulation ends in 2005, with the future simulations beginning in 2006 and continuing until the end of the century. In the more recent CMIP6 ensemble, the historical simulation ends in 2014, the future simulations begin in 2015 and, as CMIP5, continue out to 2100. This means that the versions of climate models that you choose to work with (i.e., whether you are using CMIP5 or CMIP6 models) will dictate what historical modelled data are available for you to use as a baseline.
It is best practice to use a baseline period from the historical component of the simulation and not one that spans both the historical and future components. While there is not a great difference among the emissions scenarios in the early years of the future simulations, it is best to avoid including years which are part of the future simulation.
For instance, if you’re using CMIP5, the most recent 30-year baseline that would include only modeled historical data is 1976-2005. The period 1971-2000 is a popular option.
With CMIP6, on the other hand, you could choose a more recent baseline such as 1985-2014, or stick to complete decades with a baseline of 1981-2010. You could also use the same 1971-2000 baseline as with CMIP5.
You might choose a more recent baseline for its relevancy, such as opting for 1981-2010 over 1971-2000 if you’re using CMIP6. However, if you aim to compare your work with previous studies, it would be best to use the same baseline as the earlier work.
Moreover, it’s crucial to note that ClimateData.ca uses a baseline period of 1971-2000 in its delta calculations. The delta, or change, is the difference between the 30-year averages of a future time period (e.g., 2041-2070) and the reference period (or baseline) for a given climate variable. Using this period as a baseline means you could obtain the delta values directly from the website, and it allows you to see the differences in climate projections between CMIP5 and CMIP6 for the same baseline period.
While modelled historical data is ideal for comparing future projections with simulated past conditions, historical observation data is invaluable for linking to past impacts. This will give you a fuller picture and help you examine the relationships between climate data and climate impacts experienced (e.g., looking at what impacts happened during very hot days in heatwaves).
In conclusion, the selection of your baseline period should be carefully documented and clearly explained. The choice depends on your project’s specific goals, the climate model ensemble you’re using, and the other data you’re planning to incorporate in your work.
Stay tuned for more posts in our “Ask a Climate Expert” series, where we’ll continue to break down complex climate concepts into digestible information. Until then, keep the questions coming and happy data hunting!