Scenarios and climate models

Scenarios and climate models are the building blocks for future climate projections.

Emissions scenarios

Human activity is causing climate change. However, we don’t know exactly how humans will behave in the future. Nor can we know how emissions of greenhouse gases will change. Emissions scenarios are a way to help us understand what the future could look like. These scenarios provide a range of possible futures, based on a range of future emissions.

A set of scenarios referred to as Representative Concentration Pathways (RCPs) are in common use to study future climate change. RCPs are designed to provide plausible future scenarios of human emissions patterns. These include consideration of future greenhouse gas emissions, deforestation, population growth and many other factors.

Based on best practices in the global science community, the Government of Canada usually presents 3 RCPs (Figure 1):

RCP8.5: high global emission scenario. This scenario indicates global average warming levels of 3.2 to 5.4°C by 2090.
RCP4.5: medium global emission scenario, includes measures to limit (mitigate) climate change. This scenario indicates global average warming levels of 1.7 to 3.2°C by 2090.
RCP2.6: low emission global scenario, requires strong mitigation actions. This scenario indicates global average warming levels of 0.9 to 2.3°C by 2090.

Other futures are also possible, but limiting consideration to a few RCPs is easier for research and communication. The pathway that unfolds in reality will depend on societies’ choices.

Figure 1: Change in global average temperature relative to the 1986-2005 reference period

This figure shows changes in global average temperature, relative to the 1986 to 2005 reference period, simulated by 29 global climate models from the Coupled Model Intercomparison Project, Phase 5 (CMIP5).

Long description

This figure shows both historical simulations and future projections in global average annual temperature from 1900 to 2100 under three emissions scenarios. Historical simulations are for the period from 1900 to 2005. Future projections are for the period from 2006 to 2100 based on three global emission scenarios: low (RCP2.6), moderate (RCP4.5), and high (RCP8.5). Data are presented as a shaded band surrounding a thick line, corresponding to the range of model projections and the median, respectively. RCP2.6 has the smallest projected increase in global average temperature reaching approximately 1 degree change by 2100 with the range between approximately 2 and 0 degrees. In RCP4.5 the median reaches approximately a 1.9 degree change by 2100 with the range between approximately 1 and 3 degrees. RCP8.5 shows the largest increase in global average temperatures with the median reaching approximately 4 degrees by 2100 and the range between approximately 2.9 and 5.6 degrees.

Climate models and projections

The climate is affected by many elements, including ocean temperatures, clouds, rainfall and vegetation growth. Each of these processes can be simulated in a climate model. These models are so complex it can take weeks to run one simulation, even with supercomputers. To decrease computing time as much as possible, climate models divide the Earth up into large grid cells. For global climate models (GCMs) that cover the globe, grid cells are often larger than 100 kilometres (km).

For impact, vulnerability or adaptation studies, you may need data from grid cells smaller than 100 km. For example, precipitation extremes often occur on much smaller scales than 100 km. Also, local temperatures can be affected by local landscape features. One way to get finer resolution data (smaller grid cells) is from dynamic or statistical downscaling.

Dynamically downscaled models are also known as regional climate models (RCMs). RCMs simulate the climate of a smaller region, relying on information provided by GCMs. RCMs’ grid cells are usually from 10 to 50 km in size. RCMs use the laws of physics to simulate the local climate.

Statistically downscaled models use statistical relationships between local climate variables (such as precipitation) and large-scale variables (such as atmospheric pressure). The relationship is then applied to projections from GCMs to simulate local climate.

Climate models project future climate conditions based on the assumptions in the RCPs (Representative Concentration Pathways). In other words, a climate projection shows how certain elements of climate, such as the average temperature in a region, could change based on one RCP. Although climate models are based on the laws of physics, different climate models can use different methods to simulate these laws when simulating the climate. So, even for the same RCP, projections from different climate models can differ.

Managing uncertainty in climate projections

We can’t say for certain how the climate will change in the future. This is because:

we can’t predict the exact amount of greenhouse gases future human activity will produce
we can’t perfectly model the Earth’s climate system

Use of different scenarios or RCPs help deal with the first issue. For the second issue, multiple climate models, each constructed somewhat differently, are used. There is no one best climate model. Plus, some models are better at capturing different aspects of the climate than others. Dealing with all these results can be daunting. Fortunately, we can account for the range of model results in a simpler way.

The results from many climate models are grouped to create a “multi-model ensemble.” Then, percentiles (from statistics) are used to summarize the range of results from that ensemble. A percentile is the maximum value of a percentage of all results. To summarize the range of results, the 25th and 75th percentiles are commonly used, along with the median (50th percentile). For example, the 25th percentile for temperature increase means that 25% of individual model results show the same or less warming.

Learn about how these climate information concepts are translated into practice in decision-making.