Climate Forecasts: How may the climate change in a given area?

“Climate” in a location can be most simply described as the average (or central tendencies of) weather over specific periods of time. Climate forecasts are developed by using Global Circulation Models (GCMs), which is a mathematical representation of the physics driving climatic processes. The current GCMs typically link dynamic models of the atmosphere and the ocean so that the interactions between these two aspects of the climate system can be captured. Before they are used to predict future changes, models are carefully calibrated by tuning them until they are accurately able to predict past climate conditions.

Most agree that GCMs are currently the best way to predict climate change and they are constantly being improved. There are however, two complications for decision makers to consider when using GCMs. First, irreducible uncertainties are an inherent element of climate change, regardless of the sophistication of modelling techniques. Some of the issues regarding uncertainty are discussed in the Irreducible Uncertainty section of this site.

The second challenge presented by GCMs that may be of particular relevance for local decision makers dealing with adaptation is the resolution of these models. GCMs are built by creating a grid that covers the entire globe. Predictions are made by resolving the model equations for each square and then linking this to the surrounding squares. The sheer complexity of the models limits the resolution at which predictions can be generated. Currently, most GCMs generate information for grid squares that are about 2° square, which means that in British Columbia, for example, grids are several hundred kilometres square.

Within the climate science community, significant resources have been invested into finding ways to downscale GCM data so that regional predictions can take local variations into account. This work is ongoing, and while specific downscaling has been conducted for some regions, the most commonly available information is directly available from GCMs. This information does indicate the potential general changes within a given region. However, it is unclear how much of an advantage increased resolutions of predictions will be for local decision makers. To some extent, the irreducible uncertainties may mean that the lower resolution GCMs are as useful as higher resolution downscaling in immediate planning activities

Regional Climate Forecasts

Forecasts of relevance to a particular region can be generated in two ways. First, data on the region in question can be extracted directly from a GCM. These data are fairly low resolution (large scale), but show general trends and expectations. Second, climate downscaling can be used to generate more specific forecasts for a particular region. These data will have higher resolution but are much more expensive and difficult to generate. Either way, developing regional scenarios and creating influence diagrams that correlate these scenarios to regional systems can help make climate data more useful for decision making.

Accessing Regionally Relevant Data from a GCM

The Pacific Climate Impacts Consortium (PCIC) has made their Regional Analysis Tool available to the public. This tool has the capacity to present regional data for North America from 10 GCMs, and has global scale data from an additional 15 GCMs. This tool does not use regionally downscaled data, but shows the results from CGMs for the particular region in question.

The PCIC tool allows one to use data from multiple GCMs under several emission scenarios to generate maps and basic statistics describing potential impacts on a range of variables. Variables that can be explored include mean temperature, mean precipitation, soil moisture, wind speed, and many more. By considering these variables, planners can start to think about how regional systems might interact with these predicted changes.

The PCIC Regional Analysis tool can be found at: http://pacificclimate.org/tools/regionalanalysis/

Accessing Tools and Data for Regional Downscaling

Developing regionally downscaled climate predictions is more difficult than accessing regionally relevant GCM data. However, the Canadian Climate Change Scenarios Network provides both some of the basic data required and two downscaling tools. These two tools are the Statistical Downscaling Model (SDSM), and a weather generator (LARS-WG). This centre also has maps and data available from Canada’s regional climate model (CRCM).

In addition to the tools for downscaling climate models, results of downscaling predictions for particular regions are also available. For example, some work has been done linking GCM predictions to the effects produced by altitude that can only be captured by more specific downscaling.

Both of the downscaling tools provided through this Canadian Climate Change Scenarios Network can be used to create regional scenarios and are available at:

http://www.ccsn.ca/The_Network/The_Network-e.html

Maps generated through partial downscaling conducted by the BC Royal Museum in conjunction with the Pacific Climate Impact Consortium are available at:

http://www.pacificclimate.org/resources/climateimpacts/rbcmuseum

Downscaling climate data is a strategy for generating locally relevant data from GCMs. For details on how to do so effectively, review the section, How does Downscaling Work?

Creating Regional Scenarios

Two factors are particularly important in shaping regional climate forecasts:

1. The GCM used to generate the forecast; and
2. Assumptions about GHG emission generation over time.

The number of relevant GCM forecasts available through the PCIC tool depends on the specific variables (e.g., temperature, precipitation, wind speed), region, and time of interest. Similarly, there are several possible GHG emission scenarios for which data can be accessed. When combined, these two factors create a wide range of possible results for the variables in question.

In order to make the diversity of results useable for decision makers, it makes sense to create a smaller set of regional scenarios. For example, based on differences in GCM and emissions assumptions a set of ‘high, medium and low’ scenarios can be built to show the range of impacts on the selected variables of interest. This can help to identify regional vulnerabilities.

A recent example of the creation of a regional scenario of this type is found in work prepared for British Columbia’s biodiversity plan - http://www.biodiversitybc.org/assets/Default/BBC Major Impact Climate Change.pdf.

Using the PCIC regional analysis tool, the full range of GCM predictions across all scenarios were collected. This collection gives the full range of predictions for the province. To make this information more useful, only three combinations of emission scenario and GCM were selected. These three combinations represented a “high” range for climate change, a “medium” and a “low” level. Decision makers could start to predict what the broad range of possible impacts might be using these combinations for seasonal and annual predictions for temperature and precipitation.

Why Would You Want to Explore Multiple GCMs?

Climate modelling is inherently uncertain but this does not mean that forecasts do not have value. One way to make decisions despite this uncertainty is to consider the range of possible climate outcomes instead of relying on single forecasts.

Because each GCM incorporates slightly different assumptions about how the climate works, each generates different results. Decision makers can make more resilient decisions by incorporating a range of these results in their considerations.

The variation in GCM outcomes can also be combined with variation in greenhouse gas emission scenarios.