Models Put Spin on Future Climate

Once only the territory of prophets and fortune tellers, forecasting the future has become more a matter of science than soothsaying over the past half century. Today, supercomputers capable of performing billions of arithmetic operations per second make it possible to predict the behaviour of many natural systems using mathematical models of the processes that make them tick.

Such models are particularly important when it comes to studying the impact of human activities on the environment, because they allow scientists to plug-in changes in certain variables and then see the possible consequences. Over the past four decades, results from increasingly complex models of our global climate system have raised awareness of one of the most serious environmental issues facing the world today: the dramatic warming effects of increases in carbon dioxide, methane and other heat-trapping "greenhouse gases" on the earth's atmosphere.

Using various scenarios for future emissions of air pollutants, climate modellers have looked ahead 100 years and glimpsed a world in which global temperatures may be nearly 4.5°C higher on average than they were in 1985. Warming is expected to be even greater in Canada, and some parts of the Arctic could see average increases three times that magnitude. These warmer temperatures could have devastating implications—including a drastic reduction in sea ice and snow cover, changes to our water cycle and supply, and the enhanced survival of pests. Ecosystems and animals, such as polar bears, that can't adapt to their new surroundings could disappear forever.

Just how accurate is this picture of the future? Understanding how well climate models work means first understanding how they work. Simply put, climate models are very large computer programs that simulate the functioning of our global climate system in three spatial dimensions and in time. Climate models are based on the laws of physics, which govern the ways in which matter and energy interact. Woven into this framework are equations describing the different processes within the climate system, and how they respond to internal and external changes.

While early climate models represented only atmospheric processes, today's "coupled" models recognize that our climate system involves a myriad of complex interactions connecting the atmosphere, oceans, land surfaces and polar ice masses. Scientists simulate this system by linking individual models of each of these different components with the various processes by which they exchange energy and mass.

Climate models divide the air, land and oceans into a three-dimensional grid made up of thousands of interacting cells. As conditions in one cell change, they also have an influence on their neighbours. Computer-simulated versions of our climate have shown surprisingly realistic variability on time scales from hours to centuries. If you could step inside one of these models, you would experience weather and year-to-year changes in climate similar to those in the real world.

Involved in climate modelling since the 1970s, Environment Canada's work is based out of its Canadian Centre for Climate Modelling and Analysis (CCCma) in Victoria, British Columbia. The CCCma is one of a dozen centres around the globe currently developing coupled climate models, and one of only four whose models have been used by the Intergovernmental Panel on Climate Change to assess evidence of human effects on climate.

Environment Canada created its first coupled model in the mid-1990s, and is currently testing a third-generation version. These models run on a mammoth supercomputer at the Department's weather centre in Dorval, Quebec. The system's mind-boggling capacity to perform 128 billion arithmetic operations per second lets the climate model simulate three years of weather in a single day.

Canada's global climate model is made up of two main components: a general circulation model of the atmosphere and one of the ocean. The atmospheric component divides the atmosphere into a three-dimensional grid with 10 vertical layers extending a total of 30 kilometres above the earth's surface. Each cell in the grid has a horizontal resolution or "width" of about 300 kilometres. The atmospheric component simulates day-to-day weather—that is, the movement, temperature, pressure and density of air, clouds, the transfer of radiative energy through the atmosphere, and the hydrological cycle.

Illustration showing how anthropogenic (human-caused) influences, such as emissions of the greenhouse gases carbon dioxide and methane, affect climate by trapping heat in the atmosphere.

The ocean component, which simulates the ocean's circulation and water properties, has 29 vertical layers and a horizontal resolution of 150 kilometres. It reproduces the large-scale features of the ocean's circulation, as well as important water properties such as temperature, density and salinity. The model also has a sea-ice component that allows ice to form and melt as it exchanges heat with the ocean and atmosphere, and to move naturally with the currents and winds. Finally, there is a land-surface component that calculates variations in soil moisture and surface temperature, evaporation, and reflectivity.

Coupling these components is no easy task. The ocean and atmospheric models must first be individually "spun up" to a state representative of the present climate—a process that takes only a few decades of simulated time for the atmospheric component, but several thousand years of simulated time for the ocean component, which evolves much more slowly. Once they have been coupled, inaccuracies in the modelled flow of heat and moisture between the ocean and atmosphere are difficult to avoid, and would eventually cause the simulated climate to drift if not corrected using a "flux adjustment". This adjustment has become smaller with each successive generation of the climate model.

Trends and variations in average global surface temperatures as simulated by the Canadian Global Coupled Model 1. The thin, lower line is the model's control run; the thickest line the model's greenhouse-gas and aerosol runs; and the other line (which ends in the year 2000) shows observed climate trends.

An important test of a climate model is whether it can reproduce the changes in global mean temperature observed during the past 150 years when it is run with the known changes in greenhouse-gas and aerosol concentrations. Although current models pass this test, scientists point out that there is room for improvement. For example, the influences of many surface features, such as mountains and vegetation, can only be estimated because the resolution of the model does not allow it to "see" features smaller than 300 kilometres across.

Two other areas of the model that require improvement are those dealing with clouds and aerosols. Clouds, which can both reflect and trap solar rays, are difficult to represent not only because their basic physics isn't well known, but also because they occur on a small scale and are highly chaotic. Aerosols are tiny particles in the atmosphere that are a product of fossil fuel combustion. Although they have a direct cooling effect on climate by reflecting the sun's energy back into space, they also have an uncertain indirect impact resulting from their effect on the reflectivity and longevity of clouds.

In an effort to address these and other shortcomings, Environment Canada is actively developing new models with collaborators in Canada's universities. A third-generation coupled model is now running in test mode. Its atmospheric component has many improved physical processes and 32 vertical layers extending 50 kilometres above the earth's surface. It also has a multi-layer land-surface model that includes a vegetative canopy, flowing rivers that carry runoff to the ocean and a representation of ice sheet processes that operate in Greenland and Antarctica. While tests on the third-generation model continue, further work is under way to incorporate the carbon cycle—the continuous transfer of carbon back and forth between the atmosphere and living organisms—into the model.

In the meantime, researchers at the Université du Québec à Montréal are refining a regional climate model that is nested in Environment Canada's global model and operates at a finer horizontal scale of 45 kilometres. This regional model uses data from the larger-scale model in much the same way that regional weather forecasts use data from global forecasts. Already tested in a series of simulations of the current and future climate in western Canada, the model will soon be used to create the first high-resolution assessment with a dynamical model of how global climate change will affect the different regions of Canada.

Schematic comparing some of the differences between the structures of the second-generation Canadian Global Coupled Model (left) and the third-generation version of the Model (right).

While there will always be room for improvement, today's models provide us with remarkably realistic forecasts of the future state of our environment. Such forecasts are not only useful in the formulation and implementation of policies to reduce environmentally harmful human activities—they also allow us to develop strategies for adapting to the impacts of droughts, floods and other related risks of climate change.

What does the Canadian Global Coupled Model (CGCM1) predict for the future? Average global surface temperatures increase by about 1.7ºC above 1985 levels by 2050 and nearly 4.5ºC by 2100. Increases over land are higher than over the ocean, with the former warming by 6ºC and the latter by 3.5ºC by 2100. In Canada, cold extremes become less severe and less frequent with time, while extreme maximum temperatures become hotter and more frequent. Average global precipitation increases by about 1 per cent by 2050 and 4.5 per cent by 2100. By 2090, precipitation over most of Canada increases by 10-20 per cent, and damaging precipitation could double in frequency. Combined with warmer temperatures, North America experiences a notable decrease in available soil moisture. Major changes occur in sea ice coverage in the Northern Hemisphere, with the annual mean coverage decreasing by about 40 per cent by 2050 and virtually disappearing by 2100. Average global sea-level rises about 40 centimetres by the last two decades of the 21st century, mainly as the result of the thermal expansion of ocean waters.

  Home |  Air |  Atmospheric Science |  Climate Change |  Environmental Action |  Habitat |  Pollution |  Species at Risk |  Technology |  Water |  Weather |  Wildlife