TABLE OF CONTENTS MBIS: A MID-TERM PROGRESS REPORT EXPEDITION STUDYING ARCTIC OCEAN AND GLOBAL CLIMATE CHANGE HUMAN DIMENSIONS OF GLOBAL CHANGE-ARCTIC RESEARCH CGCP NOW CMOS GLOBAL CHANGE FORUM IGBP UPDATE IPCC-SECOND ASSESSMENT REPORT IDNDR-THE YOKOHAMA STRATEGY OPINION-IGBP REPORT NO. 28: WORK PLAN 1994-1998 UVB IMPACTS CONFERENCE UNIQUE WEATHER PROJECT   CCP INFO MODELLING THE GLOBAL CLIMATE SYSTEM WEATHER AND CLIMATE: INFORMATION FOR AGRICULTURE JAMES BRUCE WINS 1994 IMO PRIZE

MODELLING THE GLOBAL CLIMATE SYSTEM Henry Hengeveld Atmospheric Environment Service   Introduction Many aspects of human economy and everyday life, particularly in resources sectors such as agriculture, forestry, fisheries and energy, are highly sensitive to extreme weather events and other variations of climate. Hence, understanding the way climate functions and responds to change has become vitally important. In recent years this importance has been underscored by the realization that global warming, induced by an enhancement of the earth's natural greenhouse effect, may change global and regional climates to new regimes and extremes beyond anything yet observed in human history. One of the central features of the global research effort to understand the climate system has been the use of numerical models--and, in particular, powerful computer-based representations of the global climate system known as general circulation models (GCMs). These tools have emerged internationally as the only practical and timely manner of investigating the behaviour of the complete climate system and studying its response to the forces that effect it. Anatomy of a GCM The underlying concept of numerical modelling is fairly simple and straightforward. Climatic processes are based on fundamental physical laws that can be expressed mathematically. If we write the equations that describe a certain climatic process, we, in effect, have created a model of it. Change certain values in the equations, and the model will tell us how that process will respond to a different set of conditions. By adding more equations to describe more processes, we can eventually create a model of the climate system as a whole. GCMs are powerful climate models that attempt to replicate the structure of the global climate system and the operation of its major processes and feedbacks in a sophisticated, three-dimensional and time-dependent manner. They simulate the passage of solar and heat energy throughout the climate system. This energy flow, and its effect on various elements of the climate system, such as temperature, precipitation, soil moisture, and other important climatic variables, are calculated in a series of time steps and at a large number of regularly spaced locations on the earth's surface and within the atmosphere. For each time step and spatial point, the system is allowed to change, transfer energy, momentum, and moisture, and respond to feedbacks until a stable pattern emerges. The more points there are, the higher the resolution of the model and the finer the detail it can simulate. Many climatic processes, such as clouds, thunderstorms, precipitation, evaporation, and soil moisture, generally take place on scales considerably smaller than the resolution capabilities of GCMs. These are known as sub-grid-scale phenomena. Since the GCM model cannot handle these directly, they must either be given fixed values or estimated through a process, known as parameterization, which links them physically and statistically to the larger scale variables, such as air pressure, temperature, and humidity, that the model can resolve. The average cloud characteristics for a given area, for example, can be parameterized in terms of the temperature and atmospheric moisture at that location. When a certain relative humidity is reached, the model forms clouds. The results can be tested against climate records to determine how well the parameterization works, and whether additional improvements are needed. In spite of its sophistication, a GCM is still only an approximation of reality. Even the most powerful supercomputers available today cannot handle all the detail needed to give a complete description of the climate system. Nor do we fully understand all of the processes that affect climate. Therefore, the model must simplify or estimate some features of the system and simply ignore others that are not considered important enough to significantly affect the outcome of the model's calculations. Consequently, the world as seen by a GCM bears a strong resemblance to a three-dimensional map made of Lego pieces. All of the major features are recognizable, but much of the fine detail is missing. The objective, as the models evolve, is to fill in more of the missing detail and make the models more realistic. First generation GCMs, like most of those used during the 1980s, were quite limited in the amount of detail they could represent. A second generation of GCMs emerged in the late 1980s, with higher resolution, more detailed representation of features and processes, and more sophisticated parameterizations. One of these is the Atmospheric Environment Service's GCM2, which incorporates virtually all of the features considered state-of-the-art for a second-generation model. It has a relatively high horizontal resolution of about 600 km and simulates horizontal heat transport in oceans that provides a better representation of the oceanic heat distribution and its effects on regional climates. Sea ice, cloud properties, land surface processes and other climatic elements are all also simulated more realistically. What Do the Models Tell Us? The most common application of GCMs is to determine the sensitivity of the climate system to a change in one of its key elements. Strictly speaking, the result is not a prediction of climate change, though it does have some predictive value. Instead, it is an answer to a "what if" question--in the case of global warming, what would be the response of the climate system if the concentration of greenhouse gases in the atmosphere increased significantly? The classic experiment for testing the climate's sensitivity to higher greenhouse gas concentrations is known as a 2 x CO2 or doubled carbon dioxide equilibrium response experiment. It is, in effect, two experiments in one. It begins with a present climate simulation, which not only serves as a reference for comparing the results of other experiments, such as a 2 x CO2 experiment, but also as a control for verifying and fine-tuning the model's performance. The better the approximation of the observed properties and behaviour of the real climate system, the greater the confidence in its results when used for climate change experiments. The model is then reprogrammed, with the carbon dioxide concentration set to twice its present level (2 x CO2), and run again until it reaches a new equilibrium. The difference between the two sets of results is the model's projection of the climate's equilibrium sensitivity to a doubling of carbon dioxide. Even with today's most powerful supercomputers, capable of performing 2 billion operations or more a second, a typical climate change experiment can consume more than 1000 hours of computer time. In general, GCMs reproduce the large-scale features of the present global climate--such as the north-south distribution of pressure, temperature, wind, and precipitation--with considerable fidelity. However, on a regional or sub-continental scale, they are less reliable. This is especially true for precipitation and precipitation-dependent measures such as soil moisture and snow accumulation. These results, together with the broad agreement between various types of GCM models on the general nature and direction of change projected under a 2 x CO2 experiment, give us reason to believe that the large-scale pattern of such change is substantially correct. The world is indeed almost certain to become warmer. What we don't know with sufficient confidence, however, is how great the warming will be or how it will affect temperature, precipitation, evaporation, soil moisture and other important climatic features on a regional scale. Nor do we have enough certainty about when these changes will take place. Reducing the Uncertainties A number of international modelling groups, including Canada's, are now at work on a third-generation GCM. Canada's new model will build on the successful foundation provided by GCM2, incorporating improvements to many of its present features while adding some important new capabilities. These include improved treatment of clouds and land surface characteristics and the addition of a detailed treatment of the chemical and radiative processes in the middle atmosphere. The most significant feature of the third-generation model, however, will be the inclusion of a three-dimensional, fully circulating, interacting ocean that can help the model represent the important transfer of heat between the ocean's surface layer and the ocean depths. This will help the model simulate the time delay and some of the critical feedbacks of the ocean response to atmospheric change. Adding this feature will make it possible to conduct a much more realistic type of experiment, known as a transient response experiment, in which the climate changes continuously in response to gradually increasing levels of greenhouse gases, rather than to a sudden leap in concentration as in equilibrium response simulations. To mobilize the necessary expertise required to undertake this enormous task, Environment Canada is developing a Climate Research Network in collaboration with partners in the universities, the private sector, and other federal and provincial government departments. The network will consist of a series of collaborative research groups located across the country, with each group focusing on a particular area of scientific uncertainty. It will have access to the new supercomputer facilities acquired to handle the increased computing power and speed required to run the new model, and a high-speed data link will allow researchers from every region to work together. As GCM capabilities are enhanced, especially for regional simulations, they will find other important applications as well. Reliable climate forecasts that identify the principal climatic trends of the coming season are a realistic possibility. They could also warn of major anomalies, such as those associated with El Ni¤os or volcanic eruptions, and could have considerable value for many sectors of society. But for the moment the pressing demand for further development of GCM expertise will continue to be the need for more reliable information for policy-makers dealing with the challenge of global warming. (The preceding article is extracted from a new Climate Change Digest special report (CCD 94-01), published under the same title. Copies of this report can be obtained, free of charge, from the: Climate Products and Publication Division of the Climate and Atmospheric Research Directorate, Atmospheric Environment Service 4905 Dufferin St. Downsview, Ontario Ontario, M3H 5T4 tel: (416) 739-4431 fax: (416) 739-4380).