Climate models are computational tools used to study the dynamics of the climate system and predict future climate changes. Climate models have developed historically through the growth of mathematical techniques, computational capacity of computers, data of present and past climates, and theories of component climate processes. Climate models have progressed from very simple models for regional weather prediction, to sophisticated general circulation models that simulate interactions of the atmosphere, oceans, land surfaces, ice, and biosphere. Climate models are of great relevance because they examine the effects of increased atmospheric CO2 on the Earth’s average temperature. As such, climate models have been the primary source of data used by scientific communities and government organizations to draw conclusions about global warming.
As developed in the early 1920s, the first proper model of climate-related events was a mathematical technique used for numerical weather prediction. This model treated a local weather area as a grid of cells. Using a set of basic equations, the model could, in principle, calculate how differences in pressure between adjacent cells determined wind speed and direction across the represented weather area. Lewis Fry Richardson, a pioneer of numerical weather prediction, attempted to apply the model to actual weather conditions. Richardson’s attempt was unsuccessful because he was unable to perform calculations faster than the weather occurred.
During the late 1940s and early 1950s, computers were developed that could reliably perform complex calculations much faster than human beings. American meteorologists used computers to build models of weather simulation, one of the earliest of which was run on a computer named ENIAC. This particular model divided North America into a grid of cells. Taking the known weather conditions for each grid, the model calculated how air should move across the cells. When compared to the real weather that emerged, the model proved partially accurate, though the applicability of the model was limited by computational capacity and scientific knowledge of weather and climate processes.
As scientists developed more successful weather models, scientific focus expanded during the 1950s and early 1960s to include the development of simple climate models. In the earliest of these models designers analyzed the effects of geography and topography of mountain ranges on airflow across North America. They also simulated how energy and momentum moved through the atmosphere, and were able to predict wind patterns with some accuracy. Additionally, simple models began to represent equilibrium in the atmosphere by incorporating calculations for balancing incoming solar radiation with outgoing radiation reflected from the Earth. By the mid-1960s, these developments gave some degree of credibility to climate modeling; models could now simulate coarse processes of the Earth’s atmosphere.
General circulation models
In the late 1960s and early 1970s, computers advanced to the point of performing complex calculations in short timescales. Better observational data and measurements were made available through satellites and ground measurements. These conditions facilitated the emergence of three-dimensional atmospheric general circulation models, which represented climate as a comprehensive system. As they developed, general circulation models treated the atmosphere as layers and incorporated processes of convection, evaporation, and rainfall. The models could simulate the transfer of radiation vertically through the atmosphere, the reflectivity of sunlight from snow and ice, and basic seasonal changes. During this period, oceans were represented in models and coupled with the atmosphere, though these early models treated the ocean as a slab, without any unique dynamics of its own. General circulation models permitted analysis of how the movement of radiation through the atmosphere was affected by water vapor and CO2.
Throughout the 1970s, Syukuro Manabe built and enhanced a climate model to analyze the relationship between CO2 changes and climate change, a concept known as climate sensitivity. Specifically, the model investigated how a doubling of atmospheric CO2 concentration would change temperature. The first estimates indicated that temperature would increase by approximately 3.6 degrees F (2 degrees C). In this way, models raised awareness among the scientific community that rising CO2 in the atmosphere could lead to increases in the Earth’s temperature, a phenomenon known as greenhouse warming.
By the 1980s, atmospheric general circulation models permitted analysis of climate component processes occurring on smaller spatial and temporal scales. Also, models of this period coupled the atmosphere and oceans in a complex way by simulating the exchange of heat between both components. Coupled models could simulate changes in sea ice, as well as the emergence of deserts and rainfall areas. As the use of coupled models advanced, research into atmospheric changes in CO2 revealed serious consequences for the processes driving the overturning of ocean waters, which could potentially cease with higher levels of CO2.
Models and climate change analysis
Models provided information that ultimately increased public concern about the Earth’s climate. As a result, the validity of models themselves became an issue of increasing importance. By the 1990s, projects were initiated in which models were systematically compared to one another, in relation to a growing and uniformed body of climate data. Such projects provided the science of climate modeling with a form of quality control and a standard of reproducibility of results. Using data about conditions during the most recent ice age (which were gained primarily from ice-core measurements), one way of testing the validity of models was to see how well they could simulate the climatic and oceanic conditions characteristic of glacial periods. An additional test of validity involved using models to predict consequences of unique, real-time events, such as climatic responses to volcanic eruptions. Models were successful in these various tests. Models gained further credibility as they served as the primary source of data for the newly-formed Intergovernmental Panel on Climate Change (IPCC). In the mid-1990s, the IPCC used climate model data to draw its conclusion that a human influence on climate had been detected.
Models that have developed over the past several years are extremely complex. Relative to the simple models that emerged in the 1950s, current models can carry out simulations in much shorter time periods and represent many component processes simultaneously, including the atmosphere, oceans, glaciers and ice sheets, land surfaces, and biological and chemical activities linked with human economic life. Models now inform understanding of a range of possible future climatic changes and their impacts. Current models that simulate the present climate and account for CO2 show a severe risk of future global warming. Based on these data, the broadly accepted view among scientists, policymakers, and the public is that rising CO2 levels are warming the Earth.
The success of complex models has not eliminated the need for simple models. The latter models still provide valuable information about independent processes and climatic changes occurring on small spatial and temporal scales. Thus, they serve to develop basic theoretical understanding that enables complex models to simulate broader interactions between the atmosphere, oceans, biosphere, and other component processes. Along with models of intermediate complexity, such as those used to study long time-series corresponding to glacial processes, simple models are regarded as part of a hierarchy of models upon which complex models are based.
In the future, models will continue to generate highly-complex, detailed three-dimensional representations of climate dynamics. Nevertheless, challenges remain that prevent models from providing a perfect simulation of real climate. Scientific understanding of many of the key process that control climate sensitivity (such as clouds, vegetation, and ocean convection) is incomplete. Consequently, these processes cannot be represented in detail in climate models. Additionally, the predictive capabilities of climate models are linked to their performance in reproducing the historical record, which is largely limited to geological data and relatively recent global temperature observations. Processes could exist on a longer temporal scale that even the highly-complex models of today cannot take into consideration. For example, the presence of non-linear processes could potentially change the behavior of climate abruptly. Future progress of climate modeling will depend on the continued commitment to developing improved mathematical and computational techniques, more data, and better theoretical understanding of climate dynamics.