Some of the strongest evidence to support the view that human activities are causing the Earth to warm comes from numerical experiments performed with state-of-the-art global climate models. These models encapsulate the current understanding of the physical processes involved in the climate system, the interactions, and the performance of the system as a whole. They have been extensively tested and evaluated using observations. Today's best climate models are now able to reproduce the climate of the past century, and simulations of the evolution of global surface temperature over the past millennium are consistent with paleoclimate reconstructions.
As a result, climate modelers are able to test the role of various forcings in producing the observed changes in global temperature. Forcings imposed on the climate system can be natural in origin, such as changes in solar luminosity or volcanic eruptions, or human-induced, such as increases in aerosol and greenhouse gas concentrations in the atmosphere.
Climate model simulations that account for such changes in forcings have now reliably shown that global surface warming of recent decades is a response to the increased concentrations of greenhouse gases and sulfate aerosols in the atmosphere. An example, from a climate model simulation performed at the National Center for Atmospheric Research (NCAR), is provided in Figure 1. When the model is integrated forward in time over the 20th Century with only information on imposed natural forcings, there is no discernible trend in global surface temperatures over the last several decades (blue line). When changes in greenhouse gas and aerosol concentrations are added to these natural forcings, however, the model not only simulates an increase in global surface temperature (red line), but it almost exactly reproduces the observed rate of change (black line). Numerous simulations for each case are run, and the solid lines represent the mean while the shaded regions indicate the "spread" about the mean. This spread reflects intrinsic natural climate variations arising from purely internal atmospheric processes as well as from interactions among the different components of the climate system, such as those between the atmosphere and oceans or the atmosphere and land.
Such results, which have also been produced by several other independent modeling groups, increase our confidence in the observational record and our understanding of how global mean temperature has changed. They also indicate the time histories of the important forcings are reasonably known, and the climate processes being simulated in models are adequate enough to make the models very valuable tools for investigating the causes and processes of past climate variations as well future climate change.
Surface temperature warming has occurred in all seasons and over much of the globe, but not uniformly. Climate models used to project future climate indicate that the largest temperature increases will occur over land relative to oceans, with the greatest warming at high latitudes of the Northern Hemisphere during the winter and spring seasons - much like the pattern we have observed over recent decades.
Modest warming will have both positive and negative impacts. A modest increase in global temperatures could increase agricultural productivity in some areas by, for instance, lengthening the growing season. But in high latitude regions, where the warming is expected to be greatest, there is already strong evidence to suggest the current warming is having strong negative impacts, such as severe coastal erosion due to retreating sea ice, increasing sea level, and thawing of coastal permafrost. The thawing of tundra is having negative impacts on buildings, roads, and industry. Higher global sea levels associated with warmer ocean temperatures mean that storm surges associated with hurricanes will be more destructive. Moreover, the rate of future warming as projected exceeds anything seen in nature in the past 10,000 years.
There is an abundance of scientific evidence that shows major and widespread climate changes have occurred with startling speed. For example, roughly half of the warming of the North Atlantic Ocean since the last ice age was achieved in only a decade, and this warming was accompanied by significant changes in climate across most of the globe. Research over the past decade has shown that these abrupt - or nonlinear - climate changes have been especially common when the climate system was being forced to change most rapidly. Thus, the rate of buildup of carbon dioxide in the atmosphere may increase the possibility of large, abrupt and unwelcome regional or global climate events.
The mechanisms of past abrupt climate changes are not yet fully understood, and climate models typically underestimate the size, speed and extent of those changes. Hence, future abrupt changes cannot be predicted with confidence. Yet, because of greenhouse warming and other human alterations of the earth system, and the long lifetime of carbon dioxide in the atmosphere, certain thresholds are likely to be crossed and we will not know we have crossed them until it is too late to alter the outcome.
The best climate models encapsulate the current understanding of the physical processes involved in the climate system, the interactions, and the performance of the system as a whole. They have been extensively tested and evaluated using observations. They are exceedingly useful tools for carrying out numerical climate experiments, but they are not perfect, and some models are better than others. Uncertainties arise from shortcomings in our understanding of climate processes operating in the atmosphere, ocean, land and cryosphere, and how to best represent those processes in models. Yet, in spite of these uncertainties, today's best climate models are now able to reproduce the climate of the past century, and simulations of the evolution of global surface temperature over the past millennium are consistent with paleoclimate reconstructions. This gives increased confidence in future projections. The shortcomings in our understanding of the processes involved in climate and how they are depicted in models arise from inadequate observations and theoretical underpinnings associated with the incredible complexity of dealing with scales from molecules and cloud droplets to the planetary-scale atmospheric circulation. These issues are addressed in several steps:
- Individual climate processes are dealt with as best as is possible given the understanding and computational limitations.
- The processes are assembled in models and then the model components are tested with strong constraints. The components include modules of the atmosphere, the oceans, the land and sea ice, and the land surface. These modules are coupled together to mimic the real world.
- The climate system model as a whole is then integrated in an unconstrained mode and thoroughly tested against observations.
One strong test is to simulate the annual cycle of seasonal variations (the changes in climate from winter to summer). Another is to simulate observed variability from one year to the next. Yet another is to simulate past climate, even going back in time thousands or millions of years tested against records from ice cores, tree rings, and other "proxy" data.
As our knowledge of the different components of the climate system and their interactions increases, so does the complexity of today's climate models. Also, many of the most pressing scientific questions regarding the climate system and its response to natural and anthropogenic forcings cannot be readily addressed with traditional models of the physical climate. One of the open issues for near-term climate change, for example, is the response of terrestrial ecosystems to increased concentrations of carbon dioxide. Will plants begin releasing carbon dioxide to the atmosphere in a warmer climate, thereby acting as a positive feedback, or will vegetation absorb more carbon dioxide and hence decelerate global warming? Related issues include the interactions among land use change, deforestation by biomass burning, emission of greenhouse gases and aerosols, weathering of rocks, carbon in soils, and marine biogeochemistry.
Exploration of these questions requires a comprehensive treatment of the integrative Earth system. In order to address these emerging issues, physical models are being extended to include the interactions of climate with biogeochemistry, atmospheric chemistry, ecosystems, glaciers and ice sheets, and anthropogenic environmental change. These new "Earth System Models", however, will require large investments in computing infrastructure before they can be fully utilized.
Climate varies naturally. We consider natural variability as resulting from purely internal atmospheric processes as well as from interactions among the different components of the climate system, such as those between the atmosphere and oceans or the atmosphere and land. However, the most significant forcings with impact on climatic time scales are generally imposed upon the climate system.
External forcings arise from a wide array of processes covering a range of spatial and temporal scales. "Natural" external forcings include changes in the global configuration of the continents, the slow increase of solar luminosity that occur over hundreds of millions of years, variations in the Earth's orbit, and the injection of aerosols high into the atmosphere by explosive volcanic eruptions. Human emissions of carbon dioxide and other greenhouse gases, the local emission and suspension of small (aerosol) particles on timescales of minutes to days, and changes in land use are some examples of anthropogenic forcings.
The global temperature variations reflected in ice core records from the distant past reflect the influence of natural external forcings on the climate system. However, these reconstructions of past temperature swings have also demonstrated that the projected rate of global temperature change exceeds anything seen in nature in the past 10,000 years.
Greenhouse gas concentrations in the atmosphere are now higher than at any time in at least the last 750,000 years. It took at least 10,000 years from the end of the last ice age for levels of carbon dioxide to increase 100 ppmv to 280 ppmv, but that same increase has occurred over only the past 150 years to current values of over 370 ppmv. About half of that increase has occurred over the last 35 years, owing mainly to combustion of fossil fuels and deforestation. In the absence of controls, future projections are that the rate of increase in carbon dioxide amount may accelerate, and concentrations could double from pre-industrial values within the next 50 to 100 years.
Global average temperature increases in recent decades are primarily due to increasing greenhouse gas concentrations in the atmosphere resulting from human activities, such as the burning of fossil fuels. Yet, the magnitude of the anthropogenic influence on regional climate remains uncertain. A principal reason is because the effects of human activities are superimposed on the background "noise" of natural climate variability, which can be very large regionally.
Global warming does not mean that temperature increases are spatially uniform or monotonic: some places warm more than the average and some places cool. Regional changes in temperature are often associated with changes natural patterns (or modes) of the atmospheric and oceanic circulation, such as the El Niño/Southern Oscillation (ENSO) phenomenon. Changes in the climate system from human activities may affect these modes, however, so quantifying the anthropogenic and natural components of the observed warming on regional scales remains a difficult and critical research question.
Many global climate models, for instance, project changes in the statistics of ENSO variability with global warming, specifically of greater ENSO activity marked by larger interannual variations relative to the warmer mean state. More El Niño events would increase the probability of weather regimes that favor, for instance, regional cooling over the North Pacific Ocean with warming over much of northwest North America. Yet, the details of ENSO are not well enough simulated in climate models to have full confidence in these projected changes, in part because the positive atmosphere-ocean feedbacks involved with ENSO mean that small errors in simulating the relevant processes can be amplified.
Thus, while it is likely that changes in ENSO and other natural modes of climate variability will occur as a result of anthropogenic climate change, their nature, how large and rapid they will be, and their implications for regional climate change around the world remain uncertain.
Claims have been made that the surface temperature measurements are tainted by the proximity of data-generating thermometers to cities. While amplified warming does occur in urban areas and is an important local phenomenon, a number of independent and recent scientific studies have shown that urbanization is a negligible effect as far as continental- and hemispheric-space averages are concerned. Over land, temperature data come from fixed weather observing stations with thermometers housed in special instrument shelters. Records of temperature from many thousands of such stations exist. Some are in urban areas. Many are not.
One concern regarding the construction of global temperature records is the variety of changes that may affect temperature measurements at an individual station. For example, the thermometer or instrument shelter might change, the time of day when the thermometers are read might change, or the station might move. These problems are addressed through a variety of procedures (for example, checking for consistency with data from neighboring stations) that have proven to be very effective. Other, perhaps more subtle influences (e.g., urbanization) are addressed either actively in the data processing stage or through dataset evaluation to ensure as much as possible that the data are not biased. For instance, several studies have compared global surface temperature time series made up of only rural stations with the "standard" global temperature time series, only to find out that there is no significant bias.