tree rings. These reconstructions are typically associated with as much uncertainty as reconstructions of surface temperature.
Climate models are often used to simulate the response of climate to variations in external forcing, including those indicated by proxy evidence. A variety of models have been used to estimate the surface temperature variations implied by the available proxy data for the last 2,000 years. Climate models can also be used to study the feedbacks that determine the response of the global mean surface temperature to external forcings and also to estimate the natural internal variability of the climate system.
The temperature of the Earth is determined by a balance of the energy entering the Earth–atmosphere system and the energy leaving the system. An energy imbalance imposed on the climate system either externally or by human activities is termed a climate forcing (NRC 2005); persistent climate forcings cause the temperature of the Earth to change until an energy balance is restored. The amount of change is determined by the magnitudes of the climate forcings and the feedbacks within the climate system that amplify or diminish the effect of the forcings (NRC 2003b). Climate forcings that directly affect the radiative balance of the Earth are termed radiative forcings and are typically measured in watts per square meter (W·m–2) (NRC 2005). Positive global mean radiative forcings result in warming of global mean surface temperatures. Volcanic eruptions, changes in the Sun’s radiative output, and the mostly anthropogenic changes in greenhouse gases, tropospheric aerosols, and land use are the main climate forcings for surface temperatures over the last 2,000 years.
The primary natural greenhouse gases are carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O). Water vapor is also a greenhouse gas that contributes the largest warming, but it is treated as a feedback because its concentration is controlled by the temperature of the atmosphere rather than by human activities. Continuous atmospheric measurements of carbon dioxide have been available since the mid-20th century from the Mauna Loa Observatory, and all of the significant greenhouse gases have been monitored since 1980 by the National Oceanic and Atmospheric Administration’s global air sampling network (Keeling and Whorf 2005). For previous decades and centuries, greenhouse gas concentrations are obtained by analyzing air bubbles trapped in cores and firn of ice in Antarctica and Greenland.
Over the glacial–interglacial cycles of the last 650,000 years, carbon dioxide varied between about 300 ppm (parts per million by volume) during warm interglacial periods and about 180 ppm during cold glacial periods (Siegenthaler et al. 2005a). Methane and nitrous oxide atmospheric concentrations during interglacial periods did not exceed 790 ppb (parts per billion by volume) and 290 ppb, respectively (Spahni et al. 2005). Carbon dioxide, methane, and nitrous oxide varied little over the past 2,000 years prior to the industrial era (Figure 10-1). Ice core measurements indicate that carbon dioxide and nitrous oxide remained within a few ppm and ppb, respectively, of their mean concentrations and within the uncertainties of the data (Raynaud et al. 2003, Gerber et al. 2003). Methane fluctuated between 600 and 750 ppb, changing with the climate and likely resulting from fluctuations in natural and early anthropo-