434
F. Garcia-Menendez et al.
The three principal sources of uncertainty in climate projections are greenhouse
gas emissions scenario, natural variability, and climate model response [3]. While
efforts to simulate climate change effects on air quality have explored the influence of emission scenarios and to a lesser degree natural variability, none have
systematically assessed the impact of climate model response. Multi-model ensembles that project the effect of climate on pollutant concentrations inherently cover a
range of climate responses across different climate-chemistry models. However, the
influence of climate model response in these ensembles cannot be discerned from
other sources of uncertainty including internal variability, parameter selection, and
scenario representation. Analyses that compare the weights of major climate uncertainties on projections of temperature and precipitation have found that, although
natural variability controls uncertainty initially and greenhouse gas emissions scenario eventually dominates, model response is consistently the second largest source
of uncertainty [4]. It is critical that impact projections based on climate projections
adequately consider their underlying uncertainties, including that related to climate
model response. Here we specifically evaluate its role and compare it to the other
major contributors to uncertainty in climate projections.
69.2 Methods
Our projections of climate-induced impacts on U.S. air quality are based on the integrated scenarios of economic activity, greenhouse gas emissions, and climate change
from the U.S. EPA Climate Change Impacts and Risk Analysis project [4, 6]. Socioeconomic activity and climate projections are generated with the MIT IGSM-CAM
integrated modeling framework [5]. Within the IGSM-CAM simulations, climate
model response is altered using a cloud radiative adjustment method to modify climate sensitivity to greenhouse gas forcing [7]. We model changes in climate under
climate sensitivities of 2.0, 3.0, or 4.5 °C mean surface temperature change per doubling of atmospheric CO 2 . The climate projections are used to drive the CAM-Chem
climate-chemistry model and simulate global air quality [2]. In the atmospheric
chemistry simulations, the effect of climate is isolated by holding emissions of conventional air pollutants and precursors at start-of-the-century levels. In addition to
variations in climate sensitivity, our simulation ensemble projects atmospheric chemistry at the start, middle, and end of the 21st century under a business-as-usual reference scenario (REF) and 2 climate policy scenarios with 2100 total radiative forcing
targets of 4.5 W m
−2 (P45) and 3.7 W m
−2 (P37). Projections of U.S. air quality
for each period and policy are based on 30-year atmospheric chemistry simulations
under 5 unique sets of initial conditions in the IGSM-CAM to account for natural
variability. The climate penalties on O 3 and PM 2.5 are estimated as the difference in
concentrations under present and future climates.
F. Garcia-Menendez et al.
The three principal sources of uncertainty in climate projections are greenhouse
gas emissions scenario, natural variability, and climate model response [3]. While
efforts to simulate climate change effects on air quality have explored the influence of emission scenarios and to a lesser degree natural variability, none have
systematically assessed the impact of climate model response. Multi-model ensembles that project the effect of climate on pollutant concentrations inherently cover a
range of climate responses across different climate-chemistry models. However, the
influence of climate model response in these ensembles cannot be discerned from
other sources of uncertainty including internal variability, parameter selection, and
scenario representation. Analyses that compare the weights of major climate uncertainties on projections of temperature and precipitation have found that, although
natural variability controls uncertainty initially and greenhouse gas emissions scenario eventually dominates, model response is consistently the second largest source
of uncertainty [4]. It is critical that impact projections based on climate projections
adequately consider their underlying uncertainties, including that related to climate
model response. Here we specifically evaluate its role and compare it to the other
major contributors to uncertainty in climate projections.
69.2 Methods
Our projections of climate-induced impacts on U.S. air quality are based on the integrated scenarios of economic activity, greenhouse gas emissions, and climate change
from the U.S. EPA Climate Change Impacts and Risk Analysis project [4, 6]. Socioeconomic activity and climate projections are generated with the MIT IGSM-CAM
integrated modeling framework [5]. Within the IGSM-CAM simulations, climate
model response is altered using a cloud radiative adjustment method to modify climate sensitivity to greenhouse gas forcing [7]. We model changes in climate under
climate sensitivities of 2.0, 3.0, or 4.5 °C mean surface temperature change per doubling of atmospheric CO 2 . The climate projections are used to drive the CAM-Chem
climate-chemistry model and simulate global air quality [2]. In the atmospheric
chemistry simulations, the effect of climate is isolated by holding emissions of conventional air pollutants and precursors at start-of-the-century levels. In addition to
variations in climate sensitivity, our simulation ensemble projects atmospheric chemistry at the start, middle, and end of the 21st century under a business-as-usual reference scenario (REF) and 2 climate policy scenarios with 2100 total radiative forcing
targets of 4.5 W m
−2 (P45) and 3.7 W m
−2 (P37). Projections of U.S. air quality
for each period and policy are based on 30-year atmospheric chemistry simulations
under 5 unique sets of initial conditions in the IGSM-CAM to account for natural
variability. The climate penalties on O 3 and PM 2.5 are estimated as the difference in
concentrations under present and future climates.