CONTROL ID: 1485487
TITLE: The Causes and Implications of Persistent Atmospheric Carbon Dioxide Biases in Earth System Models
AUTHORS (FIRST NAME, LAST NAME): Forrest M Hoffman1, 2, James Tremper Randerson1
INSTITUTIONS (ALL): 1. Earth System Science, University of California, Irvine, CA, United States.
2. Computational Earth Sciences Group, Oak Ridge National Laboratory, Oak Ridge, TN, United States.
ABSTRACT BODY: Increasing atmospheric carbon dioxide (CO2) concentrations, resulting from anthropogenic perturbation of the global carbon cycle, are altering the Earth’s climate. The strength of feedbacks between a changing climate and future CO2 concentrations are highly uncertain and difficult to predict using Earth System Models (ESMs). To reduce the range of uncertainty in climate predictions, model representation of feedbacks must be improved through comparisons with contemporary observations. In this study, the historical and future emissions-driven simulation results produced by 12 ESMs for the fifth phase of the Coupled Model Intercomparison Project (CMIP5) were analyzed. This analysis focused on simulations from fully coupled ESMs with interactive terrestrial and marine biogeochemistry models for the historical period and future period change based on the RCP 8.5 scenario. These simulations were forced with CO2 emissions, with global atmospheric CO2 mole fractions computed prognostically from atmospheric transport of these emissions, and interactions with ocean and land processes. The ability of models to accurately reproduce the observed atmospheric CO2 mole fraction trajectory over the historical period provides a broad indication of model fidelity, a necessary but not sufficient condition for credible ESM performance. Comparison of ESM prognostic atmospheric CO2 over the historical period with observations indicates that ESMs, on average, had a high bias in their predictions of contemporary atmospheric CO2. A key driver of this persistent bias was weak ocean carbon uptake exhibited by the majority of ESMs, based on comparisons with observationally-based estimates of ocean carbon inventories from Sabine et al. (2004) and Khatiwala et al. (2009). The high atmospheric CO2 bias for the multi-model mean produced radiative forcing that was too large and, therefore, unrealistically high temperature increases during the historical period. A significant linear relationship was found between the magnitude of the contemporary atmospheric CO2 biases and future CO2 levels for the multi-model ensemble. This relationship was exploited to create a “zero-bias model” estimate of the atmospheric CO2 trajectory for the 21st century. This approach reduced the spread of future atmospheric CO2 projections for the RCP 8.5 scenario and yielded radiative forcing and temperature increases during the middle of the 21st century that were lower than the multi-model mean. Our approach and results suggest that because many of the processes that contribute to contemporary carbon cycle biases persist over decadal timescales, uncertainties in future climate scenarios can be significantly reduced by tuning models to the long-term time series of CO2 obtained from Mauna Loa and other atmospheric monitoring stations.
KEYWORDS:  GLOBAL CHANGE / Earth system modeling,  BIOGEOSCIENCES / Carbon cycling,  OCEANOGRAPHY: BIOLOGICAL AND CHEMICAL / Biogeochemical cycles, processes, and modeling,  BIOGEOSCIENCES / Biosphere/atmosphere interactions.
(No Image Selected)
(No Table Selected)
Previously Presented Material: 0%
CONTACT (NAME ONLY): Forrest Hoffman
CONTACT (E-MAIL ONLY): forrest at climatemodeling dot org
TITLE OF TEAM: