Developing downscaled probabilistic climate projections for regional integrated assessments

Background/Question/Methods

The Southeast US contains the highest levels of biodiversity in North America outside of the tropics. This is partly due to the climate over the last few millennia, characterized by abundant precipitation, mild temperatures, and low climatic variability. The Intergovernmental Panel on Climate Change  concluded that it is very likely that humans are responsible for increasing the 20^th century global average surface temperature by 1.0 ^oC through the release of greenhouse gasses (GHG) into the atmosphere. This warming is expected to continue into the future and is projected to cause considerable impacts on ecosystems. Thus, mitigation of, and adaptation to the impacts of climate change on ecosystems in the Southeast will likely be the key challenge confronting natural resource managers in the coming decades. Central to this is how to best implement an adaptive management strategy given the large uncertainty associated with climate change projections. This requires a careful treatment of this uncertainty as well as methods to downscale climate projections to the scale of ecosystem processes because of the coarse spatial resolution of the models. To date, most studies use the range of GCM output to represent the full range of projection uncertainty; thus increasing the risk of underestimating structural and parametric uncertainty associated with these projections. This underestimation propagates through associated integrated assessments that use climate change projections, leading to overconfident predictions. As a result, decision-makers may insufficiently hedge against the risks associated with low probability/high impact extreme climatic events. We address this by developing a suite of regional probabilistic climate change projections for the Southeast Regional Assessment Project (SERAP).

Results/Conclusions

Two core climatic datasets are used for base projections: (1) GCM simulations from the IPCC for fully coupled global-scale climate simulations; and (2) an Earth system Model of Intermediate Complexity (EMIC) to sample the parametric uncertainty of three key climate system variables. These datasets are further post-processed through: (1) Bayesian ensemble dressing methods to estimate structural uncertainty and the accuracy of the GCMs; and (2) statistically downscaled simulations forced by boundary conditions from the GCM and EMIC runs. The probabilistic projections generated through these methods are the basis for the ecological models and simulations used for SERAP and provide natural resource decision makers with a more coherent and rigorous quantification of projection uncertainty in climate models. Finally, using probabilistic projections can provide a more basis for developing strategies to adapt to climate change under a Robust Decision Making framework.