We describe a novel approach to impact assessment that generates probabilistic distributions of climate change impacts and passes model and societal uncertainties in a continuous manner throughout the assessment process. Instead of driving impact assessment simulations based upon summary statistics, a range of idealized climate scenarios, or a subset of models and emissions scenarios, end-to-end assessment is conducted for each model under each emissions scenario. The resulting distribution of impacts may be used to elucidate internal dynamics of the system and to attach model and societal-based probabilities to individual outcomes.

To demonstrate the method, preliminary results from a study of the effect of climate change on Bangladesh's agricultural sector are presented. Analysis is conducted through an integrated framework of global climate models (GCMs) from a range of emissions scenarios in the Intergovernmental Panel on Climate Change's Fourth Assessment Report (IPCC AR4), hydrologic models (generating flood projections), and coastal models (generating sea-level rise and inundation projections), all used to drive biophysical agricultural models of major cereal crops in Bangladesh. By conducting end-to-end evaluations of each GCM/emissions scenario and combining the final agricultural impacts, probabilistic projections of cereal yield in Bangladesh are generated.