To date, no model building process can ensure full representation of the complex climate-economic processes – instead, multiple highly detailed models are put forward by individual research groups to capture some selected aspects. On the other hand, a number of the simplified integrated assessment models (IAMs) have been developed attempting to consider the full causal loop between accumulated emissions, economy and climate, and study associated uncertainty.
Here we present a simplified system dynamics IAM based on the model from Kovalevsky and Hasselmann (2014) with stochastic climate sensitivity and a nonlinear climate damage function. We explore the structural sensitivity of the long-term projections (focusing on global temperature, global economy output, GHG emissions and atmospheric concentrations) to a probabilistic distribution describing the climate sensitivity. We investigate the model robustness under different assumptions on climate sensitivity distribution. For this purpose, we use the approach suggested by Kryazhimskiy (2016), which attempts to ‘integrate’ several independent distributions representing the same variable into one posterior distribution using a Bayesian approach based on the posterior event being the one when stochastic variables in all models have the same realization. The results show that model ‘integration’ leads to a higher mean global output, emissions and concentrations and lower mean global temperature than both prior means, coming along with lower uncertainty in the integrated scenario.
The authors would like to acknowledge DG research for funding through the FP7-funded COMPLEX project #308601, www.complex.ac.uk.

1. D.V. Kovalevsky, K. Hasselmann (2014): Assessing the transition to a low-carbon economy using actor-based system-dynamic models, Proceedings of the 7th International Congress on Environmental Modelling and Software (iEMSs), 15-19 June 2014, San Diego, California, 4, 1865-1872

2. A.V. Kryazhimskiy (2016): Posteriori integration of probabilities. Elementary theory, Theory Probab. Appl., 60:1, 62–87