Uncertainty and risk in climate projections for the 21st century

Probabilistic climate projections based on two SRES scenarios, an IMAGE reference scenario and five IMAGE mitigation scenarios (all of them multi-gas scenarios) using the Bern2.5D climate model are calculated. Probability distributions of climate model parameters that are constrained by observations are employed as input for the climate model. The sensitivity of the resulting distributions with respect to prior assumptions on climate sensitivity is then assessed. Due to system inertia, prior assumptions on climate sensitivity play a minor role in the case of temperature projections for the first half of the 21st century, but these assumptions have a considerable influence on the distributions of the projected temperature increase inthe year 2100.

Upper and lower probabilities for exceeding 2°C by the year 2100 are calculated for the different scenarios. Only the most stringent mitigation measures lead to low probabilities for exceeding the 2°C threshold. This finding is robust with respect to our prior assumptions on climate sensitivity.

Further, probability distributions of total present-value damages over the period 2000-2100 for the different scenarios are calculated assuming a wide range of damage cost functions, and the sensitivity of these distributions with respect to the assumed discount rate is investigated. Absolute values of damage costs depend heavily on the chosen damage cost function and discount rate. Nevertheless, some robust conclusions are possible.