The unconditional probability distributions of future emissions and temperatures

How high should we build a dyke today, knowing that it will serve for more than 50 years? This depends on the unconditional probability distribution of future temperatures. We review the literature on estimates of future emissions for current policy scenarios and current pledge scenarios. Reviewing expert elicitations, abatement costs of scenarios, learning rates of technologies, fossil fuel supply side dynamics and geoengineering, we argue that scenarios with emissions largely beyond current policy scenarios and largely below current pledge scenarios are relatively unlikely. Based on this, we develop a transparent method to estimate unconditional probability distributions of future temperatures and temperature exceedance probabilities for use in Value at Risk stress tests in 2030, 2050 and 2100.