Generally, the main sources of uncertainty in regional climate projections are:
- The IPCC emission scenario. Different emission scenarios reflect different assumptions about the future global development (in terms of population, technology, socio-economy, energy use, etc.), which are translated into different future GHG emission levels.
- The Global Climate Model (GCM). Different GCMs describe the climate processes and feedbacks in different ways.
- The initialisation of the CGM. Different initialisation states makes different climate projections be more or less in phase with actual low-frequency climate oscillations, reflecting the natural climate variability.
- The Regional Climate Model (RCM). Different RCMs describe the climate processes and feedbacks in different ways.
Any combination of the above sources will generate a future climate that is more or less different from any other combination. Generally, it is beforehand not possible to say that the future climate simulated by a certain combination should be more realistic or probable than the climate simulated using another combination. The main way to deal with this uncertainty is to use a large ensemble of projections that encompasses different emission scenarios, GCM, initialisations and RCMs.
However, there are six main emission scenarios, on the order of 20-30 different GCMs (including different versions of the same model), in some cases with a few different initial states, and a number of RCMs that at least equals the number of GCMs. An ensemble including all possible combinations is thus huge and unattainable for computational reasons. Therefore smaller ensembles are commonly used, including a few representatives of all sources, to obtain at least an approximation of the total uncertainty although most probably underestimated.
Much research is devoted to estimating which of the uncertainty sources that are more or less important, i.e. that contributes more or less to the total uncertainty. The results are very difficult to generalise as they depend on the climate aspect considered. For example, for future mean temperature in the Mediterranean region one source may dominate but for future heavy precipitation in Scandinavia another source. In many cases, however, the GCM (and its initialisation) is found to contribute more to the total uncertainty than emission scenario and RCM.
In regional climate change impact modelling, where a subsequent model uses RCM output (in terms of temperature, precipitation, and/or other meteorological variables), using even a limited ensemble of projections may be difficult or even unattainable in light of the resources available; in many cases only one or a few projections can be handled. Based on a larger ensemble, it is sometimes possible to select projections that satisfies certain criteria of the impact modeller, e.g. a projection with a temperature increase in the close to the mean of the total range or a projection with a precipitation increase close to the high end of the range. But, again, how a certain projection relates to others depends on the climate aspect considered and it is difficult to generalise.
In SUDPLAN, initially a small ensemble of four projections is made available for uncertainty assessment. The ensemble includes projections from three of the leading GCMs: HADCM3, CCSM3 and ECHAM5. All three have simulated the future climate using emission scenario A1B, representing a medium level of future GHG emissions. Further, ECHAM5 have simulated the future climate using emission scenario A2, representing a high level of future GHG emissions. All four global projections have been regionally downscaled over Europe using the regional model RCA3. The small ensemble is designed to include the most relevant uncertainty sources for the subsequent impacts considered in SUDPLAN, based on previous experiences at SMHI. It is intended to expand the ensemble with more projections later.
For further guidance to uncertainty aspects of climate projections and impacts, as well as how it affects decision-making, visit the European Climate Adaptation Platform (CLIMATE-ADAPT).