Abstract
The estimate of future climate changes is a great challenge for scientists. Climate models, numerical models of the climate system, are very useful instruments for these estimates.
However, model simulations of future development of the climate can only be taken as projections with many uncertainties. In present paper the main sources of these uncertainties in regional climate models are reviewed and several related issues are discussed.
Generally, the uncertainties in model outputs come from inaccuracies of initial and boundary conditions, parameterization of small scale processes, and the structure of the model. Additionaly, in case of the simulations of future climate, the unknown development of forcings that influence the climate system play also a role.
The multi-model ensembles provide a tool to analyze the structural uncertainty. The methods for such analysis are often based on weighting the models according to various criteria and metrics.
Model weights are usualy based on model performance in a reference period. Two issues connected to model validation are ilustrated here, e.i. the results of model performance evaluation depend on the choice of the metrics incorporated and the reference period.