The average difference between the values of the forecasts and the observations on the long term.
For impact studies, the raw output of climate models is often not useful because of significant biases in climate model output. For example:
– temperature can be consistently too low or too high
– rainfall is too low or too high
– models provide and incorrect simulations of the seasonal patterns of rainfall (e.g. peak rainfall too soon or too late)
– climate models tend to over estimate the number of days with rain (rain days) and underestimate precipitation extremes.
Downscaling is a method that derives local- to regional-scale (10 to 100 km) information from larger-scale models or data analyses. Two main methods exist: dynamical downscaling and empirical/statistical downscaling. The dynamical method uses the output of regional climate models, global models with variable spatial resolution or high-resolution global models. The empirical/statistical methods develop statistical relationships that link the large-scale atmospheric variables with local/regional climate variables. In all cases, the quality of the driving model remains an important limitation on the quality of the downscaled information.
Means lack of precision or that the exact value for a given time is not predictable, but it does not usually imply lack of knowledge. Often, the future state of a process may not be predictable, such as a roll with dice, but the probability of finding it in a certain state may be well known (the probability of rolling a six is 1/6, and flipping tails with a coin is 1/2). In climate science, the dice may be loaded, and we may refer to uncertainties even with perfect knowledge of the odds. Uncertainties can be modelled statistically in terms of pdfs, extreme value theory and stochastic time series models.
This project has received funding from the European Union's Horizon 2020 Research and Innovation programme under Grant agreement No. 776467