Session D6: Hybrid downscaling methods: How can CORDEX benefit from statistical approaches?
Abdelkader Mezghani, Rasmus Benestad, Stefan Sobolowski
NORCE Norwegian Research Centre and the Bjerknes Centre for Climate Research
Downscaling is a key aspect in most studies of climate change and its consequences, as it reduces the gap between climate model output and observations. Two widely used methods have been developed by two separate scientific communities such as dynamical (DD) and statistical downscaling (SD). While DD is primarily based on the physical understanding of processes and phenomena, SD tries to reproduce the statistical properties (e.g. variability, extremes, …) of particular climate variables or phenomena. Here, we welcome applications combining both methods to produce reliable local climate signal including uncertainties.
Downscaling is a key aspect in most studies of climate change and its consequences, as it reduces the gap between climate model output and observations. There are two widely used methods that are developed by two separate scientific communities: dynamical and empirical-statistical downscaling. While the dynamical downscaling approaches are primarily based on the physical understanding of processes and phenomena, statistical downscaling techniques focus more on reproducing the statistical properties of a particular climate variables or phenomena such as the mean, variability, and extremes. The two different approaches have different strengths and weaknesses, which implies that they can complement each other.
In this session we welcome applications combining the two downscaling techniques and discussing how the methods can benefit from one another to produce a more reliable and robust climate and climate change signal along with associated uncertainties. This session will include contributions related to statistical analysis techniques that make use of CORDEX datasets, innovative methods of combining different sources of regional downscaled climate information/data, and statistical approaches dealing with the uncertainty inherent in these datasets.