Seasonal prediction of European summer climate: a process-based approach
Seasonal climate predictions show very limited skill over Europe, especially forthe summer season. Those predictions are usually generated in ensembles andthe skill is assessed as the mean over all ensemble members. Most scientificstudies expect an increase in skill with an increase in ensemble size...
| Main Author: | |
|---|---|
| Format: | Doctoral or Postdoctoral Thesis |
| Language: | English |
| Published: |
Universität Hamburg
2019
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/21.11116/0000-0005-0E71-D http://hdl.handle.net/21.11116/0000-0005-0E73-B |
| Summary: | Seasonal climate predictions show very limited skill over Europe, especially forthe summer season. Those predictions are usually generated in ensembles andthe skill is assessed as the mean over all ensemble members. Most scientificstudies expect an increase in skill with an increase in ensemble size. However, theensembles can spread out with increasing lead time, such that seasonal climatepredictions over Europe show a high spread and an ensemble mean with lowvariability and values around the climatological mean.Here, I show a way to refine an ensemble by grouping the members accordingto the physical process they represent. For this, I assess which processes dominatethe climate of individual European summers and confirm that the dominantseasonal process can be explained by either a meridional or a zonal pressuregradient, in their positive or negative phase. The evaluated dynamical seasonalclimate prediction model is able to represent the spatial pattern and overallfrequency of occurrence of the assessed processes, but the individual membersdisagree on the process they predict for each summer. I thus show that the highspread of the ensemble results from the ensemble members predicting a variety ofphysical processes for European summers. A mean taken over all those membersthus averages over different physical processes, which is not physically consistent.For a physically consistent prediction, I restrict the ensemble mean to thosemembers, that predict the dominant physical process in each summer, which isobtained through observations. With such a refinement, significant hindcast skillcan be achieved over many parts of Europe and the North Atlantic, showing thatthe model is capable of predicting European summers if the physical processesare considered.In line with such a process-based approach I, instead of using observations toobtain the dominant physical process in each summer, show an alternative wayin which I am able to predict the zonal pressure pattern and its teleconnections. Iassess these connections in the ... |
|---|