CRU : Research : STARDEX
STARDEX Deliverable D11
To recommend variables and extremes for which downscaling is required.
The need for downscaling for extremes: an evaluation of interannual variations in the NCEP reanalysis over Europe.
The work described in this deliverable differs in several respects from conventional GCM evaluations in order to provide specific insights into the need for downscaling.
In particular, the NCEP reanalysis is considered as a quasi-ideal GCM.
The focus is on temperature and precipitation extremes and their interannual variability.
The extensive STARDEX data sets of daily station series are used - suitably upscaled to provide a fair comparison at the scale of the GCM.
Three mutually-related measures are considered: standard correlation coefficient; ratio of variance; and debiased root mean square error - displayed using Taylor diagrams.
The results provide guidance on the need for downscaling and its dependence on season, type of extremes and region.
- Synthesis Report
- ADGB: Northern Italy precipitation
- ARPA-SMR: Emilia Romagna temperature and precipitation
- AUTH: Greek temperature and precipitation including biases
- CNRS: French Alps precipitation, focusing on reproduction of trend.
- ETH: Alpine precipitation
- FIC: squared standard anomaly and correlation coefficients for 481 European stations - coming soon
- UEA: UK temperature and precipitation including LEPS scores
- USTUTT-IWS: German Rhine precipitation and temperature including biases and reproduction of trends
Slides from these presentations may be used provided that STARDEX is duly acknowledged.
- D11 synthesis report outline Christoph Frei, October 2003
Last updated: July 2004, Malcolm Haylock