WP3: Statistical downscaling, model verification and output localization
| WP leader: | dr. Radan Huth (Institute of Atmospheric Physics, Czech Republic) |
Objectives
O3.1 Construction of statistical downscaling models for the target areas / stations and variables
O3.2 Development and implementation of techniques of localization of RCM outputs into stations
O3.3 Validation of RCM and SDS outputs
O3.4 Construction of climate change scenarios for the target areas / stations and variables
O3.2 Development and implementation of techniques of localization of RCM outputs into stations
O3.3 Validation of RCM and SDS outputs
O3.4 Construction of climate change scenarios for the target areas / stations and variables
Description of work
The list of variables other than temperature and precipitation for which the SDS models should be built,
RCM output localization performed, validation carried out, and scenarios constructed, will be specified
based on the requirements of the impact WPs (5 to 7). The same holds for a potential enhancement of
the validation criteria.
The work on statistical downscaling methods will be distributed among the institutions based on their recent expertise: IAP - MLR, ANNs, classification-based methods; CUNI - ANNs and local models in phase space; NMA - CCA, conditional weather generator; ELU - stochastic downscaling.
The techniques for the localization of RCM outputs into stations and observation up-scaling will be developed separately in indicidual institutions for different dense datasets because different geographical settings (e.g., Alpine region vs. Hungarian flatlands) may require different methodologies to be adopted.
Each participating institution will be responsible for a specific validation task: IAP - temporal and spatial characteristics, trends; NMA - links between large-scale and local surface variables; BOKU - relationships between variables in terms of evapotranspiration; CHMI - higher order statistical moments; CUNI - distribution characteristics, annual cycles and other measures of correspondence; OMSZ, ELU - evaluation of the time slice RCM runs for the Carpathian Basin using local observations; NIMH - relationships of RCM outputs with local surface variables.
The construction of climate change scenarios will be distributed among the institutions according to the SDS method employed; furthermore, IAP will utilize weather generator Met&Roll with parameters modified by GCM / RCM outputs.
Finally, the results of the project achieved for the area of central and eastern Europe will be compared with the analogous results of the ENSEMBLES project, in order to better understand the degree of the transferability of the methods.
The work on statistical downscaling methods will be distributed among the institutions based on their recent expertise: IAP - MLR, ANNs, classification-based methods; CUNI - ANNs and local models in phase space; NMA - CCA, conditional weather generator; ELU - stochastic downscaling.
The techniques for the localization of RCM outputs into stations and observation up-scaling will be developed separately in indicidual institutions for different dense datasets because different geographical settings (e.g., Alpine region vs. Hungarian flatlands) may require different methodologies to be adopted.
Each participating institution will be responsible for a specific validation task: IAP - temporal and spatial characteristics, trends; NMA - links between large-scale and local surface variables; BOKU - relationships between variables in terms of evapotranspiration; CHMI - higher order statistical moments; CUNI - distribution characteristics, annual cycles and other measures of correspondence; OMSZ, ELU - evaluation of the time slice RCM runs for the Carpathian Basin using local observations; NIMH - relationships of RCM outputs with local surface variables.
The construction of climate change scenarios will be distributed among the institutions according to the SDS method employed; furthermore, IAP will utilize weather generator Met&Roll with parameters modified by GCM / RCM outputs.
Finally, the results of the project achieved for the area of central and eastern Europe will be compared with the analogous results of the ENSEMBLES project, in order to better understand the degree of the transferability of the methods.
Milestones
M3.1 month 6: datasets prepared, validation criteria formulated, list of variables for impacts agreed
M3.2 month 12: SDS methods developed
M3.3 month 18: RCM output localization methods developed and verified on ERA40 RCM runs
M3.4 month 24: validation and comparison of RCM and SDS models completed
M3.5 month 30: climate change scenarios for both time slices completed
M3.6 month 36: comparison with ENSEMBLES outputs finished
M3.2 month 12: SDS methods developed
M3.3 month 18: RCM output localization methods developed and verified on ERA40 RCM runs
M3.4 month 24: validation and comparison of RCM and SDS models completed
M3.5 month 30: climate change scenarios for both time slices completed
M3.6 month 36: comparison with ENSEMBLES outputs finished
