Project Details
Projekt Print View

Comparison of Metrics for Probabilistic Climate Change Projections of Mediterranean Precipitation (COMEPRO)

Subject Area Physical Geography
Term from 2014 to 2019
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 256844028
 
Final Report Year 2019

Final Report Abstract

The aim of the COMEPRO project was to contribute to the interpretation and evaluation of PDFs of future climate change from multi-model ensembles, focussing on future temperature and precipitation changes along the Mediterranean Basin including extremes of precipitation. Various methods of assessing the model performance were compared with each other and combined with different available observational data sets. The work plan and gained results can be subsumed under three major research issues: (1) development and comparison of a large spectrum of established and novel statistical metrics to derive weighting factors for climate models, (2) a statistical downscaling approach for deriving model weights, based on the varying capability of the global climate models to represent large-scale modes of variability, key predictors for statistical downscaling and non-stationarities in the predictorpredictand relationships, (3) the relationship between model characteristics and model performance. Overall, our study revealed a considerable improvement of model performance of CMIP5 over CMIP3 that is still outperformed by the CORDEX regional climate model simulations. The application of model weights affects the PDFs of future climate change in a rather diverse way, depending on the season, region, scenario and climate variable. Some of the metrics, especially the ones including the Bayesian theorem, have the potential to reduce the ensemble uncertainty in probabilistic climate projection. Partly, the PDFs of projected climate change can be quite narrow. Most methods appear rather stable when altering the considered time period or reference data. Thus, the approach is very effective in changing the PDFs of future climate change and, hence, appropriate when this is major objective, e.g. for adaptation research. Statistical downscaling provides a substantial contribution to the aspect of model evaluation and weighting, especially when model weights are derived according to the varying capability of global climate models to represent key predictors for statistical downscaling like circulation and humidity variables for the case of assessing precipitation. The effect appears to be more consistent for extreme rather than seasonal precipitation in the Mediterranean basin. The consideration in the model weighting of the models’ capability to reproduce non-stationarities between predictors and predictands has an immense effect on the ranking results for the CMIP models. Another important result is that the convection scheme might be a useful criterion when a sub-ensemble of GCMs is to be formed from a larger one with the aim of reducing the ensemble spread. Two aspects of the original work plan have not been successful: the assessment of model uncertainty along the whole process chain from radiation processes over evaporation, advection, convection and cloud processes to (extreme) precipitation has revealed marginal effect but only when the climate models are grouped according to the implemented convection scheme. The statistical downscaling approach on the basis of circulation types originating from particular classification techniques has been thoroughly tested but was not found to be appropriate for our purpose. In contrast, we were positively surprised by the fact that the model weights are quite consistent across different metrics. Thus, they represent a robust instrument to affect PDFs of climate change, e.g. for applications in planning and adaptation measures. Yet, the weighting algorithm has to be tailored thoroughly to a given research issue in a specific region and season with respect to a well-defined aspect of climate change.

Publications

  • (2015): Präzisierung von Niederschlagsprojektionen im Mittelmeerraum durch bestmögliche Gewichtung von Multimodellensembles. 34. Jahrestagung des AK Klima, Hattingen: 37
    Kaspar-Ott I., Hertig E., Pollinger F., Ring C., Paeth H., Jacobeit J.
  • (2016): A comparison of metrics for assessing state-of-the-art climate models and implications for probabilistic projections of climate change. Clim Dyn.
    Ring, Christoph; Pollinger, Felix; Kaspar-Ott, Irena; Hertig, Elke; Jacobeit, Jucundus; Paeth, Heiko
    (See online at https://doi.org/10.1007/s00382-017-3737-3)
  • (2016): Development and comparison of weighting metrics for probabilistic climate change projections of Mediterranean precipitation. EGU General Assembly, Vienna 2016. Geophysical Research Abstracts 18: 12422
    Kaspar-Ott I., Hertig E., Pollinger F., Ring C., Paeth H., Jacobeit J.
  • (2016): Empirische Gewichtung von Multi-Modell Ensembles für probabilistische Klimawandelprojektionen im Mittelmeerraum, 35. Jahrestagung des AK Klima, Bad Dürkheim: 57
    Pollinger, F., Ring C., Pollinger F., Kaspar-Ott I., Hertig E., Jacobeit J., Paeth H.
  • (2016): Fingerprinting zur Evaluation von Klimamodellen und Abschätzung der Unsicherheiten über Zukunftsprojektionen. 35. Jahrestagung des AK Klima, Bad Dürkheim: 54
    Ring C., Pollinger F., Kaspar-Ott I., Hertig E., Jacobeit J., Paeth H.
  • (2016): Gewichtung von Klimamodellen mittels statistischer Downscaling-Ergebnisse für den Niederschlag des Mediterranen Raums. 35. Jahrestagung des AK Klima, Bad Dürkheim: 46
    Kaspar-Ott I., Hertig E., Jacobeit J., Ring C., Pollinger F., Paeth H.
  • (2017): Change points in predictors-predictand relationships within the scope of statistical downscaling. Int. J. Climatol. 37: 1619-1633
    Hertig, E., Merkenschlager, C., Jacobeit, J.
    (See online at https://doi.org/10.1002/joc.4801)
  • (2017): Development and comparison of metrics for evaluating climate models and estimation of projection uncertainty. EGU General Assembly, Vienna 2017. Geophysical Research Abstracts 19: 6774
    Ring C., Pollinger F., Kaspar-Ott I., Hertig E., Jacobeit J., Paeth H.
  • (2017): Gewichtete Multimodellensemble-Projektion von Niederschlagsextremen im Mittelmeerraum unter Anwendung von statistischem Downscaling. 36. Jahrestagung des AK Klima, Rauischholzhausen: 70
    Keupp L., Kaspar-Ott I., Hertig E., Pollinger F., Ring C., Paeth H., Jacobeit J.
  • (2018): Extreme precipitation in the Mediterranean area – weighted multi-model ensemble projections using statistical downscaling. EGU General Assembly, Vienna 2018. Geophysical Research Abstracts 20: 19559
    Keupp L., Hertig E., Kaspar-Ott I., Pollinger F., Ring C., Paeth H., Jacobeit J.
  • (2019): Weighted multi-model ensemble projection of extreme precipitation in the Mediterranean region using statistical downscaling. Theor. Appl. Climatol.
    Keupp, L., Hertig, E., Kaspar-Ott, I., Pollinger, F., Ring, C., Paeth, H. & J. Jacobeit
    (See online at https://doi.org/10.1007/s00704-019-02851-7)
  • (2019): Weighting CMIP3 & CMIP5 models with respect to Mediterranean precipitation in a statistical downscaling framework. EGU General Assembly, Vienna 2019. Geophysical Research Abstracts 21: 1833
    Kaspar-Ott I., Hertig E., Kaspar S., Pollinger F., Ring C., Paeth H., Jacobeit J.
  • (2019): Weights for general circulation models from CMIP3/CMIP5 in a statistical downscaling framework and the impact on future Mediterranean precipitation. Int J Climatol.
    Kaspar-Ott I, Hertig E, Kaspar S, Pollinger F, Ring C, Paeth H and Jacobeit J
    (See online at https://doi.org/10.1002/joc.6045)
 
 

Additional Information

Textvergrößerung und Kontrastanpassung