A21F-0217
Detection and Attribution of Simulated Climatic Extreme Events and Impacts: High Sensitivity to Bias Correction

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Sebastian Sippel1, Friederike Elly Luise Otto2, Matthias Forkel1, Myles Robert Allen3, Benoit P Guillod2, Martin Heimann1, Markus Reichstein1, Sonia I Seneviratne4, Kirsten Thonicke5 and Miguel D Mahecha1, (1)Max Planck Institute for Biogeochemistry, Jena, Germany, (2)University of Oxford, ECI/School of Geography and the Environment, Oxford, United Kingdom, (3)University of Oxford, Physics, Oxford, United Kingdom, (4)ETH Swiss Federal Institute of Technology Zurich, Zurich, Switzerland, (5)Potsdam Institute for Climate Impact Research, Potsdam, Germany
Abstract:
Understanding, quantifying and attributing the impacts of climatic extreme events and variability is crucial for societal adaptation in a changing climate. However, climate model simulations generated for this purpose typically exhibit pronounced biases in their output that hinders any straightforward assessment of impacts. To overcome this issue, various bias correction strategies are routinely used to alleviate climate model deficiencies most of which have been criticized for physical inconsistency and the non-preservation of the multivariate correlation structure. We assess how biases and their correction affect the quantification and attribution of simulated extremes and variability in i) climatological variables and ii) impacts on ecosystem functioning as simulated by a terrestrial biosphere model.

Our study demonstrates that assessments of simulated climatic extreme events and impacts in the terrestrial biosphere are highly sensitive to bias correction schemes with major implications for the detection and attribution of these events. We introduce a novel ensemble-based resampling scheme based on a large regional climate model ensemble generated by the distributed weather@home setup[1], which fully preserves the physical consistency and multivariate correlation structure of the model output. We use extreme value statistics to show that this procedure considerably improves the representation of climatic extremes and variability. Subsequently, biosphere-atmosphere carbon fluxes are simulated using a terrestrial ecosystem model (LPJ-GSI) to further demonstrate the sensitivity of ecosystem impacts to the methodology of bias correcting climate model output.

We find that uncertainties arising from bias correction schemes are comparable in magnitude to model structural and parameter uncertainties. The present study consists of a first attempt to alleviate climate model biases in a physically consistent way and demonstrates that this yields improved simulations of climate extremes and associated impacts.


[1] http://www.climateprediction.net/weatherathome/

<< Previous Abstract | Next Abstract >>