PCMDI

CAPT

Cloud Feedbacks

CMIP5

CMIP3

Other MIPs

Software

Publications

Google Calendar

Lab Calendar


Site Map

UCRL-WEB-152471

Privacy & Legal Notice

Thanks to Our Sponsors:

PCMDI > WCRP CMIP3 Model Output > Diagnostic Subprojects Printer Friendly Version
 
<< Back to WCRP CMIP3 Subprojects

  • Timbal, B., P. Hope and S. Charles, 2008: Evaluating the consistency between statistically downscaled and global dynamical model climate change projections. Journal of Climate, 21(22), 6052-6059.

The consistency between rainfall projections obtained from direct climate model output and statistical downscaling are evaluated. Results are averaged across an area large enough to overcome the difference in spatial scale between these two types of projections and thus make the comparison meaningful. Undertaking the comparison using a suite of state-of-the-art coupled climate models for two forcing scenarios presents a unique opportunity to test whether statistical linkages established between large-scale predictors and local rainfall under current climate remain valid in future climatic conditions. The study focuses on the south-west corner of Western Australia, a region which has experienced recent winter rainfall declines and for which climate models project, with great consistency, further winter rainfall reductions due to global warming. Results show that as a first approximation the magnitude of the modeled rainfall decline in this region is linearly related to the model global warming (a reduction of about 9% per degree), thus linking future rainfall declines to future emission paths. Two statistical downscaling techniques are used to investigate the influence of the choice of technique on projection consistency. In addition, one of the techniques was assessed using different large-scale forcings, to investigate the impact of large-scale predictor selection. Downscaled and direct model projections are consistent across the large number of models and two scenarios considered; i.e. there is no tendency for either to be biased; and only a small hint that large rainfall declines are reduced in downscaled projections. Amongst the two techniques, a Non-homogeneous Hidden Markov Model provides greater consistency with climate models than an analogue approach. Differences were due to the choice of the optimal combination of predictors. Thus statistically downscaled projections require careful choice of large-scale predictors in order to be consistent with physically based rainfall projections. In particular it was noted that a relative humidity moisture predictor, rather than specific humidity, was needed for downscaled projections to be consistent with direct model output projections.


Last Updated: 2008-10-22

<< Back to WCRP CMIP3 Subprojects
 
For questions or comments regarding this website, please contact the Webmaster.
 
Lawrence Livermore National Laboratory  |  Physical & Life Sciences Directorate