Further Discussion With Zhongfeng Xu On The Value Of Dynamic Downscaling For Multi-Decadal Predictions

In the post

Question And Answer On The Value Of Dynamic Downscaling For Multi-Decadal Predictions

two colleagues of mine and I discussed the significance of their new paper

Xu, Zhongfeng and Zong-Liang Yang, 2012: An improved dynamical downscaling method with GCM bias corrections and its validation with 30 years of climate simulations. Journal of Climate 2012 doi: http://dx.doi.org/10.1175/JCLI-D-12-00005.1

This post continues this discussion with  Zong-Liang Yang of the University of Texas in Austin and Zhongfeng Xu of the Institute of Atmospheric Physics of the Chinese Academy of Science.

Following is the comment by Zhongfeng, with my responses embedded.

Dear Roger,

Thank you for your interest to our paper.

In terms of your comments “their results show that they are not adding value to multi-decadal climate projections”. I think the comment is not accurate enough. We did not compare the climate changes simulated by IDD and TDD in the paper.

My Comment:

What you and Liang have very effectively documented are systematic errors in the observationally unconstrained model runs. You did not compare climate change, but you do show that the model results are biased. This bias is an impediment to skillful multi-decadal forecasts as it shows errors in the model physics and dynamics at that level. The elimination of these errors in the unconstrained runs is a necessary condition for skillful multi-decadal global model predictions.

Zhongfeng continues

So it’s too early to make conclusion whether IDD has adding value to climate change simulation.

My Response

To show skill, one has to show that changes in regional climate statistics between your control and your “future” are skillfully predicted. For model predictions in the coming decades, it is not enough to predict the same climate statistics, one must also skillfully predict changes to these statistics. Otherwise, the impact community could just as well use reanalyses.

Zhongfeng continues

 I guess it’s possible that IDD improves climate change projection when the GCM does a good job in producing climate change signals but producing a bad climatological means.

My Response

This cannot be correct. If the climatological means are in error, there are clearly problems in the model physics and dynamics. Also, what evidence do you have that the GCM does a good job in terms of multi-decadal predictions? [please see my post http://pielkeclimatesci.wordpress.com/2012/05/08/kevin-trenberth-is-correct-we-do-not-have-reliable-or-regional-predictions-of-climate/]

Zhongfeng continues

I will pay more attention to the IDD performance in climate change projection in our future study. I will keep you updated if we find some interesting results.

My Response

I look forward to learning more on your study. Thanks!

Zhongfeng continues

BTW: The IDD does significantly improve the projection of climatological mean. It’s still better than TDD which shows larger bias than IDD in projecting climatological means.

My Response

However, the global model multi-decadal predictions still are run with these biases. Even if you use IDD for the interior, the global model still has these errors meaning they have substantive physics and/or dynamic problems.

Zhongfeng’s comment

 Thank you for all your comments. They are very informative and make me thinking more about this dynamical downscaling study.  ^_^

My Reply 

I have also valued the discussion. I will add this as a weblog post follow-up. Your paper is a very important addition to the literature but the bottom line message is, in my view, documentation of why the impacts communities (e.g. for the IPCC assessments) should not be focusing on this methodology as bracketing the future of regional climates.

source of image

Comments Off

Filed under Climate Models, Debate Questions, Q & A on Climate Science

Comments are closed.