There is a new paper, sent to me by Zong-Liang Yang, which examines the skill of multi-decadal global climate models to predict climate, as well as a method to correct systematic biases that exist in the parent global model. The paper is
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
The abstract reads [highlight added]
An improved dynamical downscaling method (IDD) with general circulation model (GCM) bias corrections is developed and assessed over North America. A set of regional climate simulations are performed with the Weather Research and Forecasting (WRF) model version 3.3 embedded in the National Center for Atmospheric Research’s (NCAR’s) Community Atmosphere Model (CAM). The GCM climatological means and the amplitudes of interannual variations are adjusted based on the National Centers for Environmental Prediction (NCEP)-NCAR global reanalysis products (NNRP) before using them to drive WRF. In this study, the WRF downscaling experiments are identical except the initial and lateral boundary conditions derived from the NNRP, original GCM output, and bias corrected GCM output, respectively.
The analysis finds that the IDD greatly improves the downscaled climate in both climatological means and extreme events relative to traditional dynamical downscaling approach (TDD). The errors of downscaled climatological mean air temperature, geopotential height, wind vector, moisture, and precipitation are greatly reduced when the GCM bias corrections are applied. In the meantime, IDD also improves the downscaled extreme events characterized by the reduced errors in 2-year return levels of surface air temperature and precipitation. In comparison with TDD, IDD is also able to produce a more realistic probability distribution in summer daily maximum temperature over the central United States-Canada region as well as in summer and winter daily precipitation over the middle and eastern United States.
This is a clearly written, very important paper, but for reasons the authors did not emphasize. First, however, I disagree with their first sentence in the paper where they write
An accurate regional projection of future climate and its impacts on society and environment have become crucial for public policy and decision-making.
Policymakers do not require accurate regional projections to make intelligent decisions as discussed in my son’s book The Climate Fix. Indeed, if one claims skill that actually does not exist, this is misleading policymakers.
The conclusion of the Xu and Yang 2012 contains the text
The most significant improvement of summer precipitation appears in the central United States-Canada region where the TDD overestimates precipitation by 0.5-1.5 mm d-1. The overestimated precipitation over the central United States-Canada region in TDD leads to a higher moisture content and enhanced evaporation, which in turn leads to a cold bias of surface air temperature. These significant errors in precipitation and surface temperature are largely removed in the IDD due to the GCM bias corrections.
The 2-year return level of summer daily maximum temperature simulated by the TDD is underestimated by 2-6°C over the central United States-Canada region. In contrast the bias is generally less than ±1°C in the IDD experiment.
which illustrates the level of error in the parent global model. Errors of this level would be expected outside of the regional climate model domain and as data is inserted into the regional model through the lateral boundary conditions (or through spectal nudging). The IDD makes an improvement in the regional domain since the model results are trained by real world data (the reanalysis). Such real world constraint, of course, is not available in predictions for the coming decades. The authors did not report on this limitation.
The bias correction, rather than providing a solution to improving multi-decadal climate model predictions, actually shows how poorly the models are doing. Providing results of these model predictions to policymakers as skillful projections is not appropriate.
I have e-mailed Liang (who is an internationally very well-respected colleague who I have the privilege to publish with; see) and wrote the following
I have read the paper and it is a very valuable new addition to the literature on dynamic regional downscaling. It involves the use of type 2/type 3 to assess one of the questions regarding the value of type 4, and shows, in my view, that as a type 4 application the GCMs are inadequate. A bias correction cannot remedy this deficiency for regional projections in the coming decades.
While you did not discuss this in the paper, your results, therefore, are in support of the findings that Rob and I reported on in our article in EOS.
You write, for example,
“A new dynamical downscaling method with GCM bias corrections for the regional projection of further [future] climate was developed and validated by comparing the GCM-driven WRF simulations to the NNRP-driven WRF simulation.”
which shows that the GCMs have systematic biases. These biases certainly influence the physics in the parent model (and show that these physics have serious problems in faithfully replicating the real climate system). For future predictions, there is no reanalysis data to bias correct towards reality. Thus this approach should not be used to claim skillful future projections.
For impact studies for the historical record, of course, one would just use the reanalyses. The GCM- regional downscaling would be informative for sensitivity model runs (e.g. land scale change effects, aerosols) by running with and without certain forcings, where the reanalysis is the control.
The future projections also have another requirement to overcome and that is, even IF they could recreate the historical climate statistics without bias correction, they must be able (in comparison to the real world data) be able to skillfully predict CHANGES in the regional climate statistics.
I would like to also post the announcement of the paper on my weblog, and discuss; let me know if you would want to first prepare a guest weblog post…..
I am pleased Liang shared this paper with me, and look forward to his response to my comments.