The stratospheric response and surface influence in a bias-corrected model.
<p>We investigate the effect of systematic model biases on teleconnections influencing the Northern Hemisphere wintertime circulation. We perform a two-step nudging and bias-correcting scheme for the dynamic variables of the ECHAM6 atmospheric model to reduce errors in the model climatology relative to ERA-Interim. The developed scheme is efficient in removing errors in model&#8217;s climatology. In particular, large negative bias in December-February mean zonal stratospheric winds is reduced by up to 75%, significantly increasing the strength of the Northern Hemisphere wintertime stratospheric polar vortex. <!-- I think calling increase in vortex strength &#8221;a result&#8221; somewhat shifts the focus. The result is reduced error, not increased strength. I mean technically it is the same, but perception of the result is a bit different. What do you think? -->The bias-corrections are applied to the full atmosphere or stratosphere only.</p><p>We compare the response of bias-corrected and control runs to internal stratospheric variability and surface forcings that are important on seasonal timescales: Siberian snow cover in October; the Quasi-Biennial Oscillation (QBO); and ENSO. We find the bias-corrected model has the potential for a strengthened and more realistic response to the teleconnections, either in the stratospheric or surface response. In particular, the bias-corrected model has a strong QBO teleconnection which modulates the extratropical polar vortex and sea level pressure variability in a manner similar to that seen in observations. The Siberian snow forcing with the stratosphere-only bias-corrections also leads to an enhanced surface response relative to the control.<!-- Given considerable remaining biases in tropospheric waves in the stratosphere-only bias corrected run I would not overemphasize the results in this particular run especially because full bias-corrected run does not show large surface response. But it is Ok to mention this result in the abstract. --> The mechanism behind the sensitivity of the teleconnections to model biases is discussed.</p>