How do we deal with uncertainty connected with atmospheric circulation?

Transcription

1 How do we deal with uncertainty connected with atmospheric circulation? Ted Shepherd Grantham Professor of Climate Science Department of Meteorology University of Reading

2 Some addi-onal background

3 Circula-on aspects of climate such as the NAO exhibit pronounced chao%c variability on mul%- decadal %mescales, with generally no clear long- term trend This confounds the detec-on of anthropogenic changes In the early 2000 s the NAO trend since 1960 was ajributed to climate change; what would we say now?

4 Surface temperature (TS) changes are generally robust, and are predictable even in the near term In contrast, need large ensemble to determine forced response of precip For Eurasia/North Atlan-c, there is about a 30% chance of 55- year trends in SLP or precip being of opposite sign to the anthropogenic signal; yet change in risk is not small PDFs of DJF trends from 2005 to 2060 in the Eurasian/North Atlantic sector Control With climate change Single- model ensemble from Deser et al. (2012 Clim. Dyn.)

5 Contribu%ons to uncertainty of decadal- mean DJF precipita%on projec%ons Hawkins & SuJon (2011 Clim. Dyn.)

6 There has been a belief that storm tracks will shi] poleward under climate change, but the winter%me North Atlan%c seems different ERA- Interim climatology shows three preferred tracks Mean CMIP5 response to RCP 8.5 in late 21 st century Zappa et al. (2013 J. Clim.)

7 However the models have large systema%c biases in storm- track posi-on; generally too zonal and/or too far equatorward Mean CMIP5 response of winter-me storm track density to RCP 8.5 in late 21 st century Mean CMIP5 bias Zappa et al. (2013 J. Clim.)

8 SPARC DynVar CMIP5 analysis suggests GHG- induced changes in winter-me Arc-c sea- level pressure are affected as much by changes in the stratospheric polar vortex as by tropical upper tropospheric warming or by Arc%c surface warming Response to stratosphere is NAO- like and opposite in sign to that from tropical warming, in the same sense as the response to SSWs, Sigmond & Scinocca (2010) and Scaife et al. (2012) Tropical warming Arctic warming Stratosphere Manzini et al. (JGR, in press)

9 There is a connec%on between the circula%on response to forcing and variability The winter-me 500 hpa geopoten-al height response to thermodynamic forcings (here SST or sea- ice anomalies) consists of a direct baroclinic response and an indirect barotropic response which projects on modes of variability Deser et al. (2004 J. Clim.)

10 Modes of atmospheric variability behave somewhat like a damped spring The spring constant reflects the internal dynamical feedbacks Longer- term varia-ons reflect changes in forcing Circula-on response to forcing is stronger for a longer -mescale mode of variability (i.e. for a slacker spring) Formalized in the Fluctua%on- Dissipa%on Theorem x + r x +ω 2 x = F x = F /ω 2 = τ 2 F

11 Evidence for the FDT: Response of the tropospheric jet to a stronger stratospheric vortex depends sensi-vely on la-tude The authors argue the response depends on the strength of the eddy feedbacks within the troposphere Garfinkel, Waugh & Gerber (2013 J. Clim.)

12 More evidence: The zonal wind response to an applied torque is linear in the strength of the torque, and increases (linearly) with the decorrela-on -mescale of the zonal- wind anomalies, as expected from the Fluctua-on- Dissipa-on Theorem In these experiments, the decorrela-on -mescale is varied by changing the surface drag (while enforcing the same jet loca-on) Chen & Plumb (2009 JAS)

13 NWP models seem to sa%sfy their momentum balance in rather different ways: e.g. ECMWF has very small drag from subgridscale orography, and much more from the PBL scheme WGNE Drag Project, Report No. 1, A. Zadra (2013)

14 In the SH, models have much too long SAM %mescales, especially in late spring/early summer Most, but not all, of this model bias in (tropospheric) SAM -mescale arises from the stratosphere Could have implica-ons for SAM response to forcing CMAM CMAM with stratospheric AM variability suppressed Obs Simpson, Hitchcock, Shepherd & Scinocca (2011 GRL)

15 Kidston & Gerber (2010 J. Clim.) argued that the SH jet la-tude bias was responsible for the SAM %mescale bias in models Obs CMAM Yet bias- correc-ng the climatological tropospheric jet in CMAM does not reduce the bias in SAM -mescale Lesson: cannot rely on correla%ons; must break feedback loop between eddies and mean flow to iden-fy biases Simpson, Hitchcock, Shepherd & Scinocca (2013 J. Clim.)

16 In CMAM, the summer-me SAM -mescale bias arises from lack of damping of the SAM by planetary wave k=3 Posi-ve forcing of SAM by synop-c- scale eddies is OK This bias is evident only in the summer season The same bias is evident in all the CMIP5 models Simpson, Shepherd, Hitchcock & Scinocca (2013 J. Clim.)

17 Some ques%ons Do model biases majer for projec-ons? There may be biases in processes that are not apparent in the climatologies because of compensa-ng errors (e.g. sfc drag) Biases may show up in the variability If so, can we relate the biases to uncertainty in projec-ons (so- called emergent constraints )? Not much success so far in circula-on metrics Can we bias- correct the model projec-ons? Need to dis-nguish between simple cases (the rela-ve change is robust) and more complex cases Can we use the Fluctua-on- Dissipa-on Theorem quan-ta-vely? More generally, what (if anything) does internal variability tell us about the response to external forcing?