Envision that project would have two components:

1. Proposed Project for multiweek prediction of Southern (and Northern) Hemisphere Stratospheric Warmings and Sustained Low-SAM Background: Since 1979; only 1 major sudden stratospheric warmings (SSW) in SH has been observed (2002) but some minor warmings have occurred (2009, 2012) SSW in NH are more common but still viewed as unusual events Cause of the SH SSW and why they are so infrequent is unknown (presumably winds too strong most of winter, and wave driving is relatively weak) SSW are associated with subsequent sustained periods of low SAM. SAM is associated with distinctive surface climate anomalies throughout SH, so prediction of the warmings and could be an important “window of opportunity” for useful multiweek prediction of climate same thing holds for NH SSW and shift to low NAM/NAO Aim of subproject is to determine: i) predictability and causes of unique SH SSWs in 2002 (and 2012) ii) predictability of multi-week tropospheric anomales following SSW (both NH and SH) does resolving/initializing the stratosphere provide for improved multiweek predictions of tropospheric climate? iii) development of diagnostics/products to monitor real time predictions

2. Envision that project would have two components: • 1) What is current predictive capability for Strat warmings themselves (SNAP)? • sensitivity to initialization in stratosphere • sensitivity to resolution in stratosphere • sensitivity to parameterized gravity wave drag, etc • role of pre-conditioning that primes warming (getting at why there are so few warmings). Is the pre-conditioning predictable? • 2) Can the SSW be exploited for extended range prediction of tropospheric climate due to sustained shift to low phase of the SAM/NAM after the warming? • how well predicted are the subsequent regional climate impacts (e.g. reduced rainfall across subtropical Australia for subsequent shift to low SAM) • Additional outcomes would include: • Understanding of pre-conditioning for the warmings (why are there so few) • mechanism by which stratosphere drives sustained shift of the SAM/NAM • improved assessment of the predictability of SAM/NAM (including its sources) • development of forecast products to monitor real time predictions

3. Thompson et al

4. Surface climate anomalies in October following anomalous SH polar cap ozone in September looks like typical low SAM anomalies

5. An apparent warimg/shift to low SAM also occurred in Oct 2012

6. Polar cap (60-90) column ozone October 2012 2002 It appears there may have been an strong warming in 2012, at least based on occurrence of high ozone in Oct. This would have been the only other strong warming besides 2002

7. Zonal mean zonal wind at 10mb 2012 2002 Although 2012 warming is interesting, it was much weaker than the 2002 event)

8. Anomalies

9. 2002 2012 SLP Anomalies

10. Cooling events also occur and are associated with shifts to high SAM

11. Next few slides look at how well predicted was the warming/low SAM in 2012 Look at POAMA and NCEP/CFS

12. POAMA forecasts of low SAM following 2012 SSW in id Oct Time series of observed SAM from 1 Nov POAMA ensemble from 11 Oct not much hint of low SAM for forecasts initialized prior to SW but some hint of sustained low SAM if initialized when SAM is already low

13. looks better with NCEP but still under-predicted beyond 10 days

14. Next slide is a word of caution: it is difficult to show that resolving the stratosphere leads to better predictions of the SAM. We note however, that we probably don’t have the L85 version of UM7.3 set up optimally (e.g. forecasts in troposphere are worse in first few days, which should not happen)

15. Skill score T rmse over south polar cap 1 11 21 31 41 Roff et al. UM 6.4 L50 relative to L38 1979-2008, 15 Nov start ERA-40 I.C.s UM7.3 (HadGEM3 settings) L85 relative to L38 Start times 1 Sep, 1Oct, 1 Nov 1982-2009 ERA-Interim I.C.s BoM experience is that fully resolving the stratosphere does provide improved predictions of the SAM after about 20 days, but improvement is modest (ie reduction in RMSE by about 5%). Note that these results are for all start dates and include strong and weak vortex variability. Impacts might be much larger if only look at forecasts when vortex is highly perturbed.

16. Stratospheric Network for the Assessment of Predictability (SNAP) Three year (2013-2016) program funded by UK NERC and WCRP/SPARC (an ‘emerging activity’) • The scientific aims of the project are to quantify: • current skill in forecasting the extra-tropical stratosphere, • the extent to which accurate forecasts of the stratosphere contribute to improved tropospheric predictability, • the partitioning of any gains in predictability with a well-resolved stratosphere between improvements in the estimation of initial conditions and improvements in the forward forecast. • The centrepiece of SNAP will be to design and perform a new intercomparison of stratospheric forecasts • This will also leave a legacy of datasets to be used by a broad community of researchers

17. Project Partners • Forecasting centres: provide staff time, computer time • Met Office (UK); Environment Canada; Met Research Institute (Japan); Naval Research Laboratory (USA); Bureau of Meteorology (Australia) • Research organizations: provide staff time, analysis • Exeter University (UK); New York University (USA) • More partners welcome (S2S seems pretty obvious; hhh) • E-mail: sparc.snap@gmail.com

18. Next Steps • January 2013: form Steering Committee • April 2013: First Workshop – review existing science and future directions; design a stratospheric predictability experiment; produce an experimental strategy (Frederic will attend?) http://www.met.reading.ac.uk/~sws05ajc/DynVar_SNAP_Workshop/ • From June 2013: start running stratospheric predictability experiments