1. Separating the Wheat from the Chaff: Finding a Concensus Prediction For the Strength of Solar Cycle 24 D.A. Biesecker NOAA/SEC

2. It’s not the predictions Wheat –E. Kihn –M. Dikpati –D. Hathaway –D. Pesnell –T. Hoeksema –L. Svalgaard –R. van der Linden –H. Lundstedt –R. Thompson –DOD Representative –O.C. St. Cyr ex officio –J. Kunches ex officio Chaff –Biesecker The panel will convene for the first time in October Preliminary Prediction due in April, 2007

3. Overview What the heck do I know? Who wants to know about the next solar cycle –Pretty much everyone –Why? What are some of the prediction techniques –Statistical –Precursor What else is out there? –Sudden Turn-on –The skeptics Coming up with a consensus prediction –How did the last panel do?

4. Who wants a prediction? Almost everyone, it seems –Satellite operators Mission planning –18 year predictions (1 yr study, 2-yr build, 15-yr mission) Parts selection –Want to use previously qualified components »Don’t want to qualify components for a more severe environment unnecessarily Insurance costs –Mission lifetime and premature failures –NASA –DOD

5. Who wants a prediction? –HF Communications Small cycle – fewer available frequencies Lots of activity – more interruptions –Ham operators – the last line of defense in communications –Want to know how many storms to expect Electric Utilities Media And pretty much everyone else

6. What is an average solar cycle? Avg is 11.0±1.5 yrs [8.2, 15] Peak is 113.7±39.5 [48.7, 164.5] Rise to Max is 4.7±1.4 yrs [2.8, 7.5] Fall to Min is 6.3±1.3 yrs [3.5, 10.2] –Isn’t this good enough for predicting?

7. What predictions are out there? Precursor, Spectral Analysis, Climatology, Recent Climatology, Neural Network… As of May 5, 2006 11 published so far –8 below average –3 above average AuthorYear of Publication Peak SSN Hathaway et al.1999Strong Dikpati et al.2006>155 Gholipour et al.2005145 Kane1999105 Wang et al.2002101 Duhau200388 Schatten200580 Svalgaard200575 Badalyan et al.2001<50 Maris et al.2004Low Clilverd et al.2004Weak

8. The Solar Cycle

9. Accurate Predictions by 30 Months The amplitude and starting time, can be accurately determined by about 30 months from the start of the cycle. Doesn’t meet most ‘forecast’ requirements YEARS AFTER START OF CYCLE

10. Cycle 23 Prediction Prediction at month 30

11. Longer Forecast Techniques Techniques other than curve-fitting or auto-regression are needed to predict cycle amplitudes at times near or before sunspot minimum. 1) Use the average cycle. 2) Use trends or periodicities in cycle amplitudes. 3) Use information from cycle statistics. 4) Use information from other cycle indicators, specifically the geomagnetic indices. specifically the geomagnetic indices.

12. Secular Trend Since Maunder Minimum The Group Sunspot Number shows a significant secular increase in cycle amplitude since the Maunder Minimum.

13. Multi-Cycle Periodicities? After removing the secular trend, there is little evidence for any significant periodic behavior with periods of 2-cycles (Gnevyshev-Ohl) or 3-cycles (Ahluwalia) There is some evidence for periodic behavior with a period of about 9-cycles (Gleissberg).

14. Even-Odd Until Cycle 23 the past 6 Odd Numbered Cycles were larger than the preceding Even Numbered Cycle (8 out of a possible 11 cycles have met this ‘criterion’) 22 20 18 16 14 12 10 8 6 4 2

15. Geomagnetic Precursors Geomagnetic activity around the time of minimum seems to give an indication of the size of the next maximum. Ohl used aa indexOhl used aa index Feynmann split the aa index into two componentsFeynmann split the aa index into two components Thompson used only aa resulting from recurrent stormsThompson used only aa resulting from recurrent storms

16. Prediction Method Errors (Prediction-Observed) Testing Precursor Techniques 1) Back up in time to the beginning of each of the last five cycles. 2) Using only information from earlier times, recalibrate each technique and apply the results to that cycle. 3) Compare the predictions with the actual numbers.

17. Predicting the Solar Cycle With A Statistical Model Why not use a statistical model? –Climatology is a pretty standard forecasting technique. If it’s happened before, it’s likely to happen again. Wang et al. “The Prediction of Maximum Amplitudes of Solar Cycles and the Maximum Amplitude of Solar Cycle 24” Chin. J. Astron. Astrophys. 2, 557-562, 2002.

18. A statistical method A n : ‘rise to max’ D n : ‘fall to min’ M n : sunspot number K n : cycle length Rise Time SSN SSN ~ Rise Time –HRV 1, 5, 7, 9, 19, 21 M=312-37*A n –LRV All even cycles and 3, 11, 13, 15, 17, 23 M=263 – 38*A n

19. Predicting Cycle 24 K(n, n+1) = 1.95D n - 3.14 –For cycles “Similar” to 23 Based on M and A Cycle 24 is ‘Even’ so use LRV curve SSN = 101.3±18.1 Rise Time SSN

20. So lets look at dynamo based methods The basics –Large-scale polar fields on the decline of the solar cycle are converted to poloidal field in the next cycle Strength of polar fields → peak of next cycle

21. Schematic summary of predictive flux-transport dynamo model Shearing of poloidal fields by differential rotation to produce new toroidal fields, followed by eruption of sunspots. Spot-decay and spreading to produce new surface global poloidal fields. Transport of poloidal fields by meridional circulation (conveyor belt) toward the pole and down to the bottom, followed by regeneration of new toroidal fields of opposite sign. Courtesy of M. Dikpati

22. Precursor Method – Small Cycle L. Svalgaard et al. (2005) Predicting a small Cycle 24 –Even after the polar field reversal the old polarity flux still holds on in the poles –Poleward moving fields don’t fully fill pole until ~3 years before solar minimum

23. Svalgaard et al. continued Dipole Moment = ABS(North – South) –Assume SSN=0 if DM=0 Fit Cycles 22 & 23 –SSN = 0.6286*DM Cycle 24 DM (so far) –119.3 SSN = 75 CycleDM (μTesla) Observed SSN Predicted SSN 22245.1158.5154.1 23200.8120.8126.2 Cycle 22 Cycle 23 Cycle 24

24. Precursor Method – Large Cycle Dikpati et al (2006) Flux Transport Dynamo –Fully account for meridional circulation –Circulation takes 17-21 years to transport polar fields down to the shear layer –So, cycle N is influenced by cycles N-1, N-2, and N-3

25. Large-scale dynamo processes (i) Generation of toroidal (azimuthal) field by shearing a pre-existing poloidal field (component in meridional plane) by differential rotation (Ω- effect ) (ii) Re-generation of poloidal field by lifting and twisting a toroidal flux tube by helical turbulence (α- effect) (iii) Flux transport by meridional circulation = FLUX-TRANSPORT DYNAMO < Sun’s “memory” of past cycles controlled by meridional circulation.

26. Dynamos with Meridional Flow Recent Dynamo theories incorporate a deep meridional flow to transport magnetic flux toward the equator at the base of the convection zone. They explain the equatorward drift of activity, the poleward drift of weak magnetic elements on the surface, length of the cycle from the speed of the flow, and give a relationship between polar fields at minimum and the amplitude of future cycles. Dikpati and Charbonneau, ApJ 518, 508-520, 1999

27. Hathaway finds the sunspot cycle period is anti-correlated with the drift velocity at cycle maximum. The faster the drift rate the shorter the period. R=-0.5 95% Significant Drift Rate – Period Anti-correlation

28. Hathaway also finds that the drift velocity at cycle maximum is correlated to the amplitude of the second following (N+2) cycle maximum. The correlation is much weaker for the immediately following maximum. R=0.7 99% Significant Drift Rate – Amplitude Correlations

29. Cycle 24 Amplitude Prediction Based on the fast drift rates at the maximum of the last (22 nd ) cycle (red oval – northern hemisphere, yellow oval – southern hemisphere) Hathaway predicts a larger than average amplitude for the next cycle (24 th ). Slow drift rates during cycle 23 indicate a very small cycle 25.

30. Simulating relative peaks of cycles 12 through 24 (Dikpati, de Toma & Gilman, 2006) Dikpati et al. reproduce the peaks from cycle 16 through 23 They predict cycle 24 will be 30-50% larger than cycle 23 SSN = 157-181

31. Skeptic(s) Letter to Nature 442, 26 “Unpredictable Sun leaves researchers in the dark.” Tobias, S., Hughes, D., Weiss, N. –“The model proposed by Mausumi Dikpati and her team… relies on parametrization of many poorly understood effects. Although such parametrized models have been widely (and legitimately) used to explore specific features of dynamo processes, they have no detailed predictive power. Indeed, there is vociferous debate in the field, not just about the size of many of the effects included in…many…people's models but even their signs. Moreover, the dynamo equations are extremely nonlinear; the solar dynamo is believed to exist in a state of deterministic chaos, making prediction intrinsically yet more difficult. Any predictions made with such models should be treated with extreme caution (or perhaps disregarded), as they lack solid physical underpinnings.”

32. The Yohkoh SXT Data Does the activity turn on ‘instantly’? Saba, Strong, and Slater 72,000 full-disk thin Al filter images –Passband: 3 to 50 Å –Temperature: 1 to 50 MK Selection towards the quieter times (<C-level flares) Complete, calibrated, aligned, despiked, and background subtracted Sum all pixels to get total X-ray flux Yields: one datum per image Yohkoh Full-disk Image Taken in the Thin Aluminium Filter

33. How Much Does the X-ray Sun Vary? 27-Day Running Average Minimum

34. GOES XRS Confirms the SXT Result 5-min Averaged 1 – 8 A Data Found Minimum Flux for Each Day –Selects against flares Found Similar Step as Seen by Yohkoh SXT –Steps are simultaneous –XRS significantly “harder” than SXT

35. Most of the New Activity Originates in the Active Region Belt

36. Weak Fields Show Gradual Increase

37. Flare Rate Increases Sharply at Step Total Number of Flares (C + M + X) Weighted Number of Flares (C + 10*M + 100*X) In the 7 months before the step: 30 flares In the 7 months after the step: 430 flares

38. No Old Cycle Regions Emerge After Step Northern Hemisphere Dominant (55%)

39. Is it evident in Sunspot Statistics? Monthly averaged sunspot area shows the step Sunspot Number Less Prominently

40. A Prediction Look at previous cycles –Available data types more limited –Less coverage –Calibration less well established The initial indications: Onsets of Cycles 21 and 22 also show similar steps, about 140 CRs apart –Predict Next Step: Jan 2008 (± 2 months) Cycle 22 Cycle 23 GOES 1-8 Å Minimum Daily Flux with 1-Month Boxcar Smooth 140 CRs (10.45 Years)

41. The Cycle 23 Prediction Panel Joselyn et al., EOS 78, No. 20, 1997 TechniqueLow EndHigh End Even/Odd165235 Precursor140180 Spectral135185 Recent Climatology 125185 Neural Networks 110170 Climatology75155

42. The Cycle 23 Prediction “…the panel of 12 scientists…agreed that a large amplitude solar cycle with a smoothed sunspot maximum of approximately 160 is probable…” –Observed SSN = 120.8 “The smoothed cycle maximum (before the date for the Cycle 23 minimum is confirmed) is predicted to occur in March 2000, within the range of January 1999 to June 2001” –Observed Maximum Date: March 2000

43. Summary The predictions for Cycle 24 are just as disparate as those for Cycle 23, but the Cycle 23 consensus was for a LARGE cycleThe predictions for Cycle 24 are just as disparate as those for Cycle 23, but the Cycle 23 consensus was for a LARGE cycle –Cycle 24 predictions from 50-160 Solar cycle predicting is growing ever more sophisticatedSolar cycle predicting is growing ever more sophisticated –This dynamo model must undergo independent tests to confirm its abilities and to determine the effects of actual variations in the meridional flow speed on predicted amplitudes It’s not apparent that any of the existing models will predict a Maunder Minimum and/or restart the system to recover from Maunder MinimumIt’s not apparent that any of the existing models will predict a Maunder Minimum and/or restart the system to recover from Maunder Minimum The solar cycle panel will hope to achieve a consensus by April, 2007The solar cycle panel will hope to achieve a consensus by April, 2007

44. Oh, and one more thing Request for solar cycle 24 predictions, both serious and ‘fun’ Please submit by September 9, 2006 to guarantee consideration. –For those without a specific prediction model, we’d still like your prediction. The chair of the panel promises to do something fun with the predictions. Fun for a physicist, that is. There might be a prize in it, or at least some notoriety. The chair will just need a long memory. –E-mail predictions, no later than September 9, 2006 to Douglas Biesecker (doug.biesecker@noaa.gov) For those submitting a ‘fun’ prediction (one prediction per person) Prediction for the peak, smoothed SSN for solar cycle 24 Prediction for the month and year of the peak, smoothed SSN