Reconciling the Sunspot and Group Numbers

Slides:



Advertisements
Similar presentations
Experimental Measurements and their Uncertainties
Advertisements

Design of Experiments Lecture I
Computational Statistics. Basic ideas  Predict values that are hard to measure irl, by using co-variables (other properties from the same measurement.
Calibration of Sunspot Numbers Leif Svalgaard Stanford University Eddy Symposium 22 Oct
Kriging.
Sixty+ Years of Solar Microwave Flux Leif Svalgaard Stanford University SHINE 2010, Santa Fe, NM.
Budapest May 27, 2008 Unifying mixed linear models and the MASH algorithm for breakpoint detection and correction Anders Grimvall, Sackmone Sirisack, Agne.
Long-term evolution of magnetic fields on the Sun Alexei A. Pevtsov US National Solar Observatory.
1 Reconciling Group and International Sunspot Numbers Leif Svalgaard, HEPL, Stanford University Edward W. Cliver, Space Vehicles Directorate, AFRL XII.
The new Sunspot Number A full recalibration Frédéric Clette World Data Center SILSO, Royal Observatory of Belgium L. Svalgaard Stanford University J.M.
A full revision of the Sunspot Number and Group Number records Frédéric Clette World Data Center SILSO, Royal Observatory of Belgium L. Svalgaard Stanford.
1 Confronting Models with Reconstructions and Data Leif Svalgaard HEPL Stanford University Boulder, June 2014 ESWE Workshop Presentation Session 9: Understanding.
Error bar estimates Estimates of the uncertainties on input LwN Estimates of the error model Estimates of the uncertainties on outputs Chla, bbp,cdm (Co-variance.
Forschungszentrum Karlsruhe in der Helmholtz-Gemeinschaft NDACC H2O workshop, Bern, July 2006 Water vapour profiles by ground-based FTIR Spectroscopy:
Error Propagation. Uncertainty Uncertainty reflects the knowledge that a measured value is related to the mean. Probable error is the range from the mean.
Polar Network Index as a magnetic proxy for the solar cycle studies Priyal, Muthu, Karak, Bidya Binay, Munoz-Jaramillo, Andres, Ravindra, B., Choudhuri,
Lecture II-2: Probability Review
LINEAR REGRESSION Introduction Section 0 Lecture 1 Slide 1 Lecture 5 Slide 1 INTRODUCTION TO Modern Physics PHYX 2710 Fall 2004 Intermediate 3870 Fall.
1 The History of the Sunspot Number Leif Svalgaard Stanford University, California, USA AOGS, Singapore, August 2015 WSO.
Calibration of Sunspot Numbers Leif Svalgaard Stanford University SIDC Seminar 14 Sept
1 Something with ‘W’ Leif Svalgaard March 1, 2012.
1 Reconciling Group and International Sunspot Numbers Leif Svalgaard, HEPL, Stanford University Edward W. Cliver, Space Vehicles Directorate, AFRL NSO.
1 The Spots That Won’t Form Leif Svalgaard Stanford University 3 rd SSN Workshop, Tucson, AZ, Jan
Propagation Trends Dayton 2014 Solar Maximum! But the slow decline to solar minimum in 2020 is likely to begin later this year.
Chapter 16 Data Analysis: Testing for Associations.
1 Geomagnetic Calibration of Sunspot Numbers Leif Svalgaard HEPL, Stanford University SSN-Workshop, Sunspot, NM, Sept
NON-LINEAR REGRESSION Introduction Section 0 Lecture 1 Slide 1 Lecture 6 Slide 1 INTRODUCTION TO Modern Physics PHYX 2710 Fall 2004 Intermediate 3870 Fall.
1 How Well Do We Know the Sunspot Number? [And what we are doing to answer that question] Leif Svalgaard HEPL, Stanford University Poster at ‘Solar in.
1 The Waldmeier Effect and the Calibration of Sunspot Numbers Leif Svalgaard Stanford University, California, USA David H.
12/15/20141 The 2015 Revision of the Sunspot Number Leif Svalgaard Stanford University AGU Fall Meeting SH11D-07 See also poster at Monday, December 15,
1 Revisiting the Sunspot Number “Our knowledge of the long-term evolution of solar activity and of its primary modulation, the 11-year cycle, largely depends.
1 Objective Calibration of Sunspot Numbers Leif Svalgaard Stanford University, Stanford, CA, USA. AGU Fall 2009,
Renewing our view to past solar activity: the new Sunspot Number series Frédéric Clette, Laure Lefèvre World Data Center SILSO, Observatoire Royal de Belgique,
Validation/Reconstruction of the Sunspot Number Record, E.W. Cliver Air Force Research Laboratory, Space Vehicles Directorate, Hanscom AFB, MA.
QM2004 Version1 Measurements of the  ->     with PHENIX in Au+Au Collisions at 200 GeV at RHIC PPG016 Figures with Final Approval Charles F. Maguire.
1 Reconciling Group and Wolf Sunspot Numbers Using Backbones Leif Svalgaard Stanford University 5th Space Climate Symposium, Oulu, 2013 The ratio between.
What the Long-Term Sunspot Record Tells Us About Space Climate David H. Hathaway NASA/MSFC National Space Science and Technology Center Huntsville, AL,
1 Building a Sunspot Group Number Backbone Series Leif Svalgaard Stanford University 3 rd SSN Workshop, Tucson, Jan
1 Sunspots with Ancient Telescopes Leif Svalgaard Stanford University, California, USA John W. Briggs Magdalena, New Mexico,
Chapter Nine Hypothesis Testing.
Oscillator models of the Solar Cycle and the Waldmeier-effect
Non-LTE atmosphere calculations of the solar spectrum
Chapter 7 Process Control.
The IAU Working Group on Coordination of Synoptic Observations of the Sun Alexei A. Pevtsov US National Solar Observatory.
From: The evolution of star formation activity in galaxy groups
The Waldmeier Effect and the Calibration of Sunspot Numbers
Geomagnetic activity indicates large amplitude for sunspot cycle 24
Statistics in MSmcDESPOT
Instrumental Surface Temperature Record
(an incomplete attempt) Lidia van Driel-Gesztelyi
Research methods Lesson 2.
The Effect of Weighting and Group Over-counting on the Sunspot Number
Instrumental Surface Temperature Record
The Waldmeier Effect and the Calibration of Sunspot Numbers
“The need of revising the good old Wolf numbers”
A Re-analysis of the Wolfer Group Number Backbone, I
ISSI Team: Recalibration of the Sunspot Number Series
ADF (active-day fraction) method of sunspot number calibration
Research strategies & Methods of data collection
Probabilistic Population Codes for Bayesian Decision Making
Volume 111, Issue 2, Pages (July 2016)
by A. Dutton, A. E. Carlson, A. J. Long, G. A. Milne, P. U. Clark, R
The spectral evolution of impulsive solar X-ray flares
Fast Sequences of Non-spatial State Representations in Humans
Testing the Fit of a Quantal Model of Neurotransmission
Research strategies & Methods of data collection
Encoding of Stimulus Probability in Macaque Inferior Temporal Cortex
Solar Activity Forecast for the Next Decade
MGS 3100 Business Analysis Regression Feb 18, 2016
Rony Azouz, Charles M. Gray  Neuron 
Presentation transcript:

Reconciling the Sunspot and Group Numbers ROB STCE Reconciling the Sunspot and Group Numbers Frédéric Clette Laure Lefèvre World Data Center SILSO Royal Observatory of Belgium, Brussels

Multi-century solar activity reference Visual sunspot number Our only direct record of solar activity over the last 400 years Longest scientific experiment still ongoing (B.Owens, Nature, March 2013) 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Two sunspot number time series SN= 10 Ng + Ns Measure of the number of active regions + their size Origin: R. Wolf (1849) Time interval: 1700 - now Production: 1700-1849: reconstruction from historical documents 1849-1980: Zurich Observatory 1981-now: World Data Center SILSO, Brussels Calibration: pilot station Zurich Observatory (successive primary observers) Specola Observatory Locarno (since 1981) Sunspot group number GN= 20.13 Ng More basic but applicable to cruder early observations Origin: Hoyt and Schatten (1998) Time interval: 1610-1995 Production: Single recent reconstruction Based on an extended set of raw historical data Calibration: “Daisy-chaining” of observers backwards in time Starting reference: Royal Greenwich Observatory photographic catalog (1875-1975) Very good match after 1900 Large disagreements before the 20th century: GN lower than SN by up to 40% 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

A new impulse: Sunspot Number Workshops NSO, Sac Peak, USA, Sept. 2011 Community effort started in Sept. 2011: 4 Sunspot Number Workshops: > 40 participants ISSI Team Meetings 2018-2019 (www.issibern.ch/teams/sunspotnoser/): Chairs M.Owens, F.Clette Synthesis in: Clette, F., Svalgaard, L., Vaquero, J.M., Cliver, E.W.: Space Sci. Rev. 186, 35-103, 2014 Solar Physics: Topical Issue on « Recalibration of the Sunspot Number», Volume 291 9-10, 2016, Eds. Clette, Cliver, Lefèvre, Vaquero, Svalgaard, 35 articles ISSI Bern, Switzerland Jan. 2018 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Sunspot Number SN version 2 (released July 2015) Diagram “Waldmeier” jump (1947) Schwabe - Wolf transition (1849-1864) Locarno’s variable drifts (1981-2015) 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

“Backbone” Group Number GN (Svalgaard & Schatten, 2016) Maunder Minimum (1650-1710) Royal Greenwich Obs. Photographic catalog “Greenwich” trend (1885-1915) 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Two unrelated set of corrections Sunspot Number Type of flaws: Inaccuracies in the k scaling coefficients vs the pilot observer Drift of the single pilot station Change of counting method: weighting according to spot size Methodological approach: Statistics over all available auxiliary observers : Replacing the single pilot station by multiple stable long-duration observers Double counts: Quantifying the effect of the weighting of spots Correction factors applied to the original SN series Sunspot group number Type of flaws: Inhomogeneities in the photographic catalog after 1875 Backwards propagation of scaling errors: Daisy-chaining of k coefficient Methodological approach: Replacing the photographic data by multiple visual observers after 1875 Avoiding daisy-chaining: Long-duration reference observers (backbone observers) Active-day fraction (ADF): scaling factor deduced from individual spotless days statistics Full reconstructions from all raw data No attempt to make one series similar to the other one ! The problems have different causes and occur at different times The required correction methods and data were different and unrelated 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Sample corrections for SN and GN

[SN] The Waldmeier jump : probable cause 2011-10-12 Sunspot weighting: Large spots are counted >1 (up to 5) Introduced for Zurich assistants in the late 19th century (Friedli 2016) Systematic application by the primary observer since 1926 Locarno auxiliary station trained to the method (1955): still in use ! Blind test (2008 – 2014): comparison of simultaneous standard and weighted counts (Clette & Lefèvre 2016, Svalgaard 2017): Variable inflation factor Constant at high activity: 1.177 ± 0.005 Matches the amplitude of the 1947 jump 4 4 4 =1+3 6 =3x1+3 9 = 3x3 13 = 3x1+3x2+4 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

GN : criticisms and new results Original GN series (Hoyt & Schatten 1998): daisy-chaining of k ratios between observer pairs Backwards propagation of errors Personal k coefficient Error 1 Time 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

GN : criticisms and new results 40% upward drift attributable to the use of photographic data after 1875 (Royal Greenwich Observatory catalogue) Cliver & Ling 2016, Cliver 2017 Personal k coefficient 1 Time 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

[GN] An alternate approach: active-days fraction Statistics of spotless days versus days with one group of spots or more = active days (Usoskin et al. 2016) Model hypothesis: differences in groups counts due to acuity of observers Construction of a standard “perfect” observer: RGO catalogue (1900-1975) Acuity = groups eliminated below a lower threshold in sunspot area (SS) Matching the cumulative distribution function (cdf) of the number of active days/month (ADF) for the observer cdf of RGO with different thresholds SS Cdf for actual observer On the figure, maybe forget the “of ADF” in red, or put it in both … Resulting series: similar to the original Hoyt & Schatten GN (low in 19th century) 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

[GN ] Limits and failures of the ADF Spörer Observer sampling can be varying with solar activity: Strong positive bias on ADF (underestimate of GN) Willamo et al. (2018) The method works only when ADF is below 80%: Activity is below 5-6 groups (< 50% of peak of solar cycle) Derived scale is extrapolated for high activities (cycle maxima) Les points 1/2 est expliqué dans Willamo et al. 2018 (Table 1) It is even worse than that. It works mostly for moderate cycles…not for low cycles, not for higher cycles. I added the Willamo reference. Only 1 reported group Base assumption does not consider differences in group splitting between observers: Important factor near cycle maxima Staudacher, Feb. 13 1760 (Svalgaard 2016) 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Post-correction assessment

Impact: new secular trends Original series: strong upward secular trend over last 200 years (“Modern maximum”, Solanki et al. 2004, Usoskin 2013): GN: + 40% / century (red) SN : + 15% / century (green) New SN and GN are similar and have a weak upward trend < 5 %/century 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

External validation: geomagnetic record Comparison with the geomagnetic record: solar open magnetic flux B reconstructions Latest joint re-calibration (ISSI workshops) (Owens et al. 2016) No trend between cycle maxima of mid-19th century and mid-20th century Best match with SN version 2.0 Owens et al. 2016 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

SN and GN remain distinct indices Different measurements of emerging toroidal magnetic flux Ratio between SN and GN values for the re-calibrated SN V2 series over 1945-2015 (Clette & Lefèvre 2016): Single non-linear relation valid for multiple cycles Consequence of the varying contribution of large and small sunspot groups (Tlatov 2013, Georgieva et al. 2017 ) S N =17.11 ±0.13 G N 1.07(±0.010) 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Building a full sunspot database

Key action: SN observation database Next SN version (V3): Recalculation from all available raw source data Recovery of personal logbooks, printed tables, drawings Staudacher, Feb. 13 and 15 1760 (Svalgaard 2016) Wolf’s sourcebooks E. Spée, 1895, ROB, Brussels 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Key action: SN observation database Recovering all raw input data from Zurich (internal + auxiliary stations) (Astronomische Mittheilungen der Sternwarte Zurich): Now encoded up to 1945 (WDC-SILSO) Not all data published after 1919 (external stations missing) No data published after 1945 Original sourcebooks recovered at the Specola Observatory (Locarno) in June 2018: all source data between 1945 and 1970 WW I WW II Unpublished (Zurich archives) Published 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Remaining issues and next goals Sunspot Number Status: The main corrections are included in Version 2 Consensus on the amplitude of the Waldmeier jump correction Remaining improvements: Modern period (1849-today) Mostly small local deviations: < 10%, less then 1 solar cycle Early period (18th and early 19th century): Lower accuracy and sparse data Ongoing database construction Recovery of lost Zurich sourcebooks Sunspot group Number Status: Consensus on flaws in the original H & S series Several incompatible reconstructions Flaws identified in all new methods Ongoing method evaluation: Common new GN database (Vaquero et al. 2016) Coordinated testing of methods: Joint work: ISSI workshops Focused topical working groups Goal: Consensus for each separate issue Single optimal reconstruction Corrected “of different strategies” Full end-to-end reconstruction from all original data Combination of different methods: Best method for each problem and epoch 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

VARSITI General Assembly, Sofia, Bulgaria Conclusions The study of the past sunspot record is completely revived SN and GN were and will be calibrated independently The SN and GN time series are now evolving data sets! 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

VARSITI General Assembly, Sofia, Bulgaria Stay tuned …. World Data Center – SILSO Sunspot Index and Long-term Solar Observations http://sidc.be/silso 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

VARSITI General Assembly, Sofia, Bulgaria 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

GN: new database and Maunder minimum Full revision of the original GN raw data archive (Vaquero et al. 2016) Many null values unduly interpolated by Hoyt & Schatten (1998) SV16 Higher activity levels in Maunder minimum: a weak solar cycle persists (South hemisphere) GN « Original », Hoyt & Schatten 1998 GN « Backbone », Svalgaard & Schatten 2016 GN Vaquero et al. 2015 A&Ap ML « Loose » model MO: « Optimum » model MS « Strict » model 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

SN criticisms: narrowing in on a common value Comparisons with external indices Svalgaard 2013 Lockwood et al. 2014 Rg Without SOON Rg Ag Clette & Lefèvre 2016 Rg & Ag Lockwood et al. 2016 foF2 Lockwood et al. 2016 Rg Lockwood et al. 2016 multi Direct counts Clette & Lefèvre 2016 daily SN Clette & Lefèvre 2016 mean cycle 23-24 Svalgaard 2017 monthly SN 1.16 1.18 1.20 1.10 1.12 1.14 Distinction between mean ratio : 1.14 +/- 0.02 and peak ratio : 1.177 +/- 0.005 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

VARSITI General Assembly, Sofia, Bulgaria [SN] Zurich observers Old team members (W.Brunner + assistant Brunner): Long-duration contributors since 1926 Continue to observe until late 1947 New team members: Recruited by Waldmeier starting in 1948 No overlap with pre-1948 observers First, short-duration observers (rapid changes) (Plot based on the Mitteilungen der Eidgenösisches Sternwarte Zürich) 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

[SN] Long-duration stations (> 1 solar cycle) Pre-WWII long-duration observers stop over 1938-1945 During the war: “emergency” network of local Swiss observers (mostly short duration) After WWII: progressive reconstruction of a new international network (many new stations: Arcetri, Kanzelhöhe, etc.) (Plot based on the Mitteilungen der Eidgenösisches Sternwarte Zürich) 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

[GN] Uncorrected series: a comparison Ratios with a non-calibrated series (raw un-normalized numbers) Cliver 2016 High values before 1900: SN V2, backbone GN, Cliver&Ling GN Low values before 1900: original H&S GN, ADF GN Constant or rising ratio (low series) imply a constant or degrading quality of the observations Inconsistent with known progresses of astronomical instruments 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Non-Gaussian errors and uncertainties Global statistics based on the SILSO database (1981-now, > 550.000 data) Two components in random errors (Dudok de Wit et al. 2016): Square root Observational errors Observational error : gaussian Solar activity: poisson Linear Solar activity + slow trends and individual biases Dudok de Wit et al. 2016 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Non-Gaussian errors and uncertainties Long-term variations of observer errors: decreasing dispersion in the resulting SN Steady improvement of the SN index precision Dudok de Wit et al. 2016 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

The Wolf-Wolfer transition (1877-1893) Unique interval in the SN series: SN number is the average between two observers (Wolf+Wolfer) Critical double transition: From Wolf 40mm portable refractor to Wolfer with 82mm refractor New counting rules: small spots, multiple umbrae Wolfer 1895 k coefficient 0.6 between Wolf series (1700-1893) and modern series (1893-now) Trend over 1876-1883: Wolfer gaining experience (counting progressively more spots) Mix with other assistants (mutual influence?) 1883-1893 gives a correction factor of about 0.55 (< 0.6) Should the Wolf series before 1876 be raised by 10%? 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

[GN] Improving the backbone method New “backbone” reconstruction (Chatzistergos et al. 2017): More backbone observers: directly overlapping Use of daily values instead of yearly mean values Non-parametric scaling between observers : correspondence matrices Cross-observer probability distribution functions Provides the means and dispersion of the estimated corrected value (panel b) Intermediate scale in 19th century: Lower than original backbone GN Higher than ADF GN Chatzistergos et al. 2017 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

[GN] Towards a solution ? Correspondence matrices: nice non-parametric tool Very few data points at moderate and high GN values: Lower slopes ( k factors) or non-linearity of fits are barely significant Compromise: a linear relation, but with large error bars for the higher regime. Left Fig.: Comparison of Wolf and Wolfer counts on their overlapping period. The large grey area is +/- 3 sigmas from the mean (what is called low stdv, high stdv) The orange dashed line is a representation of the linear relation computed by Svalgaard on yearly values. The other fits are based on the square/circle/triangle points without error bars (pink), or with error Bars (1 sigma, blue, green, grey). Fig. Blue: Example of the construction of the calibration matrix for Winkler (secondary observer, G) to RGO (primary, G) over 1900–1910. Panel (a) shows the original distribution matrix G vs. G: the black line has a slope of unity. The red circles with error bars depict the mean G values for each G column and their 1 sigma uncertainty. The yellow line shows the k−factor used in Hoyt & Schatten (1998). Best approach: combination of classical k-ratios and correspondence matrices 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Future big challenge: continuity in early data Optimal choice of the reference- observer chain (A. Muñoz-Jaramillo 2018): linking best observer pairs Recovering new “forgotten” observations: e.g. East Asia (Hayakawa et al. 2019) Advanced data-mining methods for sparse time series Exploitation of detailed information in sunspot drawings Use of geomagnetic indices to bridge short gaps to link “loose ends” Dudok de Wit et al. 2017 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Wolf’s historical observers: timeline Historical data used as quantitative measurements ! (counts) Not just as descriptive accounts 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

Two distinct base data sets Sunspot Number Reference data sets: Zurich observers and auxiliary stations before 1980 SILSO database 280 stations > 550,000 numbers + a few new recovered time series (1950-2015) Archived reports to Zurich Sunspot group number Reference data sets: Original Hoyt & Schatten group number database (1610-1995) Extension with new observations: 20th century (Wolfer, Koyama, Luft, etc.) SILSO database (1980-2014) Corrections and extension of early historical data, including the Maunder Minimum Vaquero et al. 2016 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria

[GN] Towards a solution ? A compromise between old stitching methods and correspondence matrices Compromise: a linear relation, but with large error bars for the higher regime. Left Fig.: Comparison of Wolf and Wolfer counts on their overlapping period. The large grey area is +/- 3 sigmas from the mean (what is called low stdv, high stdv) The orange dashed line is a representation of the linear relation computed by Svalgaard on yearly values. The other fits are based on the square/circle/triangle points without error bars (pink), or with error Bars (1 sigma, blue, green, grey). Fig. Blue: Example of the construction of the calibration matrix for Winkler (secondary observer, G) to RGO (primary, G) over 1900–1910. Panel (a) shows the original distribution matrix G vs. G: the black line has a slope of unity. The red circles with error bars depict the mean G values for each G column and their 1 sigma uncertainty. The yellow line shows the k−factor used in Hoyt & Schatten (1998). 14/6/2019 VARSITI General Assembly, Sofia, Bulgaria