The Fine Structure of Active Regions and Weak Magnetic...

The Fine Structure of Active Regions and WeakMagnetic Fields from MDI Images

Judit M. Pap

Goddard Earth Sciences and Technology Center, University of Maryland BaltimoreCounty / NASA Goddard Space Flight Center, Code 671.0 [email protected];

301-286-7511

A Living With a Star Group Science MeetingBoulder, ColoradoJuly 22-23, 2008

Proposal Team

1. Principal Investigator: Dr. Judit M. Pap (UMBC/GSFC)

2. Co-Investigators: Dr. William D. Pesnell (GSFC) andDr. Luca Bertello (UCLA)

3. Collaborators: Dr. Jerry Harder (LASP/CU) and Dr. RogerUlrich (UCLA)

4. Research is a collaborative effort with Dr. Gary Chapman andDora Preminger, CSUN/SFO

Fundamental Problems to be Understood – Goals:

1. Differences between Solar Activity Indeces and Irradiance atthe Maximum of Solar Cycle 23

2. Spectral Distribution of Total Solar Irradiance Variations

3. Detection and Parametrization of Weak Magnetic FieldsOutside of Active Regions

4. Origin and Evolution of Sunspots, Active Regions, andActivity Complexes

5. Energy Balance of Active Regions

Science Rationale – Overview

The most important environmental problems facing humanitytoday is to understand and predict global change (both natural andman-induced) as well as the rapid changes in our spaceenvironment.

CRITICAL ISSUE: What is the relative impact of natural andanthropogenic influences on changes in the Earth’s atmosphere?

REQUIREMENT: Accurate long-term total and spectrally resolvedsolar irradiance measurements are required to fully understand theresponse of the Earth’s atmosphere and climate to irradiancechanges.

Overview – Continued

UNDERSTANDING 1: The three-decade long irradiancemeasurements convinced the skeptics that solar irradiance (bothtotal irradiance and spectral irradiance at various wavelengths)varies on various time scales: from minutes to the 11-year solarcycle.

UNDERSTANDING 2: Similar to solar irradiance variations, othersolar-type of stars also show changes in their luminosity.


PROBLEM 1: The time period of interest far exceeds the lifespanof any single experiment, thus composite irradiance time seriesmust be compiled from data of several irradiance experiments.

PROBLEM 2: On time scales longer than the almost three-decadelong irradiance measurements, surrogates (magnetic activityindices) for irradiance have to be used to mimic the observedirradiance changes.


QUESTION 1: How well these indices used for irradiance modelingare reliable for predicting long-term irradiance changes?

QUESTION 2: What are the underlying mechanisms of irradiancevariations and to what extent surface magnetic activity and towhat extent global solar changes, like temperature or radiuschanges, contribute to irradiance variations?

QUESTION 3: What new tools we have/need to better predictsolar irradiance variability than the currently available empiricalmodels?

Total Solar Irradiance (TSI) – Measurements and Composites

To study the long-term variations and possible climate effect ofirradiance variations, two composites have been compiled: theACRIM composite and the PMOD composite (PMOD versiond41 61 0510 is shown here).

Willson and Mordvinov (2003) adjust the ACRIM I and ACRIM IIdata via their mutual inter-comparison with the Nimbus-7/ERBmeasurements, using TSI data as published by the originalinstrument teams. This composite shows a small (0.05%) increasefrom the minimum of solar cycle 21 to the minimum of cycle 22.

In contrast, Frohlich and Lean (1998) make adjustments to theNimbus-7 and ACRIM I data for 1980 and to the Nimbus-7 databetween 1989 and 1991, arguing about instrumental drifts andimproper correction for degradation effects on the ACRIM I andNimbus-7 data at the beginning of their operation.

Total Solar Irradiance (TSI) – MeasurementsTotal Solar Irradiance: Original Data (top) and Composite (bottom)

1360

1365

1370

1375

56

Total Solar Irradiance: Original Data (top) and Composite (bottom)

1360

1365

1370

13750 2000 4000 6000 8000

Days (Epoch Jan 0, 1980)

TIM/SORCE vers: 5

HF

ACRIM I

ERBS

ACRIM II vers: 101001

ACRIM III vers: 0503DIARAD/VIRGO 0605

VIRGO 6_001_0605

a) Original Data

0.3%

adapted and updated from: Quinn and Fröhlich, Nature, 401, p.841, 1999, with composite (vers d41_61_0605), ACRIM−II/III (vers II:101001) and VIRGO 6_001_0605 data (May 24, 2006)

Year

1364

1366

1368

56adapted and updated from: Quinn and Fröhlich, Nature, 401, p.841, 1999, with composite (vers d41_61_0605), ACRIM−II/III (vers II:101001) and VIRGO 6_001_0605 data (May 24, 2006)

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 Year

1364

1366

1368

b) PMOD Composite

Tot

al S

olar

Irr

adia

nce

(Wm

−2 )

1980 1985 1990 1995 2000 20051362

1363

1364

1365

1366

1367

1368

1369

Year

TS

I @ 1

AU

(w

/m2)

ACRIM Composite TSI Time Series (Daily Means) *

TSI trend between minima of solar cycles 21 − 23: + 0.04 %/decade

ACRIM Composite:Uses Nimbus7/ERB, ACRIM1, 2 & 3 resultsNimbus7/ERB comparisons bridge the ’ACRIM Gap’Results reconciled to ACRIM3 scale

Fractional components of ACRIM Composite:ACRIM: 87.4 %Nimbus7/ERB: 12.6 %

Nimbus7/ERB ACRIM1 Nimbus7/ERB ACRIM2 ACRIM3

Solar Cycle 21 Solar Cycle 22 Solar Cycle 23

RC Willson, earth_obs_fig9 12/01/2005* Willson & Mordvinov, GRL, 2003

Solar Activity Indices for Empirical Models

The Mg II h & k core-to-wing ratio derived from the vicinity of280 nm (as derived at NOAA/SEC, Rodney Viereck). This index isconsidered as a good proxy for faculae

Ca II K-line observations are available from several observatories,providing a good indicator for chromospheric activity

The 10.7 cm radio data which are available since 1947 andconsiderable effort has been devoted to extend irradiance modelsusing this index

The sunspot number (SSN), which is the longest solar time seriesavailable for solar modeling (other than the cosmogenic isotopes)

Full disk magnetic field strength data from the National SolarObservatory at Kitt Peak

Variations in the Activity IndicesMg II Core-to-Wing Ratio (Mg c/w)

0 1500 3000 4500 6000 7500 9000 10500Day Number 1 = November 7, 1978

0.26

0.265

0.27

0.275

0.28

0.285

0.29

0.295

0.3

Mg

II h&

k co

re−

to−

win

g ra

tioMg c/w Ratio and RCs (WL=3000)

Mgc/wRCs1&2RCs1&2&3RC3

Variations in the Activity Indices – ContinuedCa K line

0 1500 3000 4500 6000 7500 9000 10500Data between January 1, 1977 and June 10, 2005

0.075

0.08

0.085

0.09

0.095

0.1

0.105

0.11C

a K

Ful

l Dis

k In

dex

Kitt Peak Ca II K Index & 365−day Running Means

Variations in the Activity Indices – ContinuedF10.7 Time Series

0 1500 3000 4500 6000 7500 9000 10500Data between January 1, 1977 and September 30, 2005

0

50

100

150

200

250

300

350

40010

.7 c

m R

adio

Flu

x

10.7 cm Radio Flux and RCs (WL=3000)

F10.7RCs1&2RCs1&2&3RC3

Variations in the Activity Indices – ContinuedSunspot Number Time Series

0 1500 3000 4500 6000 7500 9000 10500Data between January 1, 1977 and September 30, 2005

0

40

80

120

160

200

240

280

320S

unsp

ot N

umbe

rSunspot Number and RC Trends (WL=3000)

Daily ValuesRCs1&2RCs1&2&3RC3

Variations in the Activity Indices – Continued

Analyses based on the Mg c/w ratio and the CaK index have led tothe conclusion that the long-term variations in total irradiance aremostly related to the bright features, such as faculae and thenetwork.

However, significant differences exist between these indices. Forexample, the maximum of the Mg c/w ratio is somewhat lowerduring the declining phase of solar cycle 23. In contrast, the CaKline shows a much lower maximum, and it is far from its minimumlevel during the declining phase of cycle 23.

Although the overall shape of F10.7 and SSN is similar, themaximum amplitude of F10.7 is much higher at cycle 23 relative tothe previous cycles than that of SSN.

As the long-term trend components show, while SSN does not yetshow solar minimum, F10.7 - similar to TSI indicate that we areclose to minimum.

Solar Magnetic Indices

Since 1975 full disk magnetograms are taken to produce twonumbers for each one: the average field strength (sum of all theindividual measurements divided by the number of measurements)and the average absolute field strength (same but take absolutevalue of the measurements first). Major changes occurred in 1992and 2003 when the instruments changed. Minor changes occurredat other times when software changed.

The KPVT observations were carried out in the 868.8 nm startingin 1977. For day numbers between 8146 and 8357, a differentspectrum line (550.7 nm) has been used. After day number 8357(11/30/03), the new SPMG was in use with the 868.8 nm line.

Solar Magnetic Indices – Continued

0 1500 3000 4500 6000 7500 9000 10500Data between January 7, 1977 and November 2, 2005

0

5

10

15

20

25

30

35

40

Mag

netic

Fie

ld S

tren

gth

(Gau

ss)

KPVT and SOLIS Magnetic Field Strength and RCs (WL=3000)

Magn Field CompositeRCs1&2RCs1&2&3RC4

Solar Magnetic Indices – Continued

As estimated, the magnetic field values are all on the same scale(this is probably correct to of order 10%). As can be seen, themaximum of cycle 23 was far lower than the last cycle – despitethe fact that solar irradiance was almost as high as during the laststrong cycles. By mid-2005, the KPVT data were close tominimum condition, and the SOLIS data indicate that we arealready at solar minimum conditions.

However, in the case of the absolute values of the measured fieldcan be at error due to various effects (error in the zero point level,image resolution, instrumental noise). Thus, further studies arerequired to confirm whether the low level of the absolute magneticfield strength during the declining portion of cycle 23 is entirely (orto what extent) due to the Sun.

Spectral Distribution of Total Irradiance Variations

1. UV irradiance is quite well measured, and it is estimated thatas much as 30 to 60% of total irradiance variations may comefrom UV variations

2. However, the variations above 300 nm are not well understood(or measured from space over a long time interval)

Spectral Distribution of Total Irradiance Variations – Continued

By courtesy of Dr. Jerry Harder

Analysis Techniques: Image and Time Series Decomposition

1. Image Processing: StarTool, developed within SOHO/GIprogram (Judit M. Pap, Michael Turmon, UCLA/JPL/UMBCeffort), based on Bayesian Technique, developed forSOHO/MDI, to be applied for HMI

2. Time Series Analyis: VARKIT (Variability Analysis andReduction ToolKit), developed by Dr. Ferenc Varadi (UCLA),co-worker: Dr. Judit M. Pap

3. Weak Field Recognition: Based on Instrument CalibrationCurves, originally developed for Mt. Wilson Magnetograms at535.5 nm by Roger Ulrich (UCLA) and to be applied to MDIby Judit M. Pap and Luca Bertello

Image Processing

1. StarTool, developed specifically to analyzing MDI images, usesboth full-disk (10242 pixels) magnetograms (level 1.5, about 15per day), and quasi-white-light photograms (level 2.0, about 4 perday).

2. It handles image preprocessing, synchronization, and featureregistration. Recognized that analyzing magnetograms alone is notsufficient to correctly recognize sunspots and faculae sincemagnetic field values between 25 and about 400 Gauss in MDI unitmay cause either sunspots or faculae (Turmon, Pap, Mukhtar,2002).

Identified three specific structures: sunspots (umbra andpenumbra), faculae, and background (the so-calledquiet-Sun/network).

StarTool Sample:

Synchronized full disk magnetogram and photogram, with scatterplots of sites within the two indicated regions. The top two panelsare details from a magnetogram and a photogram; the photogramwas taken at 17:58 UTC on September 7, 1997 and preprocessedto remove various instrumental effects; the magnetogram has beensynthesized from bracketing one-minute magnetograms taken at6:24 and 19:11 UTC on that day.

The selected regions are “255×725”. The scatter plots of data forthe two indicated subregions are in the bottom panels. The bipolarspot (left panel) shows clearly as having lower intensity, and thepenumbra is visible at the knee of the plot. Most encouraging, thebrightness enhancement of the decayed facula (right panel), farfrom the limb at µ = 0.81, is clearly visible, increasing withmagnetic flux strength.

StarTool Sample – Continued:

The maximum brightness enhancement amounts to about 3-4% ofthe nominal intensity level. They also show how various regiontypes, here faculae and sunspots, may be identified on the basis ofthe joint observables, but not either alone.

MDI Magnetograms and PhotogramsStarTool Intensity and Magnetic Field Relations

−250 −200 −150 −100 −50 0 50 100 150 200960

970

980

990

1000

1010

1020

1030

1040

1050

−1500 −1000 −500 0 500 1000 1500200

300

400

500

600

700

800

900

1000

1100

Application of Startool to Irradiance Time Series:

t

0 250 500 750 1000 1250 1500 1750Data between May 19, 1996 and September 28, 2000

−4

−2

0

2

4

6N

orm

aliz

ed T

ime

Ser

ies

SOHO/VIRGO and MDI ResultsTotal Irradiance and Spot, Faculae Area

VIRGO TSIMDI Spot AreaMDI Faculae Area

Application of Startool to Irradiance Time Series:

t

0 250 500 750 1000 1250 1500 1750Data between May 19, 1996 and September 28, 2000

−2

−1

0

1

2

3N

orm

aliz

ed T

ime

Ser

ies

(AR

bitr

ary

Uni

ts)

SOHO/MDI ResultsUARS/SUSIM Mg c/w Ratio and Faculae Reconstruction

SUSIM Mg c/wMDI/Faculae Area

Shortcomings of Startool:

(1.) With the current version of StarTool and using the full diskMDI images, the weak magnetic network structures cannot beidentified well, and these weaker structures are classified into thequiet Sun or background structures.

(2.) To identify and characterize the weak network magneticcomponents, as a first step we need to properly define and identifythe quiet Sun values.

(3.) For this purpose, we apply the results gained from the“standard instrument distribution curve” technique – as used toanalyze the Mt Wilson magnetograms (Ulrich et al., 2002.)

Standard Instrument Distribution Curves:

1. Since StarTool uses both the magnetograms and the intensityimages to extract features from the MDI images, these“standard instrument distribution curves” need to be derivedfrom the continuum images as well.

2. In this way, we can derive the necessary parameters for thequiet Sun values from the high resolution MDI images andthese parameters, together with the currently availableparameters for faculae and sunspots, would define the MDIweak field structures.

3. Technically, the weak field parameters will be determined bythe maximum magnetic field and intensity values of theinstrument distribution curves and the minimum values for thealready identified faculae.

Standard Instrument Distribution Curves –Continued:

1. Our approach: reduce the uncertainty and/or minimize theeffect of the zero-offset of the images.

2. Since both the zero offsets and the uncertainties are changingfrom exposure to exposure, our goal is to minimize theuncertainties (for example by 50 × 50 pixels binning) and toderive an averaged, constant zero offset.

3. Apparently, for this purpose we need to rely on the highresolution images. To derive a “constant” zero-offset, wepropose to select those exposure times when the Sun wasquiet and make the distribution curves of all the observablepixels, which are expected to be Gaussian.


(4.) Since both the uncertainty and zero-offset vary from exposureto exposure, we expect that these Gaussian curves will change alsoslightly. However, from these various exposure curves we will beable to derive an averaged curve that we can use to represent thedistribution curve of the quiet Sun or better to say, the distributioncurve of the instrument (called a “standard instrument distributioncurve”).

(5.) Because changes occurred in the MDI system during the2-month long “SOHO-vacation” in the summer of 1998, wepropose to derive and compare these curves from the observationsbefore and after the SOHO vacation.

(6.) It is still an open question whether or not the quiet-Sun isconstant over time so it is prudent to calculate these averaged,standardized distribution curves from time to time, and ifnecessary, to make adjustments.


(7.) This procedure would ensure that we set up a limit, althoughartificial but consistent, for the MDI pixel distribution due toinstrumental effects.

(8.) We propose to save and archive all of the averagedinstrumental distribution curves derived within various phases ofthe solar cycle, which may give us further information whether theso-called quiet-Sun is changing over the time.

Mt Wilson Curves:

– 1 0 – 5

B (gauss)

0 5 1 0– 4

– 3

log

(φ

)

– 2

– 1

0

Instrumental

1 2 / 2 5 / 9 8

6 / 7 / 9 6

By courtesy of Dr. Roger Ulrich

Mt Wilson Results:

1 9 8 5 1 9 9 0 1 9 9 5 2 0 0 0

1.7

1.8

1.9

2.0

2 . 1

0

50

100

150

200

SSNSSN

Sun

spot

Num

ber

Sun

spot

Num

ber

Date

By courtesy of Dr. Roger Ulrich


The above mentioned procedure has two applications:

1. Relevance to HMI: Since the HMI instrument on SDO is animproved version of MDI, and the HMI image resolution will beequal to the resolution of the high resolution MDI images (1.3 arcsec), the proposed calculation of the “standard instrumentdistribution curve” could guide also the development of the HMIcalibration factors and procedures.

2. Relevance to Current MDI Image Processing: The abovedescribed method will significantly improve the “StarTool” imageprocessing algorithm that was developed by Turmon and Pap(1997) to analyze the MDI images, and was described in detail byTurmon, Pap, and Mukhtar (2002).

Conclusions

While the sunspot number and magnetic field strength show thatthe maximum of solar cycle 23 is lower than the maxima of thetwo previous cycles, the maximum level of total irradiance, theMg II h & k core-to-wing ratio, He 1083, and F10.7 flux is higherduring solar cycle 23 than the maximum of the magnetic indices.

Further discrepancies can be found between the level of totalirradiance and various activity indices during the declining phase ofcycle 23.

Additional interesting events are the consecutive large activityoutbursts - during the declining phase of the weak cycle 23.

Conclusions – Continued

These results highlight that during solar cycle 23 solar activityindices, which describe magnetic features, cannot properly accountfor solar variability, as measured by solar irradiance; thus theseindices used in long-term irradiance variation may not account forlonger term changes.

Solar cycle 23 is the first measured example of the non-linearrelationship between solar activity (local magnetism) and solarvariability (changes in solar global properties).

Conclusions – Research to be Conducted:

Problems with indices used in irradiance models:

1. Effect of Sunspots: A simple model (Photometric SunspotIndex, PSI) has been developed to describe the effect ofsunspot darkening on total irradiance

(a) However, we lack long-term high precision sunspot areameasurements

(b) The dependence of the contrast change of sunspots overtheir evolution and the solar cycle, and effect of magneticconfiguration has not been included in PSI


2. Effect of Faculae: We do not have good faculae models

(a) In many cases, full disk and chromospheric indices areused to describe the effect of photospheric faculae

(b) From the MDI and Kitt Peak observations new effortshave been made to identify faculae

(c) However, the area measurements differ - and the faculaecontrast change is not taken into account


3. Effect of Network: We do not have good network models

(a) Network is estimated from full disk indices

(b) Attempts to combine continuum and Ca K images havebeen made to identify the network - these again describemostly the chromospheric network

(c) Further, we do not know whether the ”quiet-Sun” or theweak magnetic field are indeed quiet or they change overtime, thus contributing to long-term climate changes


4. Effect of Global Changes:

(a) Temperature changes: small changes in the photospherictemperature may explain the 0.1% solar cycle variations

(b) Results on radius changes are quite contradictive, we needto have (BOTH HMI and PICARD measurements)


Improvements in long-term solar irradiance models for climatestudies will require:

1. Long-term and better quality measurements;

2. Improvement of radiometry;

3. Improvement of the sunspot and faculae area and contrastmeasurements

4. Use of high resolution images to better account for the weakfield component which is excluded from the current empiricalmodels;

5. Most importantly, we need to understand the physics of allsurrogates and thus the origin of irradiance variations tobetter predict their effect on past and future climate changes.

Specific Science Tasks

Study high resolution MDI images to extract network and includeit to irradiance models (2nd and 3rd year)

Study the contrast changes of sunspots and faculae over time(active regions time scale and over cycle 23) (2nd year)

Include these parameters either into empirical or semi-empiricalmodels (synthetic models) (3rd year)

Study the energy budget of active region using more accuratesunspot/facular areas and their contrast (3rd year)

Systematic study of the evolution of active regions, particularlyones with δ spots (space weather application) (2nd year)

Apply results to current data (SOLIS, MDI, Solar-B) andforthcoming SDO measurements (3rd year + new proposal)

Active Regions Mosaic – AR Evolution

Tracked Active Region: Magnetogram, Photogram, Labeling1996 July 27 −− August 08

J 27

16:2

7J

3017

:56

A 0

203

:52

A 0

412

:39

J 27

16:2

7J

2716

:27

J 30

17:5

6J

3017

:56

A 0

203

:52

A 0

203

:52

A 0

412

:39

A 0

412

:39

The Fine Structure of Active Regions and Weak Magnetic...

Documents

Transcript of The Fine Structure of Active Regions and Weak Magnetic...