Peter MacNeice NASAGSFC

1
SOLAR MAGNETOGRAM SYNTHESISA Vital Component in
Space Weather Forecasting

• Peter MacNeice (NASA/GSFC)
• Joel Allred (Drexel Univ.)
• Kevin Olson (Drexel Univ.)
• Sandro Taktakishvilli (NPP)
• Marlo Maddox (NASA/GSFC)
• AISRP Workshop May 5-7, 2008
• Peter.J.MacNeice_at_nasa.gov

2
My Talk in One Slide

• Space Weather is hot!
• Forecasting model chains start with solar
  magnetograms
• New generation of models will demand much more
  from the magnetogram data
• No single magnetogram source can satisfy them
• Need to synthesize magnetograms from multiple
  sources !
• Who is going to create the infra-structure for
  this ? Modelers ? Observatories ? No ! - We
  will !
• Developing GUI driven CAD-like tool
• First check arrived two weeks ago.

3
Space Weather Primer

• Sun is the source of all transients driving space
  weather
• Most severe cases - Highly stressed coronal
  magnetic fields relax explosively
  Flares/Coronal mass ejection

• Resulting hazards
• Prompt radiation (8 minutes)
• Fast particles (e- gt 30 mins, p gt 60mins)
• CME shock driven particles ( gt 12 hours)
• Mass Ejecta (gt18 hours)
• Resulting magnetic storm

Credit SOHO Project
4
Societal Impact of Space Weather

• Power Grid failures
• Blackouts (eg Quebec, Mar 13 1989)
• long term, if high voltage transformers damaged
• Satellite failures (and over long term, reduced
  lifetimes)
• Communication and GPS blackouts
• Particle hazards to astronauts and polar flight
  passengers

5
Space Weather Primer (contd)

• Worst Case Scenario Carrington event, Sept
  1,1859
• Aurora in Havana
• No solar event of comparable magnitude in the
  technological era, by factor of 4 !
• Ice core records suggest one Carrington like
  event or bigger impacts Earth every 500 years.

Shea et al 2006
6
Space Weather Primer (contd)

• Carrington Event, Sept 1,1859
• Telegraph Disruption
• Observations made at Pittsburgh, Pa.,
  communicated by E.W. CULGAN, Telegraph manager.
• During the Aurora of Aug. 28th the
  intensity of the current evolved from it varied
  very much, being at times no stronger than an
  ordinary battery, and then suddenly changing the
  poles of the magnets it would sweep through them,
  charging them to their utmost capacity, and
  compelling a cessation of work while it
  continued. On the morning of Sept. 2nd, at my
  request the Philadelphia operator detached his
  battery, mine being already off. We then worked
  with each other at intervals as long as the
  auroral current continued, which varied from
  thirty to ninety seconds. During these working
  intervals we exchanged messages with much
  satisfaction, and we worked more steadily when
  the batteries were off than when they were
  attached. On the night of Aug. 28th the
  batteries were attached, and on breaking the
  circuit there were seen not only sparks (that do
  not appear in the normal condition of a working
  line) but at intervals regular streams of fire,
  which, had they been permitted to last more than
  an instant, would certainly have fused the
  platinum points of the key, and the helices
  became so hot that the hand could not be kept on
  them. These effects could not have been produced
  by the batteries. Boteler(2006)
• Estimates 70 billion impact on satellite
  industry (Odenwald et al 2006) - (equiv 6 months
  in Iraq, or, quarterly profits for the 3 biggest
  oil companies, or 1 Bill Gates)
• more than 80 satellites would be disabled
• Approx 100 LEO would reenter prematurely

7
Current Forecasting Models
Models begin with synoptic maps of photospheric
magnetic field static field for a given solar
rotation.

• Typical model chain
• model coronal field
• Coronal solution sets inner boundary to
  heliospheric model
• CCMC has 4 model chains (WSA, WSA/ENLIL, CORHEL,
  SWMF SC/IH)
• Principal role to model ambient corona and
  heliosphere
• Only beginning to dabble in transients

GONG
8
SWMF SC/IH (Univ. Mich)
Corona
CCMC Realtime run for April 30, 2008 using GONG
magnetogram
Inner Heliosphere
Solution at 1AU
9
WSA/ENLIL
Cone model approximation for CME
10
ENLIL-WSA Fieldline Tracing
11
Advances in Space Weather Models

• Current Generation
• use static synoptic magnetograms at 1o
  resolution
• create ambient corona and heliosphere solutions
• Next Generation
• will benefit from much better
• data sources,
• model algorithms
• computer hardware
• will create time dependent coronal models
• will use time dependent vector magnetogram data
  at 0.05o resolution
• the models physics, not the data, will define
  the resolution.

12
Next Generation Space Weather Models

• Coronal models will be
• Global
• Time dependent
• 3D MHD with adaptive mesh refinement
• Driven by observed surface flows
• Models will need to support both forecasting and
  research
• Function with latest data and archived data
  regardless of data limitations
• Models will define spatial resolution and cadence
  of magnetogram data at inner boundary
• eg global vector fields with maximum resolution
  of 1, cadence of 1 second

Where will these models get their data ?
13
Magnetogram Problems

• Magnetogram source limitations include
• Cadence and duty cycle
• Resolution
• Field of view
• Quality, particularly horizontal components of
  vector data
• Systematic errors associated with line fittings
• No coverage of far side
• Very poor polar fields
• No single source provides enough coverage !
• eg SDO 1 resolution data
• Limited FOV Vector data every 10 minutes
• Full Disk Line of Sight data every 10 minutes
• Full disk vector data every 6 hours

Need a CAD-like tool to synthesize magnetograms
from multiple sources
14
A Hypothetical Modeling Challenge

• Active Region evolution Model
• Suppose we need a model for slow evolution of
  Active Region A
• There is a second active region B on disk
• Synoptic vector magnetogram data is available
  from Kitt Peak along with individual vector
  magnetograms taken 3 times per day. However data
  for region B is poorly sampled due to instrument
  problems.
• Marshall Vector Magnetogram has data for B but at
  different times and resolution than Kitt Peak.
• Also have LOS magnetograms at selected times from
  Kitt Peak, Mt Wilson and MDI.

B
A

• How do we provide global surface vector fields
  and flow fields to an active region model at the
  spatial and temporal resolution required by the
  model?

15
Modeler Requirements

• Using this data requires,
• Ability to synthesize datasets into a greater
  whole
• More than just stitching images together
• Nobody knows what it will take to make
  synthesized maps accurate
• this really is a separate modeling challenge.
• We are not solving this problem just building
  the infra-structure to support it
• Also incorporates determination of surface flows
  to drive coronal field evolution
• Why? is it better to determine velocities from
  each magnetogram and then combine velocities,
  or combine magnetograms and determine a single
  velocity map? Not clear yet!

16
Tool Requirements

• Ability to interpolate in space (on a sphere) and
  in time
• Ability to handle many file formats
• Data usually fits, sometimes ascii
• Model customized at whim of developer
• Graphics IDL, TecPlot, OpenDx etc

• Ability to support complex processing algorithms
  (with many yet to be defined)
• Our tool is called MAGIC (MAGnetogram Integration
  and Composition)

17
MAGIC Design

• Modular design 6 components
• GUI (Python/TkL)
• A magnetogram database manager (MySQL)
• VSO interaction ?
• Lightweight magnetogram processing layer,
  executing interactive single line calls to
  Kameleon functions and simple canned Python
  routines for frequently used processing tasks (eg
  monopole subtraction)
• A third-party program execution interface
• Small suite of basic visualization tools (IDL and
  OpenDx)
• A command recorder function to facilitate batch
  processing.
• Open Source Linux Application

18
Typical Basic Use

• User requests a menu of all available data for
  time frame from database
• User selects their preferred data for each time
• User imports their model surface grids for all
  required times
• MAGIC does default (x,t) interpolation for each
  dataset to the appropriate grid
• User calls basic composition function for first
  grid - at prompt they input requested dataset
  weights or weighting rule
• MAGIC returns a composed surface vector field
  with a set of default images (Br, B?, B?, J)
• MAGIC asks if this synthesized magnetogram is
  acceptable
• No - go back and rework
• Yes - move on to next grid
• MAGIC reads in second grid
• Etc
• MAGIC outputs synthesized magnetograms in
  KAMELEON format files

19
MAGIC Design

• Backbone already in hand in CCMCs Kameleon
  Tool.
• KAMELEON (Maddox)
• two components, a file formater and an
  interpolator
• handles many file formats
• handles many model coordinate systems
• has both spatial and temporal interpolation
  functions
• portable an interactive interface and a
  callable library (from C, Fortran and IDL).

20
STATUS

• First funds only arrived in last few weeks
• Initial focus on defining a generic magnetogram
  format inside Kameleon
• Kameleon can now read, reformat, and interpolate
  on
• LOS data from Kitt Peak, Mt Wilson, SOHO/MDI
• vector data from Kitt Peak and Marshall Vector
  Magnetograph.
• Have begun initial GUI construction
• Added first python processing widget a monopole
  removal function
• Added graphics calls to compare initial data,
  interpolated data and data after processing.

21
Importing and Converting Datafiles
22
Processing Data
Example Interpolation of synoptic map onto test
model grid followed by removal of any monopole
signal.
23
Summary

• Developing a magnetogram synthesis tool
• using KAMELEON as the low-level manager of the
  data structures, I/O interfaces and basic
  interpolation layer
• Upon this foundation we add two processing layers
  (lightweight and heavyweight)
• Have added ability to ingest and interpolate most
  current magnetogram files
• Have begun building GUI and lightweight
  processing layer
• Have begun developing visualization tools to
  display different stages of data processing.

24
Modeler Requirements

• A tool is needed to do this synthesis.
• Who will develop it ?
• Not the data providers
• Not the modelers
• This is middleware but every bit as important
  as the data acquisition and the modeling.
• This infrastructure element is very much IT in
  nature.
• What will it take to eliminate this need?
• Full globe high resolution high cadence vector
  magnetograms in both photosphere and
  chromosphere! not within the next decade (or
  two?) !
• Even this does not facilitate future scientific
  analysis of historical events or older archived
  data.

25
ENLIL Heliosphere Model (Odstrcil)

• 3D MHD equations solved from 21.5rs to 2 AU
• Input at rotating inner boundary
• MHD parameters
• Output
• Magnetic field
• Velocity
• Density
• Temperature
• Two operating modes
• Ambient solution
• CME modeling using Cone model approximation