, W. B. Manchester
, B. van der Holst
, I. Sokolov
, G. Toth
, A. Vourlidas
C. de Koning
, T. I. Gombosi
, R. E. Mullinix
, A. Taktakishvili
, A. Chulaki
UCAR/NASA Jack Eddy Postdoc Fellow
Lockheed Martin Solar and Astrophysics Lab, Palo Alto, CA
CLaSP, University of Michigan, Ann Arbor, MI
APL, Johns Hopkins University, Laurel, MD
CIRES, University of Colorado, Boulder, CO
CCMC, Goddard Space Flight Center, Greenbelt MD
AGU Fall Meeting 2015
CME Forecasting Models
Coronal mass ejections (CMEs) are a major source of potentially destructive
space weather conditions (geomagnetic storm, SEPs). Due to the limited
observations, models play a vital role both for scientiﬁc understanding and for
Empirical Forecasting Models: using near-Sun CME observations (e.g.,
Gopalswamy et al. 2001), using data-mining techniques (Riley et al. 2015)
Kinematic Models: 3-D Hakamada-Akasofu-Fry version 2 (HAFv.2) model
(Hakamada & Akasofu 1982; Fry et al. 2001); Cone model (Zhao et al., 2002;
Hayashi et al., 2006).
Kinematic + MHD Heliosphere Models: ENLIL heliosphere model with the CME
cone model (Xie et al 2004; Odstrcil et al. 2005); 3D MHD model by Han et al.
(1988) with the HAFv.2 model (Wu et al., 2007). The average error in the CAT (CME
Analysis Tool)-Wang-Sheeley-Arge (WSA)-ENLIL operational model is 7.5 hours
(Pizzo et al. 2011; Millward et al. 2013).
MHD Corona Models + Magnetically Driven Eruptions: Provide magnetic ﬁeld
information at 1 AU (e.g., Usmanov & Dryer 1995, Wu et al. 1999, Feng et al. 2010,
Groth et al. 2000, Manchester et al. 2004, Lugaz et al. 2007, Toth et al. 2007, Cohen
et al. 2008, Shen et al. 2011, Zhou et al. 2012, Lionello et al. 2013)
A New Approach of Coupling AWSoM MHD with EEGGL
The analytical ﬂux rope parameters are derived from the observations.
Capability to trace the erupting magnetic ﬁeld from the Sun to 1AU.
Extensive comparison with the CME observations near the Sun and in the
AWSoM: Alfven-wave Solar Model
EEGGL: Eruptive Event Generator using Gibson-Low conﬁguration
References: van der Holst et al. 2010, Manchester et al.
2012, Jin et al. 2012, Sokolov et al. 2013, Oran et al.
2013, Jin et al. 2013, van der Holst et al. 2014
Data-driven inner boundary
condition by synoptic
Coronal heating and solar wind
accelerating by Alfven waves.
treatment of wave reﬂection,
dissipation, and heat
partitioning between the
electrons and protons.
Model starts from upper
chromosphere including heat
conduction (both collisional
and collisionless) and radiative
Adaptive mesh reﬁnement
(AMR) to resolve structures
(e.g., current sheets, shocks).
and synthesized EUV
images of the steady
state solar wind model
from SDO AIA 211,
STEREO A EUVI 171, and
STEREO B EUVI 195. The
observation time is 2011
March 7 ∼20:00 UT.
Synthesized EUV images
of the model.
between the model and
observation for different
structures of the Sun.
Gibson-Low Flux Rope
Analytical proﬁles of the GL ﬂux rope are
obtained by ﬁnding a solution to the
magnetohydrostatic equation and the
solenoidal condition (Gibson & Low
1998) through mathematical stretching
The transformed ﬂux rope appears as a
tear-drop shape of twisted magnetic ﬂux.
Lorentz forces are introduced, which
leads to a density-depleted cavity in the
upper portion and a dense core at the
lower portion of the ﬂux rope (3-part CME
The GL Flux Rope is determined by 4 parameters:
a (Fixed): determines the shape of the ﬂux rope
(Fixed): determines the initial position of the ﬂux rope before it is stretched
: determines the size of the ﬂux rope
: determines the magnetic strength of the ﬂux rope
With different GL radius/
strength parameters, a
linear relationship is
found between the ﬂux
rope poloidal ﬂux and the
CME speed near the Sun.
With the same ﬂux
rope parameters, the
CME speed is
inversely related to
the average Br
around the PIL of the
An Example of EEGGL Setting
Polarity Inversion Line (PIL)
GL Flux Rope Location
Red: Flux rope ﬁeld lines;
White: Large-scale helmet streamers;
Green: Field lines from surrounding active regions and open ﬁeld lines.
2011 February 15 CME
Plasma beta at 2.5 Rs
The eruption changes the
current sheet location
therefore the large-scale
Plasma density at 42 Mm
Waves reﬂection from the
north and south polar
coronal hole boundary.
Radial velocity at 42 Mm
Downward ﬂow due to the
expansion of the CME.
Observed by Hinode/EIS
(Harra et al. 2011)
EUV Waves (2011 March 7 CME)
Both the simulation and observation images are produced by tri-ratio running
difference method. The tricolor channels are AIA 211 (red), AIA 193 (green), and
AIA 171 (blue). The ratio in each channel is identically scaled to 0.8-1.2 for both
observation and simulation.
White-Light Images (2011 March 7 Event)
White-Light Images (2011 March 7 Event)
CME Evolution in the Heliosphere
(2011 March 7 Event)
1 AU Comparison (2011 March 7)
The second velocity and density peaks after the shock is due to the
The lower compression ratio in the model is caused by the elevated proton
temperature in the CIR region.
2012 July 12 CME Event
IMF Bz Evolution
We present this newly developed tool for simulating CMEs from the Sun to the
Earth. With increasing number of events studied, it will be possible to obtain
better empirical relations to constrain the EEGGL model. We will also utilize
vector magnetogram to specify the helicity of the GL ﬂux rope.
The computation expense using current AWSoM model is high (~40,000 CPU
hours on Pleiades). The new model AWSoM-R (Threaded Field Line Model;
Sokolov et al. 2015) may run 10~100 times faster so that real-time simulation
will be possible at CCMC.
The results will also improve our understanding about the erupting CME
structure and its evolution in the heliosphere therefore provide valuable
information for improving the current ﬂux rope models used in the numerical
There are issues with developing a forecasting model: We do not have full
coverage of magnetic ﬁeld observation on the Sun; We do not have direct
observation of the erupting CME magnetic structures. How these “missing
data” will inﬂuence our modeling capability needs to be understand in the