theoretical and numerical studies of ionospheric...
TRANSCRIPT
Theoretical and numerical studies
of ionospheric irregularities
Tatsuhiro Yokoyama
National Institute of Information and
Communications Technology (NICT), Japan
Outline
• Physics-based models
• Empirical models
• Numerical techniques
• Rayleigh-Taylor instability
• Perkins instability and E-F coupling
• Summary
What is Numerical Simulation?
Reproduce space environment virtually
in the computational space
• Density
• Velocity
• Temperature
• Electric fields
Computer
Ionosphere model
What is Numerical Simulation?
Inflow and outflow of
state variables
+ -
Ionization due to
solar radiation
Chemical reactions
O+
NO+
O2+
e-
O
N2
O2
Electric field
Geomagnetic
field lines
Effects of
electromagnetic
force
Collision with
neutral particles
State variables are updated every time step based on first-principles equations.
Equations
vit
vi vi 1
minip in vi vn g
e
miE vi B
3
2nik
Tit
qi t
nimi itmi mt
3k Tt Ti mt vi vt 2
3
2nek
Tet
qe Qe
nit
nivi Pi Li From atmospheric model
From electrodynamics model Photoionization&
Chemical reactions
Ion-neutral collisions Lorentz force
Heat flux Effect of collisions
Photoelectron heating Heat flux
Ion density
Ion velocity
Ion temperature
Electron temperature
Ionospheric Model Results
O+density Electric field
Time evolution of state variables at 400 km height
Electron temperature
Ground-to-topside model of Atmosphere and Ionosphere for Aeronomy(GAIA)
Ionosphere model
-Shinagawa and Oyama [EPS, 2006]
-Physics and chemistry for
ions (N2+, O2
+, NO+, O+ , H+)
-Grid size: 1°×5°×10-100 km
-Height range: ~100 to 3000 km
Whole atmosphere GCM
-Miyoshi and Fujiwara [GRL, 2003]
-physics and chemistry
from troposphere to thermosphere
T, u, v, w, N(O), N(O2), …
-Grid size: 2.8°×2.8°× 0.4H
Electrodynamics model
-Jin et al. [JGR, 2008]
-Solve current continuity with J=σ・(E+U×B) for electrostatic potential
-Equipotential magnetic field line
Atmosphere-Ionosphere Coupling
Rain fall rate
on ground
Electron density
distribution at 400
km height
condensation of
water vapor
→ latent heat release
upward propagation of
atmospheric waves
[Jin et al., 2011]
“Wave-4” observed by
IMAGE satellite
[Immel et al., 2006]
Available Models for Community
SAMI2 --- Ionosphere model developed by NRL (Naval Research
Laboratory, USA), 2-dimensional on the same meridional plane.
http://www.nrl.navy.mil/ppd/branches/6790/sami2
TIE-GCM --- thermosphere-Ionosphere-electrodynamics model
developed by NCAR (National center of atmospheric research, USA),
community model
http://www.hao.ucar.edu/modeling/tgcm/
GAIA --- whole atmosphere-ionosphere coupled model developed by
Kyushu Univ, Seikei Univ. and NICT, model itself not available currently
(model size is too large) but the result data is available
http://seg-web.nict.go.jp/GAIA/index_e.html
SAMI2
Model Comparison
Fang et al. (2013)
Models are not perfect.
There are large discrepancies
between various models.
Empirical Models
ρ
x/t (position/time)
ρ(x) = a x^3 + b x^2 + c x + d
observation
data
• Described by arbitrary fitting functions with coefficients
determined statistically by observation data
• The coefficients are dependent on locations, time, and
physical variables, and usually functions of such input
parameters as F10.7, Ap index,…
• examples: IRI(ionosphere), MSIS(neutral), HWM(neutral
wind), EXB drift,…
True profile
lack of observation
IRI (International Reference Ionosphere)
Input Year, Day of year, UT, Altitude, Geodetic latitude and
longitude, Sunspot number (Rz12), and Ionospheric index
(IG12).
Output Electron density, electron temperature, ion temperature,
ion composition (O+, H+, He+, NO+, O2+), ion drift, TEC
Cover Regions global, 50km-2000km
Conditions Monthly averages in the non-auroral ionosphere for
magnetically quiet conditions
Data Sources worldwide network of ionosondes, the incoherent scatter
radars (Jicamarca, Arecibo, Millstone Hill, Malvern, St.
Santin), the ISIS and Alouette topside sounders, and in
situ instruments on several satellites and rockets
Comparison Between GAIA, IRI and Ionosonde
IRI GAIA
ionosonde
(Kokubunji)
Solar maximum Solar minimum
Comparison Between GAIA, IRI and Ionosonde
GAIA, NmF2 IRI, NmF2
IRI
GAIA
ionosonde
(Wakkanai, Kokubunji,
Yamagawa, Okinawa)
For monthly average,
IRI shows very good results.
MSIS (Mass Spectrometer Incoherent Scatter)
Input Year, Day of year, UT, Altitude, Geodetic latitude and
longitude, F10.7 index (for previous day and three-month
average), and Ap index (daily or Ap history for the last 59
hours).
Output Neutral densities (He, O, N2, O2, Ar, H, and N, total mass
density) and temperature
Cover Regions global, 0km-1000km
Data Sources Several rockets, satellites (OGO 6, San Marco 3, AEROS-A,
AE-C, AE-D, AE-E, ESRO 4, and DE 2), and ISR (Millstone
Hill, St. Santin, Arecibo, Jicamarca, and Malvern) for
thermosphere.
Below 72.5 km, zonal average temperature and based on
the MAP Handbook tabulation. Below 20 km,
supplemented with averages from the National
Meteorological Center (NMC).
Comparison Between GAIA and MSIS
MSIS
GAIA
Empirical model and physics model show good agreement
at 300km but not at 110 km. There is a lack of observation
in the lower thermosphere.
Comparison Between GAIA and MSIS
MSIS
GAIA
neutral temperature at 110 km height
GAIA, Tn MSIS, Tn
They are different in mean value,
wave amplitude and phase…
HWM (Horizontal Wind Model)
Input Year, Day of year, UT, Altitude, Geodetic latitude and
longitude, F10.7 index (for previous day and three-month
average), and Ap index (daily or Ap history for the last 59
hours).
Output Neutral wind velocities (zonal and meridional
components)
Cover Regions global, 0km-1000km
Data Sources Wind data obtained from the AE-E and DE 2 satellites,
incoherent scatter radar and Fabry-Perot optical
interferometers, and MF/Meteor data
Comparison Between GAIA and HWM
HWM GAIA
zonal wind at 300 km height
GAIA, Un HWM, Un
They are in good agreement
at least at equator.
Comparison Between GAIA and HWM
HWM GAIA
zonal wind at 110 km height
GAIA, Un HWM, Un
They look very different ,
including tidal variations.
Comparison Between Empirical and Physics-Based Model
Empirical models Physics-based models
Approach Statistical approach, fitting
to arbitrary functions
Solving the physics equations
(energy, momentum…)
Prediction
capability
Climatological behaviors
(location, seasonal, solar
cycle dependences, ..),
quiet time behaviors
day-to-day variations,
disturbances due to magnetic
storms and lower atmospheric
inputs
Accuracy good where sufficient
observation data exists, and
bad for others (e.g., lower
thermosphere)
Good for some regions and
disturbances, but generally
difficult to make good agreement
with observations due to
problems of numerical treatments
usage Reference use, prediction of
climatology, input for
regional physics-based
models, etc.
prediction, analysis of physics,
numerical experiments, etc.
Computationa
l requirements
Not heavy Depends on the model and
resolution, usually heavy
How to Model?
Finite difference method
Finite volume method
Finite element method
…
i-1, i, i+1
i-2 i-1 i i+1
Initial Value Problem
0
x
cv
t
cCentral Difference 0
2
11
1
x
ccv
t
cc n
i
n
i
n
i
n
i
)(2
11
1 n
i
n
i
n
i
n
i ccx
tvcc
i-1, i, i+1, i+2
n+1
n
x
t
Initial Value Problem
0
x
cv
t
cUpwind Difference 01
1
x
ccv
t
cc n
i
n
i
n
i
n
i
)( 1
1 n
i
n
i
n
i
n
i ccx
tvcc
(if v > 0)
i-1, i, i+1, i+2
n+1
n
x
t
v
CIP Scheme
Constrained Interpolation Profile (CIP) method
Advection of density gradient as dependent variable
Yabe et al. (2001)
How to Solve Ax=b?
Direct solution (e.g., Gaussian elimination)
– An order of N3 operation. Non-zero elements are filled-in.
⇒ Suitable for dense matrix
Iterative methods (Jacobi, Gauss-Seidel, SOR)
– Matrix-Vector multiplication at each iteration
⇒ Suitable for sparse matrix
Multigrid method
bAx ULDA
))( n11nxUL(bDx
Conjugater Gradient (CG) Method
Non-stationary iterative method, while SOR-type
is called stationary iterative method.
Green: steepest descent
Red: conjugate gradient
High Performance Computing
Early supercomputer consists of vector processor.
Vector-type machine becomes expensive compared
to parallel scalar-type machine.
Graphical Processing Unit (GPU) is also used to
accelerate simulation.
Huge vector processor Parallel scalar processor
Amdahl's Law
Writing efficient source code is critical.
S: SpeedUp
P: Parallelized portion
N: Number of processors
Linear Analysis of Rayleigh—Taylor Instability
𝜕𝑁
𝜕𝑡+ 𝛻 𝑁𝑽𝑖 = 0, 𝛻 ⋅ 𝑱 = 𝛻 ⋅ 𝑒𝑁 𝑽𝑖 − 𝑽𝑒 = 0
Incompressible plasma in the F region(𝛻 ⋅ 𝑽 = 0)
Zeroth-order vertical gradient only 𝜕𝑁
𝜕𝑧
Magnetic field diercts northward 𝐵 = 𝐵𝑦
Ignore second-order perturbation
𝜕𝑁
𝑑𝑡+
𝑀𝑔
𝑒𝐵
𝜕𝑁
𝜕𝑥−
1
𝐵
𝜕𝜙
𝜕𝑥
𝜕𝑁
𝜕𝑧= 0
𝜕𝑁
𝜕𝑥−
𝜈𝑖𝑛
𝑔𝐵𝑁
𝜕2𝜙
𝜕𝑥2= 0
Linear Analysis of Rayleigh—Taylor Instability
Plane wave assumption for density and electrostatic
potential
𝜙 = 𝛿𝜙𝑒𝑖(𝜔𝑡−𝑘𝑥), 𝑁 = 𝑁0 𝑧 + 𝛿𝑁𝑒𝑖(𝜔𝑡−𝑘𝑥)
𝑖𝜔 − 𝑖𝑘𝑀𝑔
𝑒𝐵𝛿𝑁 +
𝑖𝑘
𝐵
𝜕𝑁0
𝜕𝑧𝛿𝜙 = 0
−𝑖𝑘𝛿𝑁 +𝜈𝑖𝑛
𝑔𝐵𝑁0𝑘2𝛿𝜙 = 0
Set determinant to be zero for coefficients of 𝛿𝑁, 𝛿𝜙,
resulting dispersion relation
𝜔 = 𝑘𝑀𝑔
𝑒𝐵− 𝑖
𝑔
𝜈𝑖𝑛
1
𝑁0
𝜕𝑁
𝜕𝑧
Linear Analysis of Rayleigh—Taylor Instability
When the imaginary part of 𝜔 is negative (upward
density gradient), 𝛾 in 𝑒𝑖𝜔𝑡 = 𝑒𝑖𝜔𝑟𝑡𝑒𝛾𝑡 becomes
positive and the perturbation grows with time. 𝛾 is
called the linear growth rate.
𝛾 =𝑔
𝜈𝑖𝑛
1
𝑁0
𝜕𝑁
𝜕𝑧=
𝑔
𝜈𝑖𝑛𝐿
Preferable condition for the instability growth:
– Small collision frequency -> high altitude -> eastward
electric field
– Steep vertical density gradien -> recombination in the
bottomside around sunset time
Need to consider whole magnetic flux tube
– Simultaneous sunset at conjugate E regions -> sunset
terminator is parallel to magnetic declination
Growth Rate Estimation by Global Simulation
Global model does not
have enough spatial
resolution to reproduce
plasma bubbles.
Using output parameters,
the growth rate of the
Reyleigh-Taylor instability
can be estimated.
General seasonal and
longitudinal pattern can
be well explained.
Wu (2015)
History of Plasma Bubble Modeling
In the review paper by Woodman (2009), “…The qualitative theory of
Woodman and La Hoz (1976) was soon supported by numerical
simulations (Scannapieco and Ossakow, 1976) (this sequence of events is
sometimes inverted in review papers and historical introductions). The idea
of a bubble ‘floating’ to the top was first presented by Woodman at the
1975 Gordon Conference on Space Plasma Physics where it was suggested
that the NRL code used to explain striations in Barium cloud releases could
simulate the bubble formation…”
Woodman and LaHoz (1976)
Received Jan. 26, 1976, published Nov. 1976
Scannapieco and Ossakow (1976)
(Received Apr. 5, 1976, published Aug. 1976)
High Resolution Bubble (HIRB) Model by NICT
Dipole orthogonal coordinate
Maximum spatial resolution perpendicular to B is 200 m.
(NX(B||), NY(BL), NZ(B)) = (501, 3600, 1680)
O+ (F region), NO+ (E region)
Yokoyama et al. (2014)
Medium-Scale Traveling Ionospheric Disturbances (MSTID)
630-nm airglow GPS-TEC
3m-scale irregularities
(MU radar)
Saito et al. (2001)
Perkins Instability
Perkins instability can produce banded structure.
Linear growth rate is very small.
Field line-integrated model for theoretical and
numerical studies (e.g., Kelley and Miller, 1997).
E-F coupling is important (e.g., Cosgrove et al., 2004).
0
0
Perkins (1973)
Zhou and Mathews (2006)
Simulation of Perkins Instability
First 2D simulation of Perkins instability in 1997.
Tilted bands are produced, but the growth rate is too small.
Kelley and Miller (1997)
Sporadic E (Es) Layer
Long-lived metallic ions survive even after sunset.
They are gathered by neutral wind shears and form
thin layers.
Mathews et al. (1997)
Perkins-Type Instability in the E Region
Similar mechanism also works in the E region.
Cosgrove and Tsunoda (2002)
Tsunoda et al. (2004)
Hysell et al. (2004)
Basic Idea of E-F Coupling
Westward wind on Es patch generates eastward
polarization electric field (upward ExB drift in the F region).
Haldoupis et al. (2003)
E-F Coupling Evidence
Irregularities occur simultaneously in both regions
along the same magnetic field.
F region echoes (projected on 100km)
E region echoes
F region
E region
Sakata
MU radar
Alt
itu
de
Growth Rates of Coupled Instability
The second terms of (3) and (4) come from polarization
electric field mapped from the other region.
Pc ≫ P and E
c E.
Tsunoda (2006)
Simulation Model for E-F Coupling Instability
O+ (F region), NO+ (E region),
Fe+ (Es layer), and electrons.
45⁰ inclination of B.
The altitude range is 90 -
470km. The grid spacing is
2km in each direction in F
region and 500m in the E
region.
Periodic boundary in
horizontal directions. Yokoyama et al. (2009)
E-F Coupling Simulation
Time variation on a meridional plane.
The F-peak altitude is modulated 8km, and the
conductivity variation reaches more than 30%.
Growth Rate
F-region conductivity variation reaches 10% after 700s,
while it takes 7000s by the isolated Perkins instability.
New Model with Dipole Magnetic Field
Random perturbation + zonal wind shear in the E region
Yokoyama and Hysell (2010)
Southward Propagation by E Region Neutral Wind
E region F region
Random perturbation + rotational shear in the E region
E
S
W
Es
Scale Dependence of MSTID
Key results:
Cases 3 and 4 grew most rapidly
Scale distribution of MSTIDs converged to 100-200 km
Very long frontal structures of MSTIDs were formed.
Case 1 2 3 4 5
Perturbation
scale 20km 40km 80km 160km 320km
E-region wind Zonal wind shear (60 m/s)
F-region wind Southwestward (120 m/s)
Scale Dependence of MSTID
E region
F region
Growth Rate Wavelength
Case 1 2 3 4 5
Scale
(km) 20 40 80 160 320
Polarization in Es Layer
Smaller-scale Es perturbation:
produces larger Ep by larger .
is reformed by wind shear more quickly.
small
small Ep
Large
Large Ep
Yokoyama et al. (2009)
Long Frontal Formation
Polarization process along the
wavefront makes it uniform due to
Perkins “stability” mechanism.
It does not prevent Perkins “instability”
because Jp can flow uniformly along it.
E F
Suggested References
• Global model
Fang, T.-W. et al., Comparative studies of theoretical models in the
equatorial ionosphere, Modeling the Ionosphere-Thermosphere System,
Geophysical Monograph Series, 133-144, AGU, 2013.
• Rayleigh-Taylor instability
Sultan, P. J., Linear theory and modeling of the Rayleigh-Taylor instability
leading to the occurrence of equatorial spread F, J. Geophys. Res., 101,
26,875-26,891, 1996.
Yokoyama, T., A review on the numerical simulation of equatorial plasma
bubbles toward scintillation evaluation and forecasting, Prog. Earth
Planet. Sci., 4:37, 2017.
• Perkins instability
Tsunoda, R. T., On the coupling of layer instabilities in the nighttime
midlatitude ionosphere, J. Geophys. Res., 111, A11304, 2006.
Yokoyama, T. et al., Three-dimensional simulation of the coupled Perkins
and Es-layer instabilities in the nighttime midlatitude ionosphere, J.
Geophys. Res., 114, A03308, 2009.