Mariangel Fedrizzi

CTIPe
HIGHLIGHTS and
REVIEW of
ACHIEVEMENTS
Mariangel Fedrizzi, Timothy J. Fuller-Rowell
(University of Colorado/CIRES and NOAA/SWPC, Boulder, CO)
Mihail Codrescu
(NOAA/SWPC, Boulder, CO)
Review of Achievements
CHAMP x CTIPE: 27 Oct - 01 Nov 2003
Halloween Storm (2003)
Before empirical NO
CTIPe is sampled
along the
CHAMP orbit
Empirical NO model: Marsh et al. 2004
• Based on over 2.5 years of measurements from the
Student Nitric Oxide Explorer (SNOE) scientific
satellite
NO Contours
150 km
• Lower thermosphere (100-150km)
• EOFs
• Dependent on EUV, season, auroral activity, height,
and latitude
After empirical NO
100 km
Kp = 1.6
Kp = 3.5
2007 CHAMP/CTIPe Orbit Average Comparisons
No semi-annual variation in
density included in CTIPe
400 km
R= 0.73
RMSE= 0.31
BIAS= 0.23
SD= 0.21
Possible Mechanisms for Semi-Annual
Variation:
• Seasonal variation in eddy diffusivity in the upper
mesosphere and lower thermosphere regions due to
gravity wave breaking (Qian et al., 2009)
• Thermospheric spoon mechanism associated with the
global scale interhemispheric circulation at solstice
(Fuller-Rowell, 1998)
• Asymmetry in conductivity distribution at solstice due
to inequality of solar radiation between hemispheres
(e.g, Lyatsky et al., 2001)
• Semi-annual variation in geomagnetic activity
peaking at equinoxes (Russell and McPherron, 1973)
• Conduction mode oscillation of the thermosphere
forced by the semi-annually varying Joule heating at
high latitudes (Walterscheid,1982)
Some of these mechanisms are not included in the model, and it is possible that a combination of these effects
(and various others listed by Qian et al., 2009) could be responsible for the semi-annual variation in density.
This is still an open question, and further investigation is required to understand the seasonal processes in the
upper atmosphere.
2007 CHAMP/CTIPe Orbit Average Comparisons (cont’d)
• Main purpose of this study is to simulate
the model response to short-period
variations in geomagnetic activity during
the year.
400 km
New CTIPe run including seasonal variation in
the E-field small scale variability
• To accommodate this goal the semiannual variation of neutral density in
CTIPe has been removed by introducing a
semi-annual variation in electric field
small-scale variability.
• Electric fields can directly change Joule
heating by varying the ion convection at
high-latitudes (Deng and Ridley, 2007). An
increase in Joule heating raises the neutral
temperature, which enhances the neutral
density at constant heights.
• Electric field variability changes the
distribution of Joule heating significantly,
and can introduce interhemispheric
asymmetries (Codrescu et al., 1995, 2000).
R= 0.88
RMSE= 0.17
BIAS= -0.017
SD= 0.17
2005/6/8/9 CHAMP/CTIPe Orbit Average Comparisons
RMSE: 0.39
Correlation: 0.74
Bias: -0.14
SD diff: 0.37
RMSE: 0.21
Correlation: 0.84
Bias: -0.01
SD diff: 0.21
RMSE: 0.23
Correlation: 0.88
Bias: -0.02
SD diff: 0.22
RMSE: 0.16
Correlation: 0.85
Bias: -0.01
SD diff: 0.15
2005-2009 CHAMP/CTIPe Orbit Average Comparisons
400 km
YEAR
R
RMSE
BIAS
SD
2005
0.74
0.39
-0.14
0.37
2006
0.84
0.21
-0.01
0.21
2007
0.88
0.17
-0.017
0.17
2008
0.88
0.23
-0.02
0.22
2009
0.85
0.16
-0.01
0.15
CHAMP data version 2.3 provided by Eric Sutton
Partitioning of Energy during a Geomagnetic Storm
• Agreement between model and
observations suggests that the
amount of energy influx from solar
radiation and magnetospheric
sources deposited into the
atmosphere is reasonably accurate,
enabling the model to be used to
estimate the rate of energy influx
from those sources.
• Integrated Joule heating accounts
for the majority of energy input into
the ionosphere/thermosphere
system during geomagnetic storms
(e.g., Lu et al., 1995; Knipp et al.,
2004; Fuller-Rowell and Solomon,
2010).
Partitioning of Energy during a Geomagnetic Storm
CTIPe is able to follow the
CHAMP neutral density
response with reasonable
fidelity. It is now possible to
explore the use the model
estimates of Joule heating as an
alternative index for the neutral
density response. In this study,
an index based on CTIPe Joule
Heating is derived.
Relationship between CTIPe Joule Heating and neutral
density: Joule heating index
• CTIPe Joule heating index is obtained by deriving
a relationship between the time series of CTIPe Joule
heating and CHAMP neutral density observations.
• Seasonal and solar cycle influences from the
CHAMP neutral density observations are removed,
so that the residual neutral density variations are
primarily due to geomagnetic activity. This is
achieved by subtracting the quiet-time density from
the satellite density measurements. The quiet-time
density is obtained from the empirical model
Jacchia-Bowman 2008 (JB2008; Bowman et al.,
2008b), assuming no variation from geomagnetic
activity, i.e., Ap and Dst are set to zero.
2007
Lag: 4 days
Maximum correlation:
~ 8 hours
• Cross-correlation between CTIPe Joule heating and residual neutral density as a function of time lag:
the maximum of cross-correlation is found at about 8 hours, beyond which the correlation gradually
decreases, approaching zero after a lag of about 3 days. This implies that magnetospheric energy sources
up to 3 days before are affecting the neutral density at the current time. It also implies the recovery time
is about 3 days from NO cooling and downward heat conduction. This correlation analysis provides the
filter shape by which to scale the time-series of Joule heating to best match the CHAMP neutral density.
Relationship between CTIPe Joule Heating and neutral
density: Joule heating index (cont’d)
INPUT
(Joule heating)
IMPULSE
RESPONSE
(shaping
filter)
OUTPUT
(Joule heating index)
• Convolution of CTIPe Joule heating with the shaping filter results in a new parameter (Joule heating
index) representing the integral of the product of the two functions.
Relationship between CTIPe Joule Heating and neutral
density: Joule heating index (cont’d)
• Linear regression between Joule heating index and the "observed" density residuals (CHAMP minus
quiet-time JB2008) is used to produce the new density residuals derived from Joule heating index.
• New absolute values of neutral density are obtained by adding the density residuals derived from
Joule heating index to the quiet-time JB2008 neutral density.
R= 0.92, RMSE= 0.10
BIAS= -0.01, SD= 0.10
R= 0.90, RMSE= 0.16
BIAS= -0.06, SD= 0.14
Relationship between CTIPe Joule Heating and neutral
density: Joule heating index (cont’d)
2005
Lag: 4 days
• Convolution of 2007 CTIPe Joule heating with the 2005 shaping
filter generates a new set of Joule heating index values. The linear
regression is used to obtain the new density residuals derived
from Joule heating index, which are added to the 2007 quiet-time
JB2008 neutral density to produce the absolute neutral density.
Maximum correlation:
~ 5 hours
R= 0.95
RMSE= 0.09
BIAS= -0.007
SD= 0.09
Fedrizzi et al. (Space Weather, 2012)
New Highlights
2005 CHAMP/CTIPe Orbit Average Comparisons
Orbit Averaged
400 km
Orbit-by-orbit
CHAMP/CTIPe Orbit Average Comparisons during Intense
Geomagnetic Storms
• Once every half hour SWEPAM takes measurements in a different mode (STI) which covers (nearly) the full instrument
energy range with reduced resolution (Ruth Skoug, LANL)
Nitric Oxide
- Nitric oxide empirical model (NOEM) is based
solely on SNOE observations and its use in such
studies should consider the conditions under which
those observations were made.
- SNOE data only spanned F10.7 between 90 and
~275 (the ascending phase of cycle 23).
- Kp rarely dipped below 0.5 or exceeded 5.
- Using values outside the range of forcing seen by
SNOE, you are extrapolating, which in extreme
conditions could give unphysical results (Dan
Marsh, private communication).
Marsh et al. (J. Geophys. Res., 2004)
- In NOEM’s empirical model, the timescale for
Nitric Oxide loss is essentially built into the one day
lag used for the Kp index. This was determined from
a lag-correlation of Kp with the principal component
that corresponded to the EOF associated with the
auroral NO signal.
- In CTIPe, NOEM model is driven by 24h running
average Kp.
Nitric Oxide
Marsh et al. (J. Geophys. Res., 2004)
Mlynczak et al. (J. Geophys. Res., 2010)
CHAMP/CTIPe Orbit Average Comparisons during Intense
Geomagnetic Storms
R= 0.74
RMSE= 0.39
BIAS= -0.14
SD= 0.37
R= 0.83
RMSE= 0.20
BIAS= 0.015
SD= 0.20
R= 0.88
RMSE= 0.17
BIAS= -0.017
SD= 0.17
R= 0.88
RMSE= 0.14
BIAS= 0.006
SD= 0.14
R= 0.88
RMSE= 0.17
BIAS= -0.017
SD= 0.17
R= 0.88
RMSE= 0.14
BIAS= 0.006
SD= 0.14
Joule heating Index Test
Courtesy: Bruce Bowman
JH Index derived from CTIPe
run with NO scaling factor and
JH factor 1.0
Joule heating Index Test
Courtesy: Bruce Bowman
JH Index derived from CTIPe
run with NO scaling factor and
JH factor 1.0
Summary and Future Work
• Empirical NO model improved recovery timescale of neutral density to geomagnetic activity in CTIPe
model. However, a scaling factor is needed for higher Kp values, and SABER data will be used in this
process.
• Introduction of seasonal variation in small scale electric field variability improved CHAMP/CTIPe
agreement during equinoxes. However, the ionospheric response and other possible mechanisms for
semi-annual variation in neutral density still need to be investigated.
• Agreement between model and observations suggests that the amount of energy influx from solar
radiation and magnetospheric sources deposited into the atmosphere is reasonably accurate, enabling the
model to be used to estimate the rate of energy influx from those sources.
• First results on Joule heating index development show improvement in neutral density estimates.
Further work is needed, and the analysis will be extended to all years of CHAMP/GRACE data to cover
the solar cycle.
• Even though good agreement between CTIPe and CHAMP indicates that Joule heating is reasonably
correct, we need more accurate specification of the Joule heating magnitude and spatial distribution, and
continue to improve geomagnetic forcing (e.g., open GGCM by including information from MHD code,
Weimer updated version, AMIE).
• Quantitative comparison of year-long climatology and day-to-day variability of CTIPe physics-based
model with comprehensive ground and space-based observations (e.g., FPI, ionosonde,
CHAMP/GRACE/GOCE, GUVI, COSMIC, GPS-TEC), and empirical models (HWM07, IRI2007,
NRLMSISE-00).