Sources of Variability in the Topside Ionosphere

SOURCES OF VARIABILITY IN THE

TOPSIDE IONOSPHERE

Jessica M. Hawkins

APPROVED BY SUPERVISORY COMMITTEE:

______Phillip C. Anderson, Chair

______Roderick A. Heelis

______Fabiano Rodrigues

______Russell Stoneback

______Lindsay J. King

Jessica M. Hawkins

To Charissa and Sandra for study dates; to Evan for keeping me curious with your love of learning; to Simone, Tim, Cory, Shawn, and Blake for encouragement and enduring friendship; and most of all to my parents and brother, for being my first teachers and my bedrock of love all these years

SOURCES OF VARIABILITY IN THE

TOPSIDE IONOSPHERE

JESSICA M. HAWKINS, BS, MS

DISSERTATION

Presented to the Faculty of

The University of Texas at Dallas

in Partial Fulfillment

of the Requirements

for the Degree of

DOCTOR OF PHILOSOPHY IN

PHYSICS

THE UNIVERSITY OF TEXAS AT DALLAS

May 2018

ACKNOWLEDGMENTS

I would like to thank my advisor Dr. Phillip Anderson for showing faith in me, devoting so much time to my education and patiently guiding me through this rite of passage, and to my committee for their time, advice and support. I am grateful to the physics department for sending me to several conferences, including a trip to visit the Jicamarca Radio Observatory which Dr. Fabiano

Rodrigues facilitated. Several professors and researchers took the time to mentor and answer my questions at these conferences, including Dr. Maura Hagan, Dr. Roger Varney, Dr. Anja

Stromme, and the people at NCAR who host the Advanced Study Program.

I would also like to thank my fellow graduate students, several of whom attended practice sessions for my proposal and other presentations. Brian Nathan always encouraged us grad students to help each other and get to know each other. Jessica Smith was a solid friend when exploring the conferences we attended. I had good conversations with Russell Landry and Brian about data processing algorithms over the years, and with Rob McIntosh about how to write the proposal. Weikang Lin patiently went over many practice problems with me before the quantum mechanics qualifier; his tutoring was a strong factor in helping me pass that subject.

Dr. Gregory Earle gave me my first instruction in space science as an undergraduate in his

CubeSat design class, which he put a lot of work and heart into. Thank you to my teachers at

ACE Camp who gave me my first instruction in programming through UTD when I was eleven years old, because that summer was the first time I felt like computer programming was "my

thing." Thank you to the CEDAR program and NSF for funding this research, and to Wiley for permission to reprint material published in the Journal of Geophysical Research.

April 2018

SOURCES OF VARIABILITY IN THE

TOPSIDE IONOSPHERE

Jessica M. Hawkins, PhD The University of Texas at Dallas, 2018

ABSTRACT

Supervising Professor: Phillip C. Anderson

The topside ionosphere is a dynamic region of the Earth’s upper atmosphere that couples with solar and terrestrial regions and affects communication and navigation systems. The solar extreme ultraviolet (EUV) irradiance is a major driver of variability in the topside, and further work is needed to understand how EUV drives variability, particularly in light of the recent availability of spectrally-separated EUV measurements. Variability in the topside is also driven from below, as there is increasing evidence that atmospheric effects are propagated across a broad range of altitudes. The WN4 pattern has been recognized as a signature of non-migrating tides originating in the troposphere, but its variations in the topside have yet to be comprehensively studied. Ionospheric models have struggled to accurately represent the shape and variability of topside ion density profiles, in part due to the relative scarcity of in situ measurements. Nevertheless, empirical models such as the internationally-recognized standard

International Reference Ionosphere (IRI) provide a crucial resource to scientists, and improvements to IRI are ongoing.

vii

This dissertation addresses these research needs in three studies. First, we investigated the response of the topside ionosphere to EUV variability using thermal plasma measurements from the Defense Meteorological Satellite Program (DMSP) over 23 years, a period of time never considered before. The response was found to be closely related to the composition due to the effect of this parameter on the scale height. The composition varies dramatically with the solar cycle and the SZA, and these changes in composition have a compounding effect on the response of the topside ionosphere to changes in solar EUV irradiance.

We investigated and quantified the signature of WN4 in the topside ionosphere across local time, season, and varying levels of solar activity using monthly averages of ion densities from DMSP.

This study provides compelling evidence of the effects of tropospheric weather on space weather even at DMSP altitudes, which had not been comprehensively presented before. Tidal patterns are strongly present along the magnetic equator throughout the year, and WN4 is clearest at

September equinox. During solstice months the WN4 signature is modified by zonal and meridional winds. WN4 moved eastward with increasing F10.7 and drifted eastward at about 2o

GLON per hour MLT.

To address the scarcity of in situ topside measurements needed to improve IRI, we presented a new data set of over 11,000 magnetic conjunctions between DMSP and C/NOFS in 2009-2015.

We investigated the error of the most recent version of IRI in reproducing the topside profile shape under different conditions. The model was found to underestimate the plasma scale height in the morning but matched the profile shape well in the afternoon. The percent error showed

viii

patterns consistent with a diffusive equilibrium condition in the afternoon, whereas strong ion flows in the morning produced a more complex relationship between the Ni profile shape and the plasma temperatures.

TABLE OF CONTENTS

ACKNOWLEDGMENTS …………………………………………………………………… v

ABSTRACT …………………………………………………………………………………. vii

LIST OF FIGURES …………………………………………………………………………. xii

LIST OF TABLES ………………………………………………………………………….. xvi

CHAPTER 1 INTRODUCTION …………………………………………………………... 1

1.1 The Topside Ionosphere ……………………………………………………………. 1

1.2 Geomagnetic Field Geometry, Winds and Effects on the Topside ………………… 5

1.3 Solar EUV Origin, Variability, and Effects on Topside Ionosphere ………………. 7

1.3.1 Origin and variability ………………………………………………………….. 7

1.3.2 Measurements and models …………………………………………………….. 7

1.3.3 Ionospheric effects …………………………………………………………….. 9

1.4 Atmospheric Tides and Coupling with Ionosphere-Thermosphere ………………... 11

1.4.1 WN4 and DE3 …………………………………………………………………. 11

1.4.2 DE3 propagation, WN4 in the F-region, and impact of WN4 on the upper atmosphere ……………………………………………………………………………. 13

1.4.3 Previous measurements and results in the topside ionosphere ………………… 15

1.5 Modeling of Electron Density Profiles …………………………………………….. 16

1.5.1 Scale height and transition height in the topside and modeling of topside electron density profiles ……………………………………………………………………….. 16

1.5.2 Methods of topside ion density profile reconstruction and estimation of the topside scale height …………………………………………………………………………… 17

CHAPTER 2 SCIENTIFIC OBJECTIVES AND RELEVANCE ………………………… 20

2.1 Research Overview ………………………………………………………………… 20

2.2 Relevance to Research Community ………………………………………………... 22

CHAPTER 3 DATA SOURCES AND MODELS ………………………………………... 25

3.1 TIMED SEE ……………………………………………………………………….. 25

3.2 SDO EVE ………………………………………………………………………….. 25

3.3 DMSP ……………………………………………………………………………… 25

3.4 C/NOFS ………………………………………………………………………….... 27

3.5 IRI …………………………………………………………………………………. 27

CHAPTER 4 TOPSIDE IONOSPHERIC RESPONSE TO SOLAR EUV VARIABILITY 29

4.1 Data ………………………………………………………………………………… 29

4.2 EOF Analysis ……………………………………………………………………… 39

4.3 Discussion …………………………………………………………………………. 45

4.4 Conclusions ………………………………………………………………………… 47

CHAPTER 5 WN4 VARIABILITY IN DMSP ION DENSITIES ACROSS SEASON, SOLAR CYCLE, AND LOCAL TIME ……………………………………………………………... 50

5.1 Satellites and Instrumentations ……………………………………………………. 50

5.2 Analysis Methods …………………………………………………………………. 51

5.3 Seasonal Variations ……………………………………………………………….. 53

5.4 Solar Cycle Variations ………………………………………………………...... 60

5.5 Local Time Variations …………………………………………………………….. 65

5.6 Conclusions ……………………………………………………………………...... 69

CHAPTER 6 TOPSIDE PROFILE SHAPE VARIATIONS USING DMSP / C/NOFS CONJUNCTIONS AND IRI ……………………………………………………………….. 72

6.1 Satellite Conjunctions Data Set ……………………………………………………. 73

6.2 Analysis Methods …………………………………………………………………... 77

6.3 Results ……………………………………………………………………………… 79

6.4 Conclusions ………………………………………………………………………… 88

CHAPTER 7 CONCLUSION ……..…………………………………………………...... 89

REFERENCES ……………………………………………………………………………... 93

BIOGRAPHICAL SKETCH ……………………………………………………………….. 103

CURRICULUM VITAE

xii

LIST OF FIGURES

1.1 Typical profiles of the neutral atmospheric temperature (left) and ionospheric plasma density (right.) [The Earth's Ionosphere, 1989] …………………………………………………………2

1.2 Dip latitude as a function of geographic latitude and longitude. [West and Heelis, 1996] ……….5

1.3 Weighted photoionization and photoabsorption cross sections for O+ for Torr and Torr [1979] bands and lines. …………………………………………………………………………………10

4.1 Examples of the daily averaged (a) ionospheric density, (b) H+/O+ ratio, and (c) ion temperature along with the (d) solar zenith angle in the dawn sector for two complete solar cycles from 1992 through 2014. The data are from three spacecraft - DMSP F11 (blue), F13 (red), and F17 (green). The E10.7 solar EUV proxy is shown in black in (a). …………………………………30

4.2 Daily averaged values in the dusk sector in the same format as Figure 4.1. ……………………31

4.3 Time-shifted CCCs between the density and E10.7 for both the dusk (red) and the dawn (black) sectors. …………………………………………………………………………………………33

4.4 (a) The daily averaged densities (black) for a 180-day period in 2003 along with the solar EUV flux at 28 nm (red) measured by the SEE instrument on the TIMED spacecraft and the E10.7 (blue—scale not shown—minimum 100 solar flux unit (SFU) and maximum 200 SFU). (b) The same format as (a) but for a 200-day period in 2013 with the daily averages of the SDO EVE- measured fluxes at 25.5 nm. ……………………………………………………………………34

4.5 Examples of the daily averaged (a) ionospheric density, (b) H+/O+ ratio, and (c) ion temperature along with the (d) solar zenith angle in the dawn sector for two complete solar cycles from 1992 to 2014 at +5° (red) and 5° (black). The data are from same three spacecraft as in Figure 4.1. The red and black bars at the top of (b) show when the H+/O+ ratio exceeds 20. ……………………36

4.6 Daily averaged values in the dawn sector in the same format as Figure 4.5 but for ±30° GLAT. ……………………………………………………………………………………………………37

4.7 Daily averaged values in the dawn sector in the same format as Figure 4.5 but for ±50° GLAT. ……………………………………………………………………………………………………38

4.8 Daily averages of the (a) SZA and the (b) density from the DMSP F15 (black) and F16 (red) spacecraft for 250 days during 2005. ……………………………………………………………39

4.9 The first principal component (PC1, black) for the dawn/dusk satellites in the dawn sector and E10.7 (red). ……………………………………………………………………………………42

xiii

4.10 Time-shifted CCCs between PC1 and E10.7 for both the dusk (red) and the dawn (black) sectors. ………………………………………………………………………………………….43

4.11 The second principal component (PC2, black), the density (red) at 30° GLAT, and the SZA (blue) at the same latitude. ………………………………………………………………………44

4.12 The EOFs (or eigenvectors of the covariance matrix) for the PC1 (blue) and PC2 (red). ……44

4.13 Density and temperature from two conjunctions of the DMSP (asterisks) and C/NOFS (lines) spacecraft. H+ (red) and O+ (black) densities are shown in the top plots, and temperatures are shown in the bottom plots. The vertical lines indicate the times of the conjunctions. …………47

5.1 Monthly averages of total ion density (rainbow panels) and fractional O+ (green and purple panels) measured by DMSP satellite F17 in 2009 at 17:30 LT. ……………………………….53

5.2 Monthly averages of total ion density for March, June, September, and December in 2008, measured by DMSP satellite F17 at 17:30 LT. A horizontal line through each panel approximates the latitude of peak densities (see Equation 5.1.) ………………………………55

5.3 Magnetic latitudes of the WN4 feature for March, June, September, and December in 2008, using data from F17 at 17:30 LT, showing how the peak ion densities move with season. These are the same lines overlaid on each panel in Figure 5.2. ………………………………………………56

5.4 Magnetic declination as a function of geographic longitude (GLON.) …………………………56

5.5 Total ion densities smoothed zonally using a 90-degree running median, for March, June, September, and December in 2008. The data is from DMSP satellite F17 at 17:30 LT. These monthly averages before smoothing are shown in Figure 5.2, and deviations from the running median are shown in Figure 5.6. ………………………………………………………………58

5.6 Deviations from the 90-degree running median (Figure 5.5) of total ion densities (Figure 5.2) for March, June, September, and December 2008. The data is from DMSP satellite F17 at 17:30 LT. ……………………………………………………………………………………………………59

5.7 Goodness of fit (R2) vs. month of the WN4 fitting procedure over a range of longitudes with moderate declination angle (-25.7 to 154.3 GLON). The data is from DMSP satellite F17 at 17:30-18:00 LT over the years 2008-2014. The trend is a polynomial fit to the R2 values. …60

5.8 Total ion densities measured by DMSP satellite F17 at 17:30-18:00 LT (top five panels) and F13 at ~18:15 LT (bottom two panels), averaged over the month of September, for seven years. The average value of the F10.7 solar index for each plot is given in parentheses. A horizontal line through each panel approximates the latitude of peak densities (see Equation 5.1.) ……………61

xiv

5.9 Fractional ion densities (dN/N) along the center of the WN4 feature (see Equation 5.1), for September during the seven years shown in Figure 5.8. The F10.7 solar index values are given in parentheses. …………………………………………………………………………………62

5.10 WN4 amplitude of September Ni during the seven years shown in Figure 5.8, vs. the monthly average of the F10.7 index. Amplitude has not been normalized for this figure. ………………63

5.11 WN4 amplitude of September Ni during the seven years shown in Figure 5.8, vs. the monthly average of the F10.7 index. Here, amplitude has been normalized by the mean Ni of the fit to account for changes in the background Ni due to solar cycle. …………………………………64

5.12 WN4 phase of September Ni during the seven years shown in Figure 5.8, vs. the monthly average of the F10.7 index. ……………………………………………………………………64

5.13 Magnetic latitudes of the WN4 feature for September of the same seven years in Figure 5.8, showing the variability in location of the peak densities over a solar cycle. The blue line shows the center line of the month with the lowest F10.7 value in this set (2008, F10.7=67.8), and the red line shows the center line of the month with the highest F10.7 value in this set (2001, F10.7=237.2) ……………………………………………………………………………………65

5.14 Total ion densities averaged over the month of September in six different years, measured by DMSP satellites (a) F13 at ~6:00 LT, (b) F18 at ~8:00 LT, and (c) F15 at ~9:30 LT. …………67

5.15 Total ion densities averaged over the month of September in 2014, measured by DMSP satellites F15 at ~15:48 LT, F17 at ~18:06 LT, and F18 at ~19:48 LT. …………………………………68

5.16 Estimated phase of WN4 for total ion densities measured by several DMSP satellites (see Table 5.2) during the month of September, showing the eastward trend of phase with evening magnetic local time. ………………………………………………………………………………………69

6.1 Histogram of the number of conjunctions with a given footprint (a) latitude difference and (b) longitude difference between satellites. …………………………………………………………75

6.2 (a) The magnetic local time (MLT) of all conjunctions over time. (b) Altitudes of the DMSP and C/NOFS satellites over time. (c) Geographic latitude and longitude of C/NOFS at all conjunctions. The conjunctions with a satellite altitude difference greater than 400 km are highlighted in red. ………………………………………………………………………………76

6.3 Percent error between IRI and the adjusted conjunction data vs. satellite altitude difference, separated into morning and afternoon local times. Gray scatter shows all the data points; blue trace shows a running median, with upper and lower quartiles as blue bars. …………………80

6.4 Percent error vs. magnetic local time, separated by satellite altitude difference. The panels are in the same format as Figure 6.3. ………………………………………………………………….81

6.5 Percent error vs. effective mass measured by DMSP, in the same format as Figures 6.3 and 6.4. The rows show different levels of solar cycle, represented by the F10.7 solar index. The columns separate the data into morning and afternoon local times. ……………………………83

6.6 Percent error vs. ion temperature (Ti) measured by DMSP, in the same format as Figure 6.5. …85

6.7 Percent error vs. electron temperature (Te) measured by DMSP, in the same format as Figure 6.5. ……………………………………………………………………………………………………86

6.8 (a) Ion temperature (Ti), (b) Electron temperature (Te), and (c) upward vertical ion drift (vz)as a function of magnetic local time (MLT), measured by DMSP. (d) Vertical ion drift (vz) vs. MLT measured by C/NOFS. …………………………………………………………………………87

xvi

LIST OF TABLES

5.1 Periods of on-orbit performance for the DMSP satellites and solar local time (SLT) of ascending node at launch/currently (or loss of function). ………………………………………51

5.2 Magnetic local time (MLT), amplitude prior to normalization (cm-3), phase, amplitude 2 normalized by mean Ni, and goodness of fit (R ) of WN4 fits used in this paper. All fits listed are for the month of September. ………………………………………………………………71

6.1 A quick reference for interpretation of % error. ……………………………………………….79

xvii

CHAPTER 1

INTRODUCTION

1.1 The Topside Ionosphere

The ionosphere is the ionized part of the Earth's upper atmosphere, which extends from about

90-1000 km in altitude. It is charge neutral as a whole and consists mainly of free electrons and positive ions, with some negative ions at lower altitudes. At middle and low latitudes, these electrons and ions come from photoionization due to solar radiation, mostly at extreme- ultraviolet and X-ray frequencies. At low altitudes, ions and electrons quickly recombine, but at high altitudes they persist for some time due to low recombination rates.

The ionosphere is conceptually divided into layers based on electron densities (Figure 1.1). The maximum electron density occurs at the F peak around 400 km. The altitude, density and shape of the F peak varies from day to night, with solar activity, and in response to geomagnetic activity. The "topside" is the part of the ionosphere above the F peak. The major ions in the topside are H+, O+, and He+. The primary reaction occurring at these altitudes that provides the main source and sink for H+ is the charge exchange reaction

O+ + H  H+ + O (1.1)

The major source of ionospheric O+ is photoionization and the primary loss mechanisms are reactions

O+ + e  O (1.2a) and

+ + O + N2  NO + N (1.2b) followed by

NO+ + e  N + O (1.2c)

Figure 1.1. Typical profiles of the neutral atmospheric temperature (left) and ionospheric plasma density (right.) [The Earth's Ionosphere, 1989]

The altitude where the ionosphere changes from O+ dominated to H+ dominated is termed the transition height. The O+/H+ transition height can vary significantly during the solar cycle, seasonally, and during geomagnetic storms [e.g., Garzón, et al., 2011; Hysell et al., 2009] due to thermospheric heating and winds, and electric fields. The other major constituent in the topside ionosphere is He+. The major source for He+ is photoionization while the major sink is charge

+ exchange with N2. As the density of N2 above the F2 peak is generally small, any He produced in the topside is relatively long-lived. During solar maximum, there are times when He+ can be the major light ion or even dominate [Heelis et al., 1990; Bailey and Sellek; 1990].

The density of each ion species in the topside can be roughly approximated by an exponential decay with altitude, where the decay rate is called a scale height H:

푘푇 퐻 = (1.3) 푚𝑔

H: scale height k: Boltzmann constant

T: temperature (or for a plasma, the sum of ion and electron temperatures) m: mass g: gravitational constant

A scale height is defined as the height over which the density of a particular ion, atom or molecule decreases by a factor of e. For example, because O+ is more massive than H+, O+ has a smaller scale height, and the concentration of O+ decreases more rapidly with increasing altitude than H+.

The neutral atmosphere is conceptually divided into layers based on temperature (Figure 1.1.)

The thermosphere is a region of the neutral atmosphere which overlaps with the ionosphere, characterized by a rapid increase in temperature with altitude. While temperatures in the lower atmosphere are relatively constant, thermospheric temperature varies significantly in response to solar heating.

In general, the rate of change of electron density is expressed by a continuity equation:

푑푁 = 푞 − 퐿 − ∇ ∙ (푁푣) (1.4) 푑푡

N: electron density t: time q: production rate

L: loss rate by recombination v: mean drift velocity

In certain regimes, one or more of the terms in the continuity equation may be approximately zero.

The extension of the ionosphere up to higher altitudes and into the inner magnetosphere is the plasmasphere. This is a toroidal region of high-density (10 to 103 cm-3), low-temperature (~1 eV) plasma which extends from about 1000 km up to the plasmapause, sometimes called the

"knee", a sharp outer boundary where the ion density drops by an order of magnitude. The magnetic field in the plasmasphere is mostly dipolar, and since the conductivity along magnetic field lines is very high, it is often useful to consider which regions are magnetically connected along field lines. For example, magnetic field lines at the plasmapause approximately map to the ionospheric main trough (mid-latitude trough) [Yizengaw et al., 2005], a plasma-depleted region near 60 to 65 degrees MLAT. The main trough and the plasmapause are both associated with the change in circulation pattern between the inner and outer magnetosphere, although their immediate causes and dynamics may differ from one another.

1.2 Geomagnetic Field Geometry, Winds and Effects on Topside

To a first approximation, the Earth's magnetic field is a dipole which is tilted and offset from the planetary spin axis. It is compressed by the solar wind on the day side, and streams away in a geomagnetic "tail" on the night side. In the ionosphere, the orientation of the geomagnetic field leads to ionospheric structure and phenomena that depend on latitude and longitude.

Figure 1.2. Dip latitude as a function of geographic latitude and longitude. [West and Heelis, 1996]

The magnetic field inclination, also called the dip angle, is defined as the angle between the magnetic field and a line tangent to the surface of the Earth (Figure 1.2.) A positive inclination means the magnetic field points toward the Earth. In contrast, magnetic declination is the angle between magnetic north and "true" north. Declination is defined to be positive when magnetic north is east of true north. For example, at middle and low latitudes, the geographic longitude range from about 180-270 has a positive declination, 300-360 has a negative declination, and the remaining longitude regions have close to zero declination.

The geometry of the geomagnetic field lines has a significant impact on the distribution of ionization in the topside ionosphere. Since the Lorentz force is zero for a charged particle moving parallel to the magnetic field, plasma tends to move very easily along geomagnetic field lines in regions where ion-neutral collisions are relatively infrequent, for example in the topside ionosphere where the neutral density is relatively low. When winds blow in the thermosphere, plasma is preferentially pushed by the component of the wind vector which is parallel to the geomagnetic field because mobility along the magnetic field lines is very high.

Meridional winds flow in a north-south direction, i.e. along a meridian; zonal winds flow in an east-west direction. Both meridional and zonal winds can act to redistribute plasma in the topside. For example, West and Heelis [1996] attributed longitude variations in O+ and H+ densities in the topside to a combination of zonal winds, meridional winds, and the offset between the geographic and geomagnetic equators.

Within the F region of the ionosphere, where ion-neutral collisions are infrequent, in response to any imposed electric field the ions and electrons will experience a motion perpendicular to both the electric and magnetic fields called "E cross B drift":

푬×푩 푣 = (1.5) |푩|2 v: drift velocity of plasma

E: electric field vector

B: magnetic field vector

This ExB drift is responsible for the electrodynamic lifting of plasma near the magnetic equator.

Electric fields in the E region, combined with the nearly horizontal magnetic field lines, are associated with upward ExB drifts during the day and downward ExB drifts during the night.

1.3 Solar EUV Origin, Variability, and Effects on Topside Ionosphere

1.3.1 Origin and variability

The solar EUV radiation ranges in wavelength from about 10 - 120 nm with a soft X-ray or XUV component from about 0.1 to 10 nm. It originates in the Sun's chromosphere, the transition region, and the corona and shows considerable variability with solar activity, solar rotation, and the solar cycle. The solar EUV spectrum is composed of numerous emission lines and continua whose irradiances can vary over several orders of magnitude with time scales from minutes to years [Woods and Rottman, 2002]. This variability is highly wavelength dependent with the lower wavelengths tending to exhibit greater variability (see, e.g., Woods and Rottman, 2002).

Solar activity varies on multiple time scales, most notably the 11-year solar cycle and the 27-day

Carrington rotation. A 27-day signal is visible in solar irradiance measurements when active regions on the Sun rotate in and out of view with a period of 27 days.

1.3.2 Measurements and models

Satellite observations of the solar EUV spectrum began in the 1960's with some daily observations being made by the AE-E spacecraft from 1976 to 1981 [Hinteregger, 1981]. Until the 1990's, there were only a few occasional measurements such as on the San Marco 5 satellite

[Tobiska et al., 1993] and some sounding rockets [i.e., Woods and Rottman, 1990]. Continuous

measurements began again in the 1990s with the Solar EUV Monitor (SEM) on board SOHO

[Judge et al., 1998], the Upper Atmosphere Research Satellite (UARS; Woods et al., 1996) and the Solar Radiation & Climate Experiment (SORCE; Rottman, 2005]. Unfortunately, these measurements were of little use for ionospheric or thermospheric studies, as they were either broadband (SEM) or did not properly cover the EUV range (SEM and UARS). In 2002, the launch of the Thermosphere-Ionosphere-Mesosphere Energetics and Dynamics (TIMED) spacecraft with the Solar EUV Experiment (SEE) grating spectrograph on board began providing spectrally resolved and nearly continuous measurements of the solar EUV irradiance [Woods et al., 2005]. The Extreme Ultraviolet Variability Experiment (EVE) onboard the Solar Dynamics

Observatory (SDO) spacecraft, launched in 2010, provides measurements at the highest spectral resolution to date with a high temporal cadence and a nearly 100% duty cycle at the shorter wavelengths.

A lack of simultaneous ionospheric satellite and solar EUV measurements has meant that most of the current understanding of the response of the ionosphere to solar variability has come from modeling studies. For instance, only very recently were sufficient interdisciplinary resources brought together to simulate the ionospheric response to EUV spectrum changes during a flare

(see, e.g., Pawlowski and Ridley, 2008; Smithtro and Solomon, 2008). These models rely on

EUV/XUV models for the solar irradiance spectrum which in turn rely on proxies for the solar irradiance, such as the F10.7 index because of the lack of adequate irradiance data. Although this parameter correlates reasonably well with the solar XUV/EUV flux, it clearly cannot replace it.

Several other proxies have been developed, but none of them can be used as a substitute for the

irradiance [Floyd et al., 2005]. Other recent EUV models include the SOLAR2000 model, which provides several indices that are tailored to specific needs, such as the E10.7 solar EUV proxy and the S10.7 index for the specification of the neutral density [Tobiska and Nusinov, 2006], and the NRLEUV model [Warren, Mariska, and Lean, 2001].

1.3.3 Ionospheric effects

Changes in the amount of solar EUV radiation result in corresponding changes in the energy balance of the upper atmosphere, dynamics, and photochemistry. Most of the solar energy deposited in the ionosphere and thermosphere is from radiation with wavelengths below about

100 nm. Indeed, Solomon et al. [2010] showed that the EUV is in fact the main driver of density changes in the ionosphere. Emmert, Lean, and Picone [2010] showed that the low EUV irradiance during the 2008 minimum phase led to an unprecedented decrease of the ionospheric density.

Solar EUV irradiance is also the primary source of energy for the upper atmosphere [Fuller-

Rowell, 2004]. Photoionization and photoabsorption cross sections for O+ are shown in Figure

1.3, where a larger cross section corresponds to more ionization or absorption, for a given EUV intensity and precursor species density. These wavelengths represent significant energy input, have high photoionization cross sections for O+, and are highly variable. EUV radiation creates the ionosphere through photoionization and drives changes in composition and temperature through heating. Many of these effects are most clearly seen in the topside ionosphere. For example, previous studies have found strong 27-day fluctuations in ion density at Defense

Meteorological Satellite Program (DMSP) altitudes (~840 km) [Rich et al., 2003] and measured by the Communication/ Navigation Outage Forecasting System (C/NOFS) spacecraft (~450 to

850 km) [Coley and Heelis, 2012] in response to the 27-day solar rotation. These 27-day density

Figure 1.3. Weighted photoionization and photoabsorption cross sections for O+ for Torr and Torr [1979] bands and lines.

variations are not generally observed at altitudes lower than about 500 km, possibly because of the greater influence of dynamical processes at lower altitudes. Additionally, density fluctuations measured by DMSP are much stronger during high solar activity, in contrast to the saturation effect of ion densities at the F peak (NmF2) and of total electron content (TEC) [Liu et al., 2007]. Furthermore, the scale height itself in the topside ionosphere has been observed to increase during high solar activity, so that the density variation with altitude is not simply a linear response.

Other studies have examined the response of the topside ionosphere to solar activity across longitude, latitude, and season [e.g. Greenspan et al., 1994; West and Heelis, 1994].

Longitudinal variations have been explained in terms of field-aligned transport. The components of meridional and zonal neutral winds parallel to the Earth's magnetic field lines tend to move plasma along the field lines as a result of pressure gradients produced by ion-neutral collisions at lower altitudes, and this effect varies with season and longitude. There are also strong latitudinal variations which depend on season due to the changing solar zenith angle.

1.4 Atmospheric Tides and Coupling with Ionosphere-Thermosphere

1.4.1 WN4 and DE3

Atmospheric tides are significant drivers of global ionospheric and thermospheric structure. In particular, a wave-number 4 oscillation (WN4) along the magnetic equator has been studied in several parameters including total electron content (TEC) [Scherliess et al., 2008; Wan et al.,

2008; Pedatella et al., 2011], thermospheric neutral winds [Häusler et al., 2007; Häusler and

Lühr, 2009], [O]/[N2] ratio [He et al., 2010], distribution of NO [Barth et al., 2003], the equatorial electrojet current density [Lühr et al., 2008], and electron densities [Liu and

Watanabe, 2008; Pedatella et al., 2008]. There is strong evidence that WN4 patterns in the ionosphere/thermosphere are associated with the DE3 non-migrating tide which originates in the lower atmosphere [Immel et al., 2006; Hagan et al., 2007; Wan et al., 2010], and the chain of tidal propagation extending from the troposphere up into the thermosphere and ionosphere has been studied extensively in recent years.

DE3 is the diurnal (24-hour period) component of an eastward propagating tide with zonal wave number 3. It has three peaks equally spaced in longitude when viewed in a constant universal time frame, whereas the WN4 pattern has four peaks equally spaced in longitude when viewed in a constant local time frame, due to the change in coordinates. The following equations express a tidal perturbation at a given latitude, in universal time (UT) or local time (LT), including its frequency and how to convert between UT and LT:

푓푈푇 = ∑ 퐴푠,푛cos⁡[휔푛(푡푈푇 − 휙푠.푛) − 푠휆] (1.6)

푓퐿푇 = ∑ 퐴푠,푛cos⁡[휔푛(푡퐿푇 − 휙푠.푛) − (푠 − 푛)휆] (1.7)

2휋푛 휔 = (1.8) 푛 24⁡ℎ표푢푟푠

24⁡ℎ표푢푟푠 푡 = 푡 − 휆 (1.9) 푈푇 퐿푇 360°

A : amplitude

ω : frequency tUT : universal time tLT : solar local time

φs,n: phase shift in time s : zonal wave number with positive (negative) values representing westward (eastward) phase propagation

λ : east longitude n=1 for diurnal, n=2 for semidiurnal, n=3 for tridiurnal, etc.

For DE3, n=1 and s=-3. Thus in a fixed-local time frame (LT), the observed zonal wave number of DE3 is |s-n|=|-3-1|=4, consistent with WN4. However, there are other combinations of n and s which would also look like WN4 in the LT frame, for example, the semi-diurnal component of an eastward-propagating tide with zonal wave number 2 (SE2) in the UT frame (|s-n|=|-2-2|=4), or the diurnal component of a westward-propagating tide with zonal wave number 5 (DW5) in the UT frame (|s-n|=|5-1|=4.) Therefore, there is degeneracy when converting between the two reference frames. Unfortunately, satellite observations alone often have difficulty resolving this degeneracy due to limited coverage.

DE3 is generated by latent heat release of evaporation during cloud formation [Hagan and

Forbes, 2002, 2003], as the atmosphere responds to periodic absorption of solar radiation. The distribution of the peaks is due to the land-sea distribution, i.e. three land masses plus a void over the Pacific Ocean, which affects the longitudinal distribution of convection patterns and cloud formation [Tsuda and Kato, 1989; Williams and Avery, 1996.]

1.4.2 DE3 propagation, WN4 in the F-region, and impact of WN4 on the upper atmosphere

Immel et al. [2006] proposed a physical mechanism for the propagation of DE3 into the F-region based on the ExB drift. They suggested that upward-propagating tides modulate the dynamo electric fields in the E-layer, which is associated with the ExB drift of plasma lifted up into the

F-region. This effect would be maximized at equinox, when the meridional component of E- region neutral winds at low latitudes is not dominated by inter-hemispheric flows. Jin et al.

[2008] simulated DE3 effects on the longitudinal structure of F-region zonal electric fields, in

order to investigate how this tide might modify the vertical ExB drift. They noted that under a realistic geomagnetic field configuration, longitude sectors with large declination angles would show a larger hemispheric difference in the phase of DE3.

Some additional coupling mechanisms have been suggested. Winds associated with non- migrating tides at F-region altitudes can push plasma along field lines and modify the F-region dynamo [England et al., 2010, Park et al., 2010.] Non-migrating tides have also been shown to

+ modify the [O]/[N2] ratio, which affects the concentration of O since an increase in [O] allows

+ + more production of O at a given altitude, and decreases in [N2] increase the lifetime of O

[Zhang et al., 2010.] It should also be noted that WN4 patterns can be enhanced by additional tides besides DE3 such as SPW4 (stationary planetary wave-4) and SE2 (semi-diurnal eastward- propagating wave-2) [Pancheva and Mukhtarov, 2010.]

Although many migrating and non-migrating tides are produced at lower altitudes and propagate upward, DE3 has a particularly significant impact on higher altitudes because it has a large amplitude, and because it propagates upward very efficiently. The efficient propagation of DE3 is due to its long vertical wavelength, which is possible because its origin in convective clouds extends throughout the troposphere [Williams and Avery, 1996]. Hagan and Forbes [2002, 2003] found that DE3 and SW6 (semi-diurnal westward-propagating wave-6) are the largest amplitude non-migrating tides that should reach the upper mesosphere/lower thermosphere using the

Global Scale Wave Model (GSWM). They also showed that the DE3 tide would be very effective at modifying the E-region dynamo because the DE3 wavelength is large compared to

the Hall conductivity layer and has a symmetric latitudinal structure. As tides propagate upward and the atmospheric density decreases, they are expected to increase in amplitude [Williams and

Avery, 1996].

1.4.3 Previous measurements and results in the topside ionosphere

While DE3 and WN4 have been observed and modeled extensively in the mesosphere and lower thermosphere regions in recent years, relatively few studies have investigated WN4 patterns in the topside ionosphere. WN4 was first observed in the F region equatorial ionization anomaly

(EIA) by Sagawa et al. [2005] in IMAGE/FUV satellite data. More recently, tidal signatures at low latitudes near 600 km have been observed by ROCSAT-1 in ion densities [Kil et al., 2008] and ExB drift [Kil et al., 2007; Ren et al., 2009; Fejer et al., 2008.] Bankov et al. [2009] compared F13, F15, and DEMETER ionospheric measurements during fall equinox 2004, at local times near 5:45 (F13), 9:30 (F15), 10:30 (DEMETER), 17:45 (F13), 21:30 (F15), and 22:30

(DEMETER.) They found that WN4 was almost a regular feature in O+ at the local times examined, as well as in H+ in evening through pre-midnight hours and in He+ in the morning.

Hartman and Heelis [2007] considered ion drift measurements made by F15 in 2001 and 2002 near 9:30 LT. They attributed a seasonally-dependent longitude variation to hemispheric differences in the dynamo current due to the different orientation of magnetic meridians to the day-night terminator. They also saw a WN4 pattern present throughout the year, but especially apparent during equinox. Ren et al. [2008] studied longitude variations in electron temperature

(Te) and ion density (Ni) measured by F13 from 1995-2005 near sunset. They attributed a seasonal change in the number of peaks (4 peaks at equinox, 3 peaks in June solstice, and 2

peaks in December solstice) to interactions between zonal and meridional winds and the difference in geomagnetic field declination with longitude. They also noted anti-correlations between Te and Ni.

1.5 Modeling of Electron Density Profiles

1.5.1 Scale height and transition height in the topside and modeling of topside electron density profiles

The ability to characterize and model vertical profiles of electron densities (Ne) and ion densities in the topside ionosphere is particularly important because the topside represents the main contribution to total electron content (TEC) measurements [Jakowski et al., 2002].

Reconstruction of the Ne profiles typically involves fitting some analytical shape to measurements. Several analytical shapes have been used to model the electron density profile in the topside such as Chapman, exponential, parabolic, and Epstein layer (sech-squared) functions.

An important parameter in the Chapman and related functions is the scale height (mentioned in

Section 1.1), defined as the altitude over which electron density in diffusive equilibrium decreases by a factor of e. For an idealized Chapman profile in diffusive equilibrium, the plasma scale height is directly proportional to the sum of the ion and electron temperatures, and inversely proportional to the effective mass. During the daytime, the ion temperature (Ti) increases with altitude from approximately equal to the neutral temperature (Tn) near the F-peak to approaching the electron temperature (Te); Te also tends to increase with altitude. The effective ion mass in the topside tends to decrease with altitude as the composition becomes more light ion-dominant. Therefore, the topside plasma scale height is highly variable, and can be expected to increase with altitude during the day. Nevertheless, a constant scale height is

often used to model the topside Ne profile in Chapman-type functions [e.g. Reinisch and Huang,

2001; Liu et al., 2014; Tulasi Ram et al., 2009 and references therein.] Some authors have used a linearly-varying scale height, or a scale height that changes with altitude based on some shape function, for example in the Vary-Chap function [Reinisch et al., 2006, Nsumei et al., 2012].

Another important parameter is the O+ to light ion transition height, i.e. the altitude where the concentration of O+ ions is half the total ion concentration. We showed that the topside ionosphere has a stronger response to solar energy input when it is O+ dominated than when it is

H+ dominated, due to the difference in scale height (see Chapter 4.) The composition, scale height, and transition height significantly influence the variability of the topside ionosphere, and therefore accurate modeling of these parameters is an important challenge in ionospheric physics.

1.5.2 Methods of topside ion density profile reconstruction and estimation of the topside scale height

Heelis et al. [2009] studied the O+/H+ transition height during the low solar minimum of 2008 and 2009 using in situ composition measurements from C/NOFS and found that the transition height occurred at a lower altitude and was significantly cooler with a lower density than predicted by the International Reference Ionosphere (IRI) model. The transition height (Ht) from in situ C/NOFS measurements was further studied by Tulasi Ram et al. [2015], who found that the latitudinal shape of Ht had a local minimum at the equator in 2008-2009 and at some local times in 2010, which they attributed to the O+/H+ charge exchange reaction.

Where direct in situ satellite measurements are not available, the topside electron density or ion density profiles have been estimated using radio occultation measurements (RO) from COSMIC

[e.g. Liu et al., 2008, Yang et al., 2016], GPS-based total electron content (TEC) [Stankov et al.,

2003], topside sounders [Webb et al., 2006], and combinations of these with in situ satellite measurements and/or ionosondes [Reinisch and Huang, 2001, Tulasi Ram et al., 2009,

Venkatesh, 2011, Liu, 2014]. Each method has advantages and disadvantages, and there is no general consensus as to the “best” method for obtaining the topside Ne or Ni profile shape.

It has been known for some time that the modeling of topside Ne profiles and associated parameters such as scale height and transition height depend on the method used. Each method is limited by what data sources are available, as well as assumptions inherent to the chosen profile shape, such as a constant or linearly varying scale height. However, few studies have directly compared the effectiveness of the different methods. Belehaki et al. [2005] compared the scale height from a model of topside sounders with scale heights derived from bottomside digisondes using the Reinisch and Huang (R-H) method, and found that while the overall magnitudes were similar, the R-H method variations were influenced by variations in thermospheric temperature. Tulasi Ram et al. [2010] used ion densities from C/NOFS during the low solar minimum to compare a transition height derived from fitting topside ion concentration profiles using an alpha-Chapman function, with the transition height obtained directly from in situ ion composition measurements. The direct in situ transition heights showed similar local time and latitude patterns as the equatorial ionization anomaly (EIA) and seemed consistent with interhemispheric neutral wind patterns; however, the transition heights from the fitted profiles

were generally lower and showed inconsistent variations, at least in part because they failed to account for the change in scale height with altitude. Yang et al. [2016] fitted an O+ alpha-

Chapman profile with linearly variable scale height to COSMIC RO data combined with

C/NOFS data, and found that these results better matched the in situ C/NOFS observations of transition height than using the COSMIC RO data with a constant scale height. These studies suggest that while several methods have been used to estimate the topside Ne and Ni profile shapes, further work is needed to evaluate these methods, and in particular to compare topside model results with in situ satellite measurements such as those provided by C/NOFS and DMSP.

CHAPTER 2

SCIENTIFIC OBJECTIVES AND RELEVANCE

2.1 Research Overview

This dissertation consists of three investigations: the examination of two important sources of variability in the topside ionosphere, the solar extreme-ultraviolet (EUV) irradiance and atmospheric non-migrating tides, and the ability of the International Reference Ionosphere (IRI)

2016 empirical model to accurately model the topside ionospheric scale height. We investigated and quantified how these three factors vary across local time, season, and varying levels of solar activity.

For the investigation of the response of the topside ionosphere to solar EUV variability, we present an analysis of 23 years of thermal plasma measurements in the topside ionosphere from the DMSP spacecraft. The H+/O+ ratio and density vary dramatically with the solar cycle; cross- correlation coefficients between E10.7 and the daily averaged densities are greater than 0.85. The ionospheric parameters also vary dramatically with season, particularly at latitudes away from the equator where the solar zenith angle varies greatly with season. There are also 27-day solar rotation periodicities in the density, associated with periodicities in the directly measured solar

EUV flux. Empirical orthogonal function analysis captures over 95% of the variation in the density in the first two principal components. The first principal component (PC1) is clearly associated with the solar EUV while the second principal component (PC2) is clearly associated with the solar zenith angle variation. The magnitude of the variation of the response of the topside ionosphere to solar EUV variability is shown to be closely related to the ionospheric

composition. This is interpreted as the result of the effect of composition on the scale height in the topside ionosphere. When the topside ionosphere is H+ dominated during solar minimum,

DMSP may be much less than a scale height above the F2 peak while during solar maximum, when it is O+ dominated, DMSP may be several scale heights above the F2 peak.

As mentioned in section 1.4, non-migrating tides are a major coupling mechanism between the different regions of the atmosphere and ionosphere. The wave-number 4 (WN4) pattern in the ionosphere has been recognized as originating primarily with the diurnal eastward propagating wave-3 non-migrating tide (DE3) in the troposphere, and significant effort has been devoted in recent years to understanding how tidal effects manifest in various physical parameters across a very wide range of altitudes. While DE3 and WN4 signatures have been much studied in the mesosphere-lower thermosphere region, relatively little is known about how WN4 impacts the ionosphere above the F-peak. We present an analysis of WN4 in the topside ionosphere as measured by the DMSP spacecraft, using monthly averages of total ion densities and composition binned by latitude and longitude. We found that WN4 is most strongly present during September equinox. In May-August, ion densities near 180-270 GLON are enhanced, and the WN4 pattern moves 5-10 degrees north; in November-February, ion densities are reduced in this region, and WN4 moves 5-10 degrees south. No solar cycle effects were found in the magnitude of dN/N. The longitude position of the peaks (phase) was observed to vary by about 10° with F10.7, moving eastward with increasing F10.7. The latitude variation was less than 5° and did not show a trend with F10.7. The WN4 pattern changes rapidly near dawn, but is

very constant throughout the afternoon and evening in terms of dN/N, and drifts eastward at about 2o GLON per hour MLT.

IRI is an international standard ionospheric model and an essential resource to space scientists and engineers. Studies have shown that although the topside ionosphere is an important contributor to total electron content (TEC), models struggle to accurately reconstruct topside electron density profiles due to the limited availability of in situ measurements. We address this need with a new data set of over 11,000 magnetic conjunctions between DMSP and C/NOFS from 2009 through the C/NOFS orbital decay at the end of 2015. We investigate the error of the most recent version of IRI in reproducing the topside profile shape at various local times and under different geophysical conditions. The model was found to underestimate the Ne profile slope and plasma scale height in morning but matched the profile shape well in the afternoon

(1400-1800 MLT.) Percent error decreased with increasing effective mass, particularly when

F10.7 was >100 SFU. Percent error increased with increasing Ti and Te in the afternoon but did not show a linear trend with Ti or Te in the morning. This is consistent with a diffusive equilibrium condition in the afternoon, whereas strong ion flows in the morning produce a more complex relationship between the Ni profile shape and temperatures.

2.2 Relevance to Research Community

The investigation of how the solar EUV irradiance drives variability in the topside ionosphere across varying levels of solar activity, and in particular how this variability depends on the ionospheric composition is very relevant to the space science research community. This research

addresses objectives in NASA's Recommended Roadmap for Science and Technology 2005-

2035: Objective H: "identify impacts of solar variability on Earth's atmosphere;" "H3:

Understand the role of the Sun as an energy source to Earth's atmosphere and, in particular, the role of solar variability in driving change."

Studies of the ionospheric response to EUV typically use solar EUV proxies such as F10.7, the

10.7 cm radio flux measured at Penticton, Canada, or MgII, the core-to-wing ratio of the MgII spectral line at 280 nm. While the use of proxies such as F10.7 and E10.7 capture the overall variation in the solar EUV and are valuable tools in modeling studies, they cannot capture the spectral variations in the solar EUV on a wavelength-by-wavelength basis. This is of course why solar EUV monitors such as SEE and EVE are flown on spacecraft. We used the spectrally- separated EUV data measured by the SDO and TIMED satellites, since the proxies do not capture the full variability of spectrally-separated EUV.

The investigation of the variability of the wave-number 4 (WN4) pattern, is also very relevant.

The National Research Council’s Decadal Survey for the 2013-2022 Decade identified

“Determine the dynamics and coupling of Earth’s magnetosphere, ionosphere, and atmosphere and their response to solar and terrestrial inputs” as a Key Science Goal. More specifically, the

Panel on Atmosphere-Ionosphere-Magnetosphere Interactions (AIMI) identified “Understand how tropospheric weather influences space weather” as one of three AIMI science priorities.

While DE3 propagation and WN4 signatures have been much studied in recent years in the mesosphere-lower thermosphere region, relatively little is known about how WN4 impacts the

ionosphere above the F-peak [Bankov et al. 2009]. A better understanding of atmospheric tides as a driver of variability in the topside ionosphere will yield important clues to whole- atmosphere coupling.

Finally, the investigation of the ability of the IRI 2016 empirical model to accurately model the ion and electron density profiles in the topside region, and the dependence of the error on various conditions and parameters, is important because of the large contribution of the topside ionosphere to the total electron content (TEC) [Jakowski et al., 2002], which affects GPS measurements. A special issue of Advances in Space Research [Reinisch and Bilitza, 2017] compared IRI-2012 with recent data sources in the equatorial and low latitude region. They concluded that more work is needed to better understand and model the topside electron density profiles.

CHAPTER 3

DATA SOURCES AND MODELS

3.1 TIMED SEE

The TIMED spacecraft was launched in December of 2001 into a circular orbit at 625 km. The

Solar Extreme-ultraviolet Experiment (SEE) aboard the TIMED spacecraft has provided solar

EUV measurements in the 0.1-195 nm range continuously since January 22, 2002. SEE contains an XUV Photometer System (XPS) and EUV Grating Spectrometer (EGS), and the data has been processed into a Level 3 data product [Woods et al. 2005]. SEE spectra in the range 7-27 nm are extrapolated from a single broad-band photodiode measurement, so EUV measurements in this critical wavelength range are supplemented with new measurements from SDO EVE.

3.2 SDO EVE

SDO (Solar Dynamics Observatory) was launched in February of 2010 into a circular, geosynchronous orbit (~36,000 km with a 28.5o inclination) as part of the Living With a Star

(LWS) program. The Extreme ultraviolet Variability Experiment (EVE) aboard the SDO spacecraft measures radiation in the 0.1-105 nm range with a 0.1 nm spectral resolution. Together,

TIMED and SDO provide measurements of solar flux at a wide range of wavelengths at high resolution in the XUV/EUV range with some overlap [Woods et al. 2009].

3.3 DMSP

The DMSP spacecraft fly in sun-synchronous, dawn-to-dusk and pre-midnight to pre-noon orbits at ~800 km altitude with ~99o inclinations and orbital periods near 100 in. This orbit means that

DMSP measurements from individual satellites are taken at nearly constant solar local time. The

Special Sensor-Ions, Electrons and Scintillation (SSIES) instrument package [Greenspan et al.,

1986] consists of an Electron Probe (EP), Retarding Potential Analyzer (RPA) , Ion Drift Meter

(DM), and Scintillation Meter (SM) in order to monitor the behavior of thermal plasma in the topside ionosphere. These instruments have been on every DMSP spacecraft since F10 was launched in 1990, and at times there have been up to 5 DMSP satellites operating in orbit at once, so DMSP represents a valuable long-term data set.

The RPA, SM, and DM are the instruments of interest for the studies here. The RPA consists of a series of biased grids in front of a collector. One or more grids are biased to reject a portion of the distribution of charged particles approaching the instrument, and the rest of the distribution is collected as current. Sweeping the bias voltage produces a current-voltage (I-V) curve, from which it is possible to derive a total ion density; fractional components of O+, H+, and He+; an average ion temperature, and the component of the ion drift velocity parallel to the spacecraft velocity vector. The current-voltage curves on the DMSP spacecraft are produced at a normal rate of one every four seconds for F11 – F15 and every second for F16 – F19.

The SM is similar to the RPA expect without the swept bias voltage. The total ion density is thus determined at a higher sampling rate than the RPA (12 Hz). The data used in this dissertation are one-second averages of the SM data.

The DM consists of a gridded collimator and segmented collector. The ratio of the collected current on opposite sides of the collector is used to determine the angle between the ambient ion velocity vector and the spacecraft velocity vector. Knowledge of the spacecraft velocity and attitude allows determination of the ion velocity vector in the plane perpendicular to the spacecraft velocity vector. The DM data has a sampling rate of 6 Hz for the horizontal and vertical ion drift components and is averaged to one-second for the studies here.

3.4 C/NOFS

C/NOFS was launched in April 2008 into an elliptical orbit with a perigee near 450 km, apogee near 850 km and a 13o inclination. It reentered the atmosphere in November, 2015. The Coupled

Ion Neutral Dynamics Investigation (CINDI) onboard C/NOFS carries an RPA and a DM similar to those on DMSP.

3.5 IRI

The International Reference Ionosphere (IRI) is an industry-standard empirical model maintained by the Committee on Space Research (COSPAR) and the International Union of Radio Science.

The model goes through regular cycles of review by the research community, who compare model predictions with recent measurements and make recommendations for improvement. The latest version, IRI-2016, is reviewed by Bilitza et al. [2017]. IRI provides electron densities, ion composition, and ion and electron temperatures, as well as vertical total electron content (TEC), at a user-specified date, time, and location (up to 2000 km altitude.) IRI-2016 includes two new

models for the F2 peak electron density, the height of the F2 peak (hmF2), and improvements to the ion composition at low and high solar activity.

CHAPTER 4

TOPSIDE IONOSPHERIC RESPONSE TO

SOLAR EUV VARIABILITY

The following work on the long-term, global effects of solar extreme-ultraviolet radiation (EUV) on the topside ionosphere is reprinted with permission from Wiley. It was published in the

Journal of Geophysical Research: Space Physics article "Topside ionospheric response to solar

Geophysical Union. All Rights Reserved.) We presented an analysis of measurements of ionospheric density, composition, and temperature in the topside ionosphere by the DMSP spacecraft near 800 km for two full solar cycles. These measurements provided a unique opportunity to examine the influence of solar EUV variability on the topside ionosphere over a period of time never accomplished before. The observed variations were interpreted in terms of the variability in solar EUV using the E10.7 solar EUV flux proxy as well as high-resolution solar EUV measurements from the TIMED SEE and SDO EVE instruments.

4.1 Data

To examine the influences of the solar EUV variability on the topside ionosphere, we analyzed the DMSP data for two full solar cycles, from 1992 through 2014. We produced daily averaged values of the ionospheric density, H+/O+ ratio, and ion temperature in 4-degree geographic latitude (GLAT) bins from 60° south to 60° north. Although the motions of the ionospheric plasma tend to be organized geomagnetically, we choose geographic coordinates as the solar zenith angle (SZA) is related to GLAT.

As the DMSP spacecraft fly in sun-synchronous, dawn-to-dusk and post-dusk to post-dawn orbits, one (or more) satellite is always in one of four specific local time sectors: ~1800 ascending node solar local time (ANSLT - Dusk), ~0600 ANSLT (Dawn), ~2000 ANSLT (post dusk), and ~1000 ANSLT (post dawn). Figures 4.1 and 4.2 show examples of the daily averaged

(a) ionospheric density, (b) H+/O+ ratio, and (c) ion temperature along with the (d) solar zenith angle from the dawn/dusk spacecraft in the dawn (Figure 4.1) and dusk (Figure 4.2) sectors for

Figure 4.1. Examples of the daily averaged (a) ionospheric density, (b) H+/O+ ratio, and (c) ion temperature along with the (d) solar zenith angle in the dawn sector for two complete solar cycles from 1992 through 2014. The data are from three spacecraft - DMSP F11 (blue), F13 (red), and F17 (green). The E10.7 solar EUV proxy is shown in black in (a).

Figure 4.2. Daily averaged values in the dusk sector in the same format as Figure 4.1.

two complete solar cycles from 1992 through 2014. The data are from three spacecraft - DMSP

F11 (blue), F13 (red), and F17 (green) - in the GLAT range between 3 and 7 N. There was a small seasonal variation in the SZA plus a small overall drift of SZA (ANSLT) in time for individual satellites (caused by atmospheric drag) and a difference between satellites (the satellites are not launched with exactly the same ANSLT). We primarily showed the data from the dawn/dusk spacecraft. Similar results were obtained from the data from the post-dusk to post- dawn spacecraft but there were almost three fewer years of data available from those spacecraft

(starting in late 1994) and the consistency in ANSLT was not as good. The E10.7 solar EUV proxy is shown in black in (a). There is obviously a strong correlation between the ionospheric density and the E10.7 over the solar cycle, with extreme variations in the E10.7 driving extreme variations in the ionospheric density. The cross correlation coefficient (CCC) between the density and E10.7 was 0.86 in the dawn sector and 0.85 in the dusk sector.

There was also a clear solar cycle variation in the H+/O+ ratio as shown in Figures 4.6b and 4.7b.

During solar maximum, the ionosphere at DMSP altitude (~800 km) is typically dominated by

O+ while during solar minimum, it is dominated by H+. As mentioned in the introduction, the

H+/O+ transition height can vary significantly during the solar cycle, seasonally, and during geomagnetic storms [e.g., Garzón, et al., 2011; Hysell et al., 2009] due to thermospheric heating and winds, and electric fields. This transition height can be above or below the DMSP spacecraft depending on these drivers. The solar output was extremely low during the last solar minimum leading to a very cold thermosphere and transition heights well below DMSP altitudes. This can be seen in the H+/O+ ratio, particularly in the dawn sector during the last solar minimum; the ionosphere at DMSP was almost completely H+ dominated. H+ was not found to dominate as much in the dusk sector but still showed the solar cycle dependence and was more dominant in the last minimum, generally making up more that 80% of the ionosphere at DMSP altitudes.

The ion temperature, shown in Figures 4.1c and 4.2c, was found to vary with the solar EUV, but did not show the solar cycle dependence the density and the composition did. However, there was a strong dependence on SZA, even for the small variations in SZA seen near the equator in

Figures 4.6d and 4.7d. This variation was particularly apparent after 2007 in the dawn sector.

The data is from the F16 spacecraft, which was launched into a slightly earlier ANSLT (such that the SLT on the dawn side is earlier) - notice the discontinuity in SZA. Thus the SZA was greater on the dawn side. The large dips in the ion temperature during the years ~2007 - 2011 are likely because the satellite was crossing the terminator into darkness.

Figure 4.3 shows the time shifted CCCs for both the dusk sector and the dawn sectors. The maximum CCC occurred at a time shift of one day indicating that the greatest response in the topside ionosphere to changes in the solar EUV occurs with a one-day time lag. This is

Figure 4.3. Time shifted CCCs between density and E10.7 for both the dusk (black) and dawn (red) sectors.

similar to the results from others [e.g., Rich et al. 2003; Coley and Heelis 2012]. Notice that there were several CCC peaks at other time lags. The peaks were separated by ~27 days and are

related to the solar variability associated with the 27-day solar rotation. This is illustrated in

Figure 4.4. (a) The daily averaged densities (black) for a 180-day period in 2003 along with the solar EUV flux at 28 nm (red) measured by the SEE instrument on the TIMED spacecraft, the E10.7 (blue - scale not shown), and the SZA (b) The same format as (a) but for a 200-day period in 2013 with the daily averages of the SDO EVE-measured fluxes at 25.5 nm.

Figures 4.4a and 4.4b. Figure 4.4a shows the daily averaged densities from the DMSP F15 spacecraft for a 180-day period in 2003 along with the E10.7, the solar EUV flux at 28 nm measured by the SEE instrument on the TIMED spacecraft, and the SZA. This data is from the post-dawn sector (F15 is in a post-dusk to post-dawn orbit) and thus the SZA is smaller. The

variation in the SZA during this period naturally creates a variation in the density that is overlaid on the variation due to the solar EUV. However, despite the SZA effects, the solar rotation variability was found to dominate and the CCC between the solar EUV and the density was very high at 0.9. The CCC between the E107 and the density was only slightly lower at 0.88. Figure

4.4b shows the daily averaged densities from the DMSP F17 spacecraft for a 200-day period in

2012 along with the E10.7, the solar EUV flux at 25.5 nm measured by the EVE instrument on the SDO spacecraft, and the SZA. In this case the correlation with the EVE data was 0.86 while the correlation with E10.7 was much less at less than 0.73. This illustrates that, while solar EUV proxies such as E10.7 are extremely useful, direct measurements of the solar EUV are critical for accurate modeling of the ionosphere.

The data presented so far have been from near the equator to mitigate the effects of the seasonal variations in SZA on the ionosphere in the analysis. Figures 4.5, 4.6, and 4.7 show the data in a similar format to Figures 4.1 and 4.2 but for various GLATs in the northern and southern hemispheres - black for southern hemisphere and red for northern. Also, the scale for the H+/O+ ratio is smaller by a factor of ten. The red and black bars at the top of the plots show when the

H+/O+ ratio exceeds 20. This data is only from the dawn sector. The seasonal effects were seen to naturally become larger at higher latitudes as the variation in SZA increases. There was a significant hemispherical asymmetry in the SZA and thus the density variations, particularly at higher latitudes. This asymmetry was driven by the axial tilt of the Earth and the sun- synchronous nature of the DMSP orbits. The DMSP orbit at 99, with ascending node in the evening/dusk side, means that the orbit plane was always tilted toward the sun in the northern

hemisphere and away from the sun in the southern hemisphere. Thus the SZA was smaller in the northern hemisphere during the solstices than in the southern hemisphere. This also leads to a dichotomy in the density, composition, and ion temperature measurements, particularly during winter solstice. The densities were lower in the southern hemisphere and the H+/O+ ratio was larger.

Figure 4.5. Daily averaged values in the dawn sector in the same format as Figure 4.1 but for +/- 5 GLAT.

Figure 4.6. Daily averaged values in the dawn sector in the same format as Figure 4.5 but for +/- 30 GLAT.

Figure 4.7. Daily averaged values in the dawn sector in the same format as Figure 4.1 but for +/- 50 GLAT.

There was a strong seasonal variability in the measurements due to the SZA variations along with the solar cycle variation. This was most clearly seen in the measurements at +/- 30° GLAT.

The seasonal variation fluctuated with the solar cycle and was clearly seen in all of the measurements but most dramatically in the composition measurements. H+ dominated during winter solstice in both hemispheres but was more dominant in the southern hemisphere. This dominance was greater during solar minimum than solar maximum and was more apparent in the last solar minimum into the last solar maximum than in the previous cycle. Notice that there were large dips in the ion temperature during the solar minimums. These dips did not occur during solar maximum even though the SZAs were similar.

4.2 EOF analysis

Even small variations in the SZA can cause significant changes in the ionization. That variation becomes dramatic at high latitudes but is still present near the equator. (See Figures 4.1, 4.2, and

4.5.) Figure 4.8 shows the daily averages of the SZA and the density from the DMSP F15 and

Figure 4.8. Daily averages of the (a) SZA and the (b) density from the DMSP F15 (black) and F16 (red)spacecraft for 250 days during 2005.

F16 spacecraft for 250 days during 2005. The 27 solar-ration variation is clearly visible. The difference in the SZA varied from about 11 to about 8 while the density difference varied from about 25% to about 10%. To first order, assuming the SZA dependence as predicted by the

Chapman ionization theory [Chapman, 1961; Yonezawa, 1966], the ionization should vary as the exponent of the secant of the SZA. In order to develop an improved metric of the behavior of the plasma around the peak of the F2 layer and remove the effects of the SZA on ionization,

Gulyaeva [2009] used ionosondes to derive a SZA correction factor to the ionospheric peak density. They found the correction to be a nonlinear function of the inverse of the SZA. The correction factor requires a normalization factor equal to the ratio of the SZA to the SZA at local noon.

We used Empirical Orthogonal Functions (EOFs) to separate the effects of variations in the solar

EUV flux and the effects of SZA variations in the analysis. EOFs provide a way to achieve this separation that depends purely on the data. This method is a decomposition of a signal or data set in terms of orthogonal basis functions, which are determined from the data. EOFS (or natural orthogonal components) have long been used in the atmospheric sciences [see review by

Hannachi et al., 2009] and have recently been used in the ionosphere. The method of natural orthogonal components (MNOC – Kendall and Stuart [1976]) has been used to separate out the structure of the geomagnetic field into longer and shorter wavelengths [Frynberg, 1975;

Rotenova et al., 1982], to separate geomagnetic field variations into quiet and disturbed components [Golovkov et al., 1978], and to generate a data based model of the geomagnetic field in space and time [Golokov et al., 2007]. Sun et al. [1998] used MNOC to separate the directly driven and unloading components of the ionospheric currents during substorms. A et al., [2012] used EOF analysis to construct a global ionospheric total electron content model from GPS data and was able to represent the variance in the data using only four modes. Zhao et al., [2005] used EOF analysis to decompose DMSP density data over an 8-year period into a time mean plus the sum of orthogonal functions multiplied by time-varying coefficients. They found that 95% of the variance in the data could be explained by the first three EOFs. The first two EOFs were

clearly associated with the solar EUV variation (as determined by F10.7) and the solar declination (or SZA) and were the dominant modes controlling the density variability. They interpreted the third EOF as being due the effects of wind patterns and the equatorial electric fields.

We applied such analysis to the DMSP data presented above. EOF analysis, closely related to

Principal Components Analysis and dimensionality reduction in the signal processing literature, is based on the singular value decomposition of a matrix. For any real m-by-n matrix A of rank r,

푟 푇 푇 퐴 = ⁡ ∑푘=1 휎푘푢푘푣푘 = 푈훴푉 (4.1) where the columns of U and V are eigenvectors, and Σ is a diagonal matrix of eigenvalue-square roots:

퐻 2 퐴 퐴푣푘 = 휎푘 푣푘 (4.2)

퐻 2 퐴퐴 푢푘 = 휎푘 푢푘 (4.3)

Note that since AHA and AAH are each Hermitian matrices, their eigenvectors are orthogonal.

Thus SVD is a series expansion on an orthogonal basis, and the singular values σk are ordered by convention from largest to smallest, so the first term in the series represents the largest variation in a data set A. In the present study, A is a unit-normalized, zero-meaned matrix of daily averages of total ion density measured by DMSP spacecraft as a function of geographic latitude and time. uk were the principal components, and vk were the EOFs.

Using this analysis, we found that over 95% of the variance in the data could be accounted for in the first two principal components (there are 121 for the 121 4-degree bins from -60° to 60°

GLAT). The first principal component (PC1) for the dawn/dusk satellites in the dawn sector is shown in Figure 4.9 along with the E10.7. It is very clear that the first principal component

(PC1) was associated with the solar EUV – even the 27-day rotation was captured. Indeed there was a .91 CCC for the PC1 with the E10.7.

Figure 4.9. The first principal component (PC1 - red) for the dawn/dusk satellites in the dawn sector and (black) the E10.7.

Figure 4.10 shows the time shifted CCCs between the PC1 and E10.7 for both the dusk sector and the dawn sectors. They were very similar to those shown in Figure 4.8, albeit with greater

CCCs showing the maximum CCC occurred at a time shift of one day and the peaks at 27-day separations.

Figure 4.10. Time shifted CCCs between PC1 and E10.7 for both the dusk (black) and the dawn (red) sectors.

Figure 4.11 shows the second principal component (PC2) along with the density at 30° GLAT. It is clear that PC2 was associated with the SZA effects. This is even more clear when considering the EOFs (or eigenvectors of the covariance matrix) for the first two principal components as shown in Figure 4.12. The EOFs are plotted vs. GLAT in this figure. The second EOF was near zero at the equator and maximum at high latitudes (where the SZA annual variation was the greatest), as expected. The first EOF showed little variation in latitude indicating that the solar

EUV effects were relatively independent of latitude.

Figure 4.11. The second principal component (PC2 - red) and the density (black) at 30° GLAT.

Figure 4.12. The EOFs (or eigenvectors of the covariance matrix) for the first two principal components.

4.3 Discussion

As mentioned in the introduction, studies of the 27-day variation in the topside ionospheric density have shown remarkable correlation with the solar EUV variation while this correlation is much poorer for other ionospheric measurements such as TEC. GPS TEC measurements include the density in the magnetosphere and plasmasphere, which does not necessarily mirror the density in the lower ionosphere. In fact, Rich et al. [2003] showed that the TEC measurements do not show a clear 27-day variation and suggested that, as the TEC is strongly influenced by the ionosphere around the F2 peak, the signatures in the plasma parameters are influenced by plasma dynamics that do not affect the topside ionosphere.

Figure 4.9 showing PC1 and Figures 4.1a, 4.2a, and 4.3a showing the density near the equator illustrate that density fluctuations due to solar EUV and SZA fluctuations are much stronger during high solar activity, in contrast to the saturation effect of ion densities at the F peak

(NmF2) and of total electron content (TEC) [Liu et al. 2007]. This is somewhat difficult to quantify in the density measurements as the solar EUV variability is also smaller during solar minimum. Nevertheless, the CCCs between PC1 and E10.7: 0.76 from 2012 to 2015 and 0.54 from 2006 to 2009. This relationship is clearly demonstrated by the plots of PC2 shown in

Figure 4.11.

The variation in PC2, associated with the density fluctuations produced by fluctuations in the

SZA, was much larger during solar maximum than during solar minimum, despite similar fluctuations in SZA. This was interpreted as the result of the change in the composition

associated with the change in the H+/O+ transition height in turn caused by heating/expansion and cooling/contraction of the thermosphere. The scale height in an H+ dominated topside ionosphere is a factor of 16 larger than the scale height in an O+ dominated topside ionosphere

(for the same temperature). Thus during solar maximum, when the topside ionosphere at DMSP altitudes is O+ dominated, DMSP may be several more scale heights above the F peak than during solar minimum.

This relationship can be illustrated using conjunctions of the C/NOFS spacecraft at low altitude and the DMSP spacecraft. C/NOFS was launched in April 2008 into an elliptical orbit with a perigee near 450 km, apogee near 850 km and a 13o inclination. The Coupled Ion Neutral

CINDI onboard C/NOFS carries an RPA similar to those carried on board DMSP. Magnetic conjunctions between DMSP and C/NOFS spacecraft were calculated using NASA's Satellite

Situation Center online. Two of these conjunctions, chosen when C/NOFS was at nearly the same altitude near 500 km (within 11 km), are shown in Figure 4.13; both occurred in the southern hemisphere at nearly the same latitude. The conjunction in Figure 4.13a occurred on

July 2, 2010 near solar minimum and northern summer equinox. F17 was well above the transition height, such that the ionosphere was almost all H+, while C/NOFS was near the transition height with almost equal concentrations of H+ and O+. The ion temperature measured by F17 was about 2000 K hotter. The conjunction in Figure 4.13b occurred during higher solar activity and southern summer on December 23, 2011. In this case, the transition height was raised to an altitude above F17, with very little H+ measured by C/NOFS, and the ion temperature was about 1000 K higher at F17. The scale height at DMSP (assuming a single

species - O+ or H+) in the first case was ~2500 km, while in the second case it was ~106 km.

Looking at Figure 4.11, the variation in PC2 was much smaller in 2010 when DMSP was above the transition height than in 2011 when DMSP was below the transition height.

Figure 4.13. Two conjunctions of DMSP (asterisks) and C/NOFS (lines). H+ (red) and O+ (black) densities are shown in the top panels and temperatures are shown in the bottom panels. The vertical lines indicate the times of the conjunctions.

4.4 Conclusions

We presented an analysis of 23 years of thermal plasma measurements in the topside ionosphere from the DMSP spacecraft. The solar cycle variations of the daily averaged densities, temperatures, and H+/O+ ratios showed a strong relationship to the solar EUV as described by the

E10.7 solar EUV proxy. The densities near the equator showed a very strong correlation with the

E10.7 producing CCCs greater than 0.85. The H+/O+ varied dramatically from solar maximum to

solar minimum; during solar minimum the topside ionosphere at DMSP was H+ dominated while it was O+ dominated during solar maximum. These ionospheric parameters also varied strongly with season, particularly at latitudes well away from the equator where the SZA varies greatly with season.

The time-shifted CCCs between E10.7 and density showed strong 27-day periodicities, associated with the periodicities in the solar EUV driven by the solar rotation. (These CCCs are highest when the time shift is a multiple of 27 days, in other words, when E10.7 and density are varying in-phase with each other.) The directly measured solar EUV fluxes from the TIMED

SEE and SDO EVE instruments often clearly showed this 27-day periodicity which was also reflected in the densities over shorter time periods. Examples of this correlation over periods on the order of six months showed CCCs between the solar EUV at 28 and 25.5 nm between 0.86 and greater than 0.9; the CCCs with the E10.7 were slightly lower.

The seasonal variations of the SZA make it more difficult to quantify the relationship between the solar EUV and the ionospheric parameters at latitudes away from the equator. However, EOF analysis allows separation of the effects of the solar EUV and SZA variations. The first two principal components of the analysis captured over 95% of the density variation. PC1 was clearly associated with the solar EUV showing a 0.91 CCC with the E10.7 proxy while the PC1 EOFs remained relatively constant with latitude indicating that the solar EUV effects are relatively independent of latitude. PC2 was clearly associated with the SZA variation, showing strong

correlation with the SZA and concomitant density variation at latitudes away from the equator with the EOFs having magnitudes near zero at the equator and maximum at high latitude.

The magnitude of the variation of the response of the topside ionosphere to solar EUV variability was shown to be closely related to the composition. The CCCs between the density and E10.7 and PC1 were greater during solar maximum, when the topside ionosphere at DMSP was O+ dominated than during solar minimum when it was H+ dominated. This was clearly visible in the response of the ionosphere to the SZA changes, which was much stronger during solar maximum than during solar minimum. We interpreted this as the result of the effect of composition on the scale height in the topside ionosphere. When the topside ionosphere is H+ dominated, DMSP may be much less than a scale height above the F2 peak while when it is O+ dominated, DMSP may be several scale heights above the F2 peak.

CHAPTER 5

WN4 VARIABILITY IN DMSP ION DENSITIES ACROSS

SEASON, SOLAR CYCLE, AND LOCAL TIME

In this work we examine how the WN4 pattern in the topside ionosphere varies across season, solar cycle, and local time, using monthly averages of ion density measurements by DMSP satellites over a complete solar cycle. This work is reprinted with permission from Wiley. It was published in the Journal of Geophysical Research: Space Physics article "WN4 variability in

DMSP ion densities across season, solar cycle, and local time" by J.M. Hawkins and P.C.

Anderson (©2017. American Geophysical Union. All Rights Reserved.) As mentioned before, the DMSP satellites orbit at ~840 km above the Earth, well above the F peak in sun-synchronous orbits, i.e. at a nearly constant local time (LT). Total ion density was obtained from the SM, and fractional O+ from the RPA. The present work extends previous studies of WN4 in the topside ionosphere to include multiple levels of solar activity, as well as seasonal effects in Ni at local times not previously examined.

5.1 Satellites and Instrumentation

This study uses measurements from the DMSP F13, F15, F16, F17, and F18 satellites, each at a different and roughly constant local time. Given the 28 geographic longitude bins used in this study, it takes about 2 days for at least one orbit of each satellite to pass through every bin.

Although the DMSP orbits limit what can be learned about tides on timescales shorter than a day, the constant altitude and local time simplifies data processing tremendously, and the combination of multiple DMSP satellites allows for solar cycle-length data sets and a variety of

local times.

Orbital characteristics for the relevant satellites are given in Table 5.1. Note that the following two sections rely heavily on F17 because WN4 patterns were strongly present at this local time, and because F17's local time stayed remarkably constant from 2008 through 2014, allowing us to examine the WN4 characteristics across a range of solar activities at constant LT. Ion densities were obtained from the SM, and fractional O+, H+, and He+ from the RPA,

Table 5.1. Periods of on-orbit performance for the DMSP satellites and solar local time (SLT) of ascending node at launch/currently (or loss of function).

Satellite Launch Date Loss of Function Ascending Node SLT

F13 Apr 1995 Nov 2009 1711/1810

F15 Dec 1999 2110/1630

F16 Oct 2003 2000/1820

F17 Nov 2006 1745/1755

F18 Oct 2009 2000/1950

5.2 Analysis Methods

We used monthly averages of total ion densities (Ni) to examine the variability of the WN4 feature at DMSP altitudes. Ion densities were separated into 5-degree magnetic latitude bins

(except for the data shown in Figure 5.1, which uses 5-degree geographic latitude bins) and 28 equally-spaced geographic longitude bins, and then averaged over one month. The magnetic

latitude (MLAT) used in this paper is the Corrected Geomagnetic (CGM) latitude of a point at 0 km in altitude directly below the satellite position (sub-satellite point.) CGM coordinates were calculated at AFRL based on the algorithm by Gustafsson [1970]. The magnetic equator is set as the dip equator, and low latitudes not included in the original Gustafsson model were interpolated.

A line is drawn to approximate the center of the WN4 feature in each case according to a weighted form of the equation for the centroid:

+30 3 ∑−30(푁푖푗 푥푗) 퐶푒푛푡푒푟⁡푀퐿퐴푇푖 = +30 3 (5.1) ∑−30(푁푖푗 )

where i is an index over GLON, j is a summation index over MLAT from -30 to +30, and Nij is the total ion density at (GLON, MLAT) = (i, j). Equation 5.1 yields the latitudinal position

(center MLAT) of the “bulk” of the WN4 feature in the ith GLON slice. This is important because the tidal features do not perfectly trace the magnetic equator, and we would like to find the latitudes with the most prominent tidal features prior to extracting WN4.

In order to obtain an amplitude and phase of the WN4 feature, the Ni along this center line was fit to a sine wave of fixed wavelength (90o GLON) and variable amplitude and phase (ϕ). Since a quarter wavelength is 22.5o GLON, for a given phase ϕ, the WN4 peaks are located at 22.5 – ϕ,

(90+22.5) – ϕ, (180+22.5) – ϕ, (270+22.5) – ϕ, and the WN4 troughs are at (90-22.5) – ϕ, (180-

22.5) – ϕ, (270-22.5) – ϕ, and (360-22.5) – ϕ in longitude. So for example, a WN4 fit with phase

ϕ = 15o GLON is farther west than a fit with phase ϕ = 10o GLON.

5.3 Seasonal Variations

Figure 5.1 shows the total ion density and fractional O+ measured by the DMSP satellite F17 in

2009. In this year, the F17 ascending node had a local time of ~17:30. Each panel is a global picture of the ionosphere at ~840 km using in situ measurements from a single DMSP satellite.

The rainbow color scale shows the level of total ion density (Ni.) The green-purple panels show the ion composition (fractional O+); green indicates a region where O+ is the dominant ion, and purple indicates a region of primarily light ions (H+ and He+).

Figure 5.1. Monthly averages of total ion density (rainbow panels) and fractional O+ (green and purple panels) measured by DMSP satellite F17 in 2009 at 17:30 LT.

Tidal patterns are present throughout the year, but WN4 is clearest at equinox. This is consistent with the seasonal pattern in the tropospheric driver [Zhang et al., 2012.] They studied altitude-

averaged total tropospheric heating of the DE3 tide using data from International Satellite Cloud

Climatology Project (ISCCP) and Tropical Rainfall Measuring Mission (TRMM), and found that maximum heating occurred in April-May and August-September, with the September maximum being dominant.

The centroid of the density peaks follows the magnetic equator. This is as expected, since WN4 in the topside ionosphere is thought to be a result of electrodynamic lifting, and the vertical ExB drift is greatest along the magnetic equator. The composition (fractional O+) follows the seasons and the solar zenith angle. There is more O+ in the topside ionosphere during local summer because solar heating expands the atmosphere and raises the O+ to H+ transition height [i.e.

Hysell et al., 2009.] This transition height can vary significantly depending on thermospheric heating and winds, as well as electric fields. It can be above or below the DMSP orbit altitude depending on these conditions (for example, see section 4.3 of this dissertation.)

Two additional seasonal effects are apparent in the ion densities. Figure 5.2 shows March, June,

September, and December Ni in 2008, also from the F17 satellite at a local time of 17:00-18:00, now binned by magnetic latitude in 5-degree bins from -30 to 30 MLAT. It can be seen that the

MLAT of the WN4 peaks varies significantly with season. This is more clearly illustrated in

Figure 5.3, which shows the movement of the Ni center line by season in 2008 at 17:00-18:00

LT. The WN4 feature moves 5-10 degrees northward in northern summer, and southward in northern winter. Additionally, ion densities are enhanced in the longitude range from 180 to 270 during northern summer, and reduced during northern winter. This second effect is likely a

Figure 5.2. Monthly averages of total ion density for March, June, September, and December in 2008, measured by DMSP satellite F17 at 17:30 LT. A horizontal line through each panel approximates the latitude of peak densities (see Equation 5.1.)

Figure 5.3. Magnetic latitudes of the WN4 feature for March, June, September, and December in 2008, using data from F17 at 17:30 LT, showing how the peak ion densities move with season. These are the same lines overlaid on each panel in Figure 5.2.

Figure 5.4. Magnetic declination as a function of geographic longitude (GLON.)

transport effect independent of the tidal driver, and has been noted previously by Kil et al. [2008] and Ren et al. [2008]. This longitude range corresponds to a region where the magnetic declination is positive (Figure 5.4.) During northern summer in the evening, the field-aligned components of an eastward day-to-night wind and a southward summer-to-winter wind increase

Ni in this region. During northern winter in the evening, winds blow eastward and northward, driving Ni away from this region and leading to a minimum.

This seasonal asymmetry is even clearer when we subtract out deviations from a 90-degree running median, in order to bring out longitude features independent of WN4. Figure 5.5 shows a 90-degree running median across GLON of the ion densities shown in Figure 5.2. This has the effect of smoothing the data zonally, to de-emphasize the peaks and troughs of the WN4 feature.

Again a comparison with Figure 5.4 shows a clear seasonal change in Ni in the longitude range with positive magnetic declination. The deviations from the running median, shown in Figure

5.6, again indicate a strong WN4 pattern in both March and September.

The seasonal changes in Ni across longitude regions with a strong declination angle make it difficult to reliably estimate the amplitude and phase of WN4 during solstice months. However, we attempted to extract WN4 using only 2 peaks over a range of longitudes with moderate declination angle, -25.7 to 154.3 GLON, for the years 2008-2014, using Ni data from DMSP satellite F17 at 17:30-18:00 LT. The R2 values of these fits are shown in Figure 5.7. R2 values for equinox months, especially September, were generally high, while R2 values during solstice months were either low (<0.5) or inconsistent (varying from year to year.) This suggests that

WN4 may not be the dominant tide during solstice months at this local time.

Figure 5.5. Total ion densities smoothed zonally using a 90-degree running median, for March, June, September, and December in 2008. The data is from DMSP satellite F17 at 17:30 LT. These monthly averages before smoothing are shown in Figure 5.2, and deviations from the running median are shown in Figure 5.6.

Figure 5.6. Deviations from the 90-degree running median (Figure 5.5) of total ion densities (Figure 5.2) for March, June, September, and December 2008. The data is from DMSP satellite F17 at 17:30 LT.

Figure 5.7. Goodness of fit (R2) vs. month of the WN4 fitting procedure over a range of longitudes with moderate declination angle (-25.7 to 154.3 GLON). The data is from DMSP satellite F17 at 17:30-18:00 LT over the years 2008-2014. The trend is a polynomial fit to the R2 values.

The next two sections use data for the month of September, since WN4 is clearest during equinox months.

5.4 Solar Cycle Variations

Previous studies found solar cycle effects in DE3 temperatures and winds in the E region above

120 km, but not at mesosphere/lower thermosphere heights [Oberheide et al., 2009]. Here we investigate the solar cycle dependence of WN4 in ion densities in the topside ionosphere.

Figure 5.8 shows the total ion density (Ni) averaged over the month of September for seven selected years over more than a solar cycle, from 2001 to 2014. The F10.7 solar flux index ranged from 67.8 SFU in the low solar minimum of 2008, to 237.2 SFU in the solar maximum of

2001. These were all acquired from data in the period from the F13 and F17 satellites near

Figure 5.8. Total ion densities measured by DMSP satellite F17 at 17:30-18:00 LT (top five panels) and F13 at ~18:15 LT (bottom two panels), averaged over the month of September, for seven years. The average value of the F10.7 solar index for each plot is given in parentheses. A horizontal line through each panel approximates the latitude of peak densities (see Equation 5.1.)

17:30-18:00 LT. Average F10.7 values for each month are labeled in parentheses. The figures were ordered by F10.7 rather than year and only selected years are shown as the F10.7 was similar in September for some of the years.

Figure 5.9 shows dN/N along the center line for the same months/years as Figure 5.8. There is some variation, but no trend in dN/N with solar cycle/F10.7. For instance, notice that the greatest negative excursions near 140° GLON occur in 2012 and 2013, when F10.7 was moderate. The greatest negative excursions near 225° and 315° GLON occur during 2014, again during moderate F10.7. The positive excursions are, in general, similar for all years.

Figure 5.9. Fractional ion densities (dN/N) along the center of the WN4 feature (see Equation 5.1), for September during the seven years shown in Figure 5.8. The F10.7 solar index values are given in parentheses.

We used the fitting procedure described in the Analysis Methods section to estimate the amplitude and phase of WN4 across solar cycle for these same seven years. The amplitude of the Ni fluctuations versus the F10.7 solar index is shown in Figure 5.10 and listed in Table 5.2.

Amplitude prior to normalization shows a clear increase with solar cycle since the background Ni is increasing. However, when the amplitude is normalized by the average Ni of the fit (Figure

5.11), the amplitude stays relatively constant with solar cycle, ranging from 11.5% to 21.5% of the background Ni. The phase (Figure 5.12) ranged from 9.5 to 18.1 degrees GLON over these years and showed an eastward trend with increasing F10.7.

Figure 5.10. WN4 amplitude of September Ni during the seven years shown in Figure 5.8, vs. the monthly average of the F10.7 index. Amplitude has not been normalized for this figure.

Figure 5.11. WN4 amplitude of September Ni during the seven years shown in Figure 5.8, vs. the monthly average of the F10.7 index. Here, amplitude has been normalized by the mean Ni of the fit to account for changes in the background Ni due to solar cycle.

Figure 5.12. WN4 phase of September Ni during the seven years shown in Figure 5.8, vs. the monthly average of the F10.7 index.

Lines through the center of the WN4 structure for September of these same years are shown in

Figure 5.13. While the latitude of the WN4 peaks varies slightly (<5 degrees), its variation with season is much stronger than its variability across a solar cycle (compare Figures 5.13 and 5.3.)

Figure 5.13. Magnetic latitudes of the WN4 feature for September of the same seven years in Figure 5.8, showing the variability in location of the peak densities over a solar cycle. The blue line shows the center line of the month with the lowest F10.7 value in this set (2008, F10.7=67.8), and the red line shows the center line of the month with the highest F10.7 value in this set (2001, F10.7=237.2)

5.5 Local Time Variations

The sun-synchronous nature of the DMSP satellites allows us to examine WN4 patterns at several local times in the morning and afternoon/evening. The development of WN4 in the morning is shown in Figure 5.14. Ion densities (Ni) measured by 3 DMSP satellites (F13 at

~6:00 LT, F18 at ~8:00 LT, and F15 at ~9:30 LT) were averaged separately for the month of

September. The F13 and F15 data are shown for 2001 and 2003, and the F18 data are shown for

2014 and 2015. The observed features are remarkably consistent from year to year. Note that the plots are to different scales. At 6:00 LT, tidal patterns are obscured by other rapid changes in the ionosphere at dawn, but by 9:30 LT, WN4 is a recognizable feature in Ni.

Figure 5.15 shows afternoon and evening Ni in September 2014 measured by 3 different satellites (F15 at ~15:48 LT, F17 at ~18:06 LT, and F18 ~19:48 LT.) The familiar WN4 pattern is strongly present at all of these times, showing roughly a factor of 2 increase of the peak densities over the background shape. There is no significant change in the latitude position of

WN4 over these times.

The WN4 pattern in Ni at 800 km is remarkably consistent throughout the afternoon and evening, but there are significant differences between how WN4 manifests in the afternoon/evening vs. in the morning. Unfortunately, due to the DMSP orbits, no data is available near local noon. At

9:30 LT, the peak near 180 GLON is further north than its position in the afternoon/evening, the peaks near 90 and 270 GLON are similar to their afternoon/evening latitudes, and the peak near

0 GLON is further south and significantly weaker than the other peaks. Additionally, at 9:30 LT, the peak near 180 GLON is strongest; at ~15:48 LT, the peak near 90 GLON is strongest; and by

~19:48 LT, the peak near 0 GLON is strongest.

Figure 5.14. Total ion densities averaged over the month of September in six different years, measured by DMSP satellites (a) F13 at ~6:00 LT, (b) F18 at ~8:00 LT, and (c) F15 at ~9:30 LT.

Figure 5.15. Total ion densities averaged over the month of September in 2014, measured by DMSP satellites F15 at ~15:48 LT, F17 at ~18:06 LT, and F18 at ~19:48 LT.

Data from only two, three, or sometimes four DMSP satellites are available in the afternoon/evening in any given year, so to examine the WN4 behavior across MLT, we extracted

WN4 for the month of September across multiple DMSP satellites over the years 2008-2014.

The amplitude and phase values for these fits are given in Table 5.2. Figure 5.16 shows phase as a function of MLT, color-coded by year, with the overall trend. Over these years and local times, the phase varied from 9.9 to 23.5 degrees GLON, and showed an overall eastward trend of

2 degrees per hour MLT. The WN4 amplitude (normalized by the mean Ni of the fit) varied from 14.0 to 24.6% of Ni, and did not show a trend with MLT.

Figure 5.16. Estimated phase of WN4 for total ion densities measured by several DMSP satellites (see Table 5.2) during the month of September, showing the eastward trend of phase with evening magnetic local time.

5.6 Conclusions

Non-migrating tides are in fact often the dominant source of longitudinal variability in the low- latitude topside ionosphere. We used monthly averages of ion densities (Ni) measured by DMSP satellites to investigate the seasonal, solar cycle, and local time variations of the WN4 pattern.

The main results are as follows:

1. Tidal patterns are present near the magnetic equator throughout the year, and WN4 is most clearly present at September equinox.

2. In May-August, Ni near 180°-270° GLON is increased; in November-February, Ni in this region is decreased. This is interpreted as the result of the positive geomagnetic field declination in this region. During northern summer in the evening, the field-aligned components of an eastward day-to-night wind and a southward summer-to-winter wind increase Ni in this region.

During northern winter in the evening, winds blow eastward and northward, driving Ni away from this region and leading to a minimum.

3. In May-August, WN4 moves northward 5-10° in MLAT; in November-February, WN4 moves southward 5-10° in MLAT.

4. No solar cycle effects were found in the magnitude of dN/N. However, the longitude position of the peaks (phase) was observed to vary by about 10° with F10.7, moving eastward with increasing F10.7. The latitude variation was less than 5° and did not show a trend with F10.7.

5. The WN4 pattern changes rapidly near dawn, but appears fully developed by 9:30 LT. WN4 is very constant throughout the afternoon and evening in terms of dN/N and latitude position of the peaks. The strong afternoon/evening presence of WN4 is consistent with the upward propagation mechanism of the DE3 tide into the F-region and topside ionosphere proposed by

Immel et al. [2006]. Dynamo electric fields in the E-layer associated with the ExB drift of plasma lifted up into the F-region are present in the morning and daytime, leading to a strong

WN4 signature in the topside Ni in the afternoon and evening, but this mechanism weakens or disappears at night, allowing the WN4 signature in the topside Ni to dissipate by 6:00 LT. WN4 drifts eastward in the afternoon/evening at about 2 degrees GLON per hour MLT, and there is variation in this rate from year to year.

6. Compared to the afternoon/evening WN4 pattern in September, at 9:30 LT, the 180° GLON peak is further north and strengthened, the 90° and 270° GLON peaks are similar in MLAT, and the 0° GLON peak is further south and much weaker.

Table 5.2. Magnetic local time (MLT), amplitude prior to normalization (cm-3), phase, 2 amplitude normalized by mean Ni, and goodness of fit (R ) of WN4 fits used in this paper. All fits listed are for the month of September.

Amplitude Normalized Year Satellite MLT Phase (o GLON) R2 (cm-3) Amplitude 2001 F13 18.24 55408.0 9.537 0.115 0.664 2002 F13 18.36 56631.7 10.054 0.149 0.700 2008 F16 19.81 4409.0 15.328 0.212 0.830 2008 F17 17.54 4558.7 16.128 0.166 0.839 2009 F16 19.41 4353.4 20.171 0.182 0.656 2009 F17 17.54 4608.9 19.915 0.148 0.778 2010 F16 18.92 8126.4 15.890 0.211 0.681 2010 F17 17.59 6886.8 18.108 0.169 0.688 2010 F18 20.15 4940.0 13.259 0.246 0.700 2011 F15 16.43 20971.6 19.302 0.188 0.776 2011 F16 18.36 22800.8 17.760 0.215 0.827 2011 F17 17.67 21875.4 16.691 0.229 0.808 2011 F18 20.26 13373.1 12.249 0.239 0.759 2012 F15 15.80 29465.0 21.283 0.192 0.595 2012 F16 17.79 20507.1 15.446 0.152 0.561 2012 F17 17.80 21271.9 15.016 0.166 0.593 2012 F18 20.24 11079.3 10.282 0.141 0.640 2013 F16 17.24 20042.9 16.290 0.212 0.726 2013 F17 17.95 17281.4 13.503 0.195 0.760 2013 F18 20.09 12000.1 9.927 0.218 0.773 2014 F15 15.80 31990.5 23.501 0.165 0.539 2014 F16 16.76 33121.1 20.104 0.208 0.613 2014 F17 18.11 33789.4 15.463 0.215 0.696 2014 F18 19.82 18961.1 11.785 0.162 0.570

CHAPTER 6

TOPSIDE PROFILE SHAPE VARIATIONS USING

DMSP / C/NOFS CONJUNCTIONS AND IRI

In this work, we explore the topside ionospheric altitudinal structure and the ability of the IRI

2016 empirical model to accurately model the ion density profiles and the dependence of the model error in the topside scale height on various conditions and parameters. We use a database of over 11,000 magnetic conjunctions between the C/NOFS and DMSP satellites over the years

2009-2015, together with the most recent version of the IRI empirical model. IRI is currently the

International Standardization Organization (ISO) standard for the ionosphere and as such, it is very important to know the accuracy of the specification of the topside ionospheric structure. As mentioned in section 2.2, the topside ionosphere is a large contributer to the total electron content (TEC) [Jakowski et al., 2002], which affects GPS measurements, and Reinisch and

Bilitza [2017] concluded that more work is needed to better understand and model the topside electron density profiles. The topside ionsopheric structure and composition in IRI was originally based on a relatively small sampling from in situ measurements on the IK-24, AE-C. and AE-E spacecraft. Recently, the addition of data from the C/NOFS spacecraft has improved the accuracy of IRI, especially during solar minimum conditions [Bilitza, et al., 2017]. While numerous validation studies on IRI have been performed using various ionospheric datasets, [e.g.

Bilitza 2009; Nigussie et al., 2013; Lühr et al., 2010; Araujo-Pradere et al., 2013; Wang et al.,

2016], none of them were able to test the accuracy of the altitudinal profiles of the density in the topside ionosphere for a large range of conditions and locations to any degree of thoroughness.

Several methods have been used to reconstruct topside electron and ion density profiles and to

extract important parameters such as the scale height and O+ to H+ transition height, each with their own problems. Ground-based measurements of necessity are limited in coverage and are dependent on bottomside dynamics while in situ measurements from single satellites do not provide an altitudinal distribution other than in a statistical sense, or in the case of satellites with elliptical orbits, distributed in location. Having two near-simultaneous measurements at two different points along a magnetic field line allows specification of the altitudinal profile or the scale height in the topside ionosphere and thus permits us to test the dependence of the accuracy of the IRI model in specifying the topside scale height and the dependence of the model error on various conditions and parameters.

6.1 Satellite Conjunctions Data Set

To identify magnetic conjunctions, we first used the Locator Tabular tool at NASA's Satellite

Situation Center (SSC) website to generate files of the latitude and longitude of the footprints of the magnetic field line mapped from the location of the C/NOFS satellite and the DMSP satellites to 100 km altitude in both hemispheres. The geomagnetic model used was the epoch- appropriate International Geomagnetic Reference Field (IGRF). One-minute values were produced by the SSC website for all satellites from 2009 to C/NOFS reentry in November 2015.

These values were then interpolated to a 1-second time resolution. Conjunctions were identified where the field line tracings from C/NOFS and any DMSP spacecraft were within 2º GLAT and

1.5º GLON within 5 minutes of each other. A further requirement was that the satellites themselves were within 4º GLAT and 4º GLON, and in the same hemisphere.

The density data from each satellite was averaged over a 30 second window centered on the times when the satellite footprints were closest together, and at least 15 good quality data points were required from each satellite within its window for the conjunction to be retained in the data set. We also required that the variation of the density over the 30 s window be less than 50%.

This yielded a total of 11,430 conjunctions from the year 2009 to the end of 2015. As shown in

Figures 6.1a and 6.1b, the vast majority of conjunctions had field line footprints with differences in latitude and longitude less than a few 10ths of a degree: 10931 had longitude differences less than 0.3° and 7424 had latitude differences less than 0.5°.

The local time, altitude, and latitude/longitude coverage of the data set is shown in Figure 6.2.

(The gap in coverage during 2013 was due to a government sequester which led to the C/NOFS satellite being put in standby mode.) The local times of the conjunctions (Figure 6.2a) ranged from 0430 - 0900 and 1630 - 2100 MLT in 2009 and were determined by the local times of the

DMSP orbits. The changes in MLT over the time period occurs because the DMSP orbits precess in ascending node local time to earlier times due largely to atmospheric drag. Thus, by the year

2015 some of the conjunctions (with the F15 satellite) occurred as early as 0200 and 1400 MLT.

The altitude (Figure 6.2b) of the DMSP satellites remained very constant over the time period

(between 830 and 870 km), but C/NOFS decreased in altitude significantly, particularly in 2014 and 2015.

Figure 6.1. Histogram of the number of conjunctions with a given footprint (a) latitude difference and (b) longitude difference between satellites.

The latitude, longitude, and difference in altitude between the DMSP and C/NOFS satellites are highly coupled as shown in Figure 6.2c. Due to the conjunction selection criteria and the geomagnetic dip angle, there are relatively few conjunctions close to the magnetic equator, and these occur at small altitude separations. This means that, for example, in the range

Figure 6.2. (a) The magnetic local time (MLT) of all conjunctions over time. (b) Altitudes of the DMSP and C/NOFS satellites over time. (c) Geographic latitude and longitude of C/NOFS at all conjunctions. The conjunctions with a satellite altitude difference greater than 400 km are highlighted in red.

northern hemisphere, because the magnetic equator is farther north than the geographic equator.

Additionally, the conjunctions with a large difference in altitude between the satellites (shown in red) are all farther from the magnetic equator, and therefore are only present at certain geographic latitudes and longitudes. For example, the only conjunctions with a satellite altitude difference greater than 400 km in the northern hemisphere are restricted to -150o

6.2 Analysis Methods

In our analysis, we used the total density measurements from the RPA on C/NOFS and the SM on DMSP. The RPA on DMSP has a lower temporal resolution than that on C/NOFS (2 or 4 seconds vs 0.5 or 1 second) so we use the SM as it has a much higher sampling rate (12 Hz) which is averaged to one-second samples. To ensure that the C/NOFS and DMSP instruments were similarly calibrated, we calculated a correction factor based on Ni measurements from the conjunctions where the altitude difference between the satellites was less than 20 km. In this set of 538 conjunctions, the Ni measurements between the satellites were found to be overall in good agreement, with the DMSP SM reading on average 3.6% higher than the C/NOFS RPA with a standard deviation of 11%. This correction factor was applied to the DMSP ion densities prior to further analysis.

For each conjunction in the data set, we ran IRI 2016 at the date, time, and location (latitude, longitude, and altitude) of each of the two satellites. (We used the most recent recommended settings for IRI 2016 including the NeQuick model for the topside Ne profile, URSI maps for the

F2 peak density, the AMTB-2013 model for the F2 peak height, and the RBV10+TTS05 option

for the ion composition.) We then calculated a measure of the difference in slope between the electron density (Ne) profile in the IRI model and the two Ni measurements along a field line provided by DMSP and C/NOFS, which we label as the % error between IRI and the adjusted satellite data:

퐼푅퐼⁡푁푒푎푡⁡퐶/푁푂퐹푆 퐴푑푗푢푠푡푒푑⁡퐷푀푆푃⁡푁푖 = 푁푖푚푒푎푠푢푟푒푑⁡푏푦⁡퐷푀푆푃⁡ ×⁡ (6.1a) 푁푖푚푒푎푠푢푟푒푑⁡푏푦⁡퐶/푁푂퐹푆

퐴푑푗푢푠푡푒푑⁡퐷푀푆푃⁡푁 ⁡−⁡⁡퐼푅퐼⁡푁 ⁡푎푡⁡퐷푀푆푃 %⁡퐸푟푟표푟 = ⁡ 푖 푒 ⁡× 100% (6.1b) 퐴푑푗푢푠푡푒푑⁡퐷푀푆푃⁡푁푖

This method allows us to take advantage of a unique feature of this data set, namely having two near-simultaneous measurements along a magnetic field line, in order to gain insight into the shape of the topside ion density profiles.

If the % error is greater than 0, then the “slope” of the satellite Ni measurements is steeper than the slope of the Ne profile in IRI, and IRI is underestimating the plasma scale height. For an idealized Chapman profile in diffusive equilibrium, we expect this to correlate with higher

+ + temperatures (Ti + Te) and/or lower effective plasma mass (meff), through a greater H /O ratio.

These interpretations are summarized in the table below. Of course, the assumption of diffusive equilibrium is not always valid, for example near sunrise and sunset when heating and ExB drift can produce rapid ion flows.

Table 6.1. A quick reference for interpretation of % error.

% error < 0 means: % error = 0 means % error > 0 means:

IRI model overestimated the The Ni profile slope IRI model underestimated the Ni

Ni profile slope. matched IRI. profile slope.

Plasma scale height was Plasma scale height Plasma scale height was greater smaller than predicted by IRI. matched IRI. than predicted by IRI.

If we can assume diffusive If we can assume diffusive equilibrium, then expect equilibrium, then expect higher lower Ti+Te or higher meff. Ti+Te or lower meff.

6.3 Results

Figure 6.3 shows the % error between IRI and the adjusted data vs. satellite altitude difference.

The first panel includes the complete data set, and the second and third panels are separated into morning and afternoon magnetic local times (MLT.) All the conjunctions are shown in gray. A running median is shown in blue, with the upper and lower quartiles as blue bars. The % error for the complete data set (Figure 6.3a) is 0 for a satellite distance of 0 because we have intercalibrated the satellite measurements to be the same, on average, when they are close together, as mentioned previously. As the satellite altitude difference increases in the morning, the % error increases and is consistently positive. However, in the afternoon, the median % error is within ±10% for all satellite altitude differences. This means that the IRI model tends to underestimate the Ne profile slope and the plasma scale height in the topside ionosphere in the morning, but corresponds very well on average with the satellite data in the afternoon. In all

three panels, the interquartile distance increases as the difference in altitude between the

satellites increases, indicating greater variability in the % error over a greater distance between

measurements.

Figure 6.3. Percent error between IRI and the adjusted conjunction data vs. satellite altitude difference, separated into morning and afternoon local times. Gray scatter shows all the data points; blue trace shows a running median, with upper and lower quartiles as blue bars.

The variation in % error with MLT, now separated by satellite altitude difference, is further illustrated in Figure 6.4. In the first panel, which shows the results for all altitudes, the median

Figure 6.4. Percent error vs. magnetic local time, separated by satellite altitude difference. The panels are in the same format as Figure 6.3.

% error varies from 20-40% in the morning, but is less than 10% in the afternoon and evening.

As the altitude difference increases, the % error naturally increases, but, again, IRI underestimates the slope and plasma scale height in the morning, but represents the data well in the afternoon and evening. For all panels shown, the interquartile distance is smaller during the day than pre-dawn or post-sunset, indicating less % error variability during the daytime. The median % error also changes rapidly pre-dawn and post-sunset; the increase in scale height pre- dawn, and the decrease in scale height post-sunset, appear to be stronger than changes in scale height predicted by the model.

Figure 6.5 shows % error vs. the effective mass (amu) measured by DMSP. The columns separate the data into morning and afternoon local times, and the rows separate the data into low, medium, and high levels of the F10.7 index. An effective mass of 1 indicates a completely H+ ionosphere at DMSP, while an effective mass of 16 indicates entirely O+. Note that at high levels of F10.7 in the afternoon (bottom right panel) the ionosphere at DMSP is almost always

O+ dominated, whereas at solar minimum in the morning (second row, second column), the ionosphere is more often H+ dominated.

The top left panel shows that on average across the whole data set, the % error decreases as the effective mass increases. This is as expected, since a decreasing % error indicates a decreasing plasma scale height, and the plasma scale height is inversely proportional to effective mass.

However, the different panels show that there are subtle differences in this effect in the morning

Figure 6.5. Percent error vs. effective mass measured by DMSP, in the same format as Figures 6.3 and 6.4. The rows show different levels of solar cycle, represented by the F10.7 solar index. The columns separate the data into morning and afternoon local times.

vs. the afternoon and at different parts of the solar cycle. Comparing the panels in the left column, this effect is more pronounced at medium and high levels of F10.7 (>100 SFU) than at low levels of F10.7 (<100 SFU.)

Figure 6.6 shows the % error vs. ion temperature (Ti) measured by DMSP, in the same format as

Figure 6.5, separated into low, medium, and high levels of F10.7, and into morning and afternoon local times. Similarly, Figure 6.7 shows % error vs. electron temperature measured by

DMSP. In both figures, in the afternoon (right column), the % error increases as temperature increases, particularly for higher temperatures and lower F10.7 values (F10.7<150 SFU.)

However, in the morning (center column) there is not a linear trend of % error with Ti. or Te.

Ti and Te at DMSP (Figures 6.8a and 6.8b) can each rise by 2000 K between 5:00 and 7:00 MLT.

During this time, the vertical ion flows at DMSP (Figure 6.8c) jump from downward to upward, and the vertical ion flows at C/NOFS (Figure 6.8d) may be either upward or downward. In contrast, the median and upper and lower quartile Ti and Te at DMSP at 14:00-18:00 MLT remain constant, and the vertical ion flows at both DMSP and C/NOFS stay close to 0 with little variation (small interquartile distance.) The topside ionosphere at 14:00-18:00 MLT is in diffusive equilibrium, but between 5:00 and 7:00 MLT, it is not. This explains why in Figures

6.6 and 6.7, % error increases linearly with Ti and Te, but in the morning, % error shows a more complex relationship with Ti and Te that may not correspond to a simple Chapman plasma scale height.

Figure 6.6. Percent error vs. ion temperature (Ti) measured by DMSP, in the same format as Figure 6.5.

Figure 6.7. Percent error vs. electron temperature (Te) measured by DMSP, in the same format as Figure 6.5.

Figure 6.8. (a) Ion temperature (Ti), (b) Electron temperature (Te), and (c) upward vertical ion drift (vz)as a function of magnetic local time (MLT), measured by DMSP. (d) Vertical ion drift (vz) vs. MLT measured by C/NOFS.

6.4 Conclusions

We presented a new data set of over 11,000 magnetic conjunctions between the DMSP and

C/NOFS satellites from 2009 to the end of 2015. This is a valuable long-term data set that covers a variety of solar cycle conditions, from the minimum in 2009 to the maximum in 2014, and multiple local times, longitudes, and altitudes. The conjunctions provide two near- simultaneous in situ satellite measurements of topside ionospheric parameters, a unique and important addition to existing methods of reconstructing topside ion and electron density profiles, and in particular to topside modeling in IRI.

We used the conjunction data set together with the most recent version of the IRI empirical model to investigate how the error in the IRI profile shape varies with local time and under different geophysical conditions, and found significant differences in the % error between morning and afternoon. In the morning, % error was consistently positive, indicating that IRI underestimates the Ne profile slope and plasma scale height in the morning, but IRI matched the profile shape well in the afternoon. % error decreased with increasing effective mass, particularly when F10.7 was >100 SFU. % error increased with increasing Ti and Te in the afternoon, but did not show a linear trend with Ti or Te in the morning. This is consistent with a diffusive equilibrium condition in the afternoon (14:00-18:00 MLT), whereas strong ion flows in the morning produce a more complex relationship between the Ni profile shape and temperatures.

CHAPTER 7

CONCLUSION

In its Recommended Roadmap for Science and Technology 2005-2035, NASA called the space science community to “identify impacts of solar variability on Earth’s atmosphere.” The first study in this dissertation found that the solar EUV as described by the E10.7 solar EUV proxy correlated strongly with daily averaged densities, temperatures, and H+/O+ ratios measured by the DMSP spacecraft. During solar minimum, the topside ionosphere at DMSP was H+ dominated while it was O+ dominated during solar maximum. The cross-correlations between

EUV and densities showed strong 27-day periodicities associated with periodicities in the solar

EUV driven by the solar rotation. While the use of solar EUV proxies such as E10.7 are common in studies of the ionospheric response to EUV, proxies do not capture the full variability of spectrally-separated EUV. Therefore we also used the spectrally-separated EUV data measured by the SDO and TIMED satellites. We showed examples of higher correlations of densities with the 28 nm line from TIMED and the 25.5 nm line from SDO than with the proxy.

NASA’s Recommended Roadmap also called the space science community to “Understand the role of the Sun as an energy source to Earth’s atmosphere and, in particular, the role of solar variability in driving change.” The role of solar variability in driving change in the topside ionosphere was found to be closely related to the composition. We used EOF analysis to separate the effects of the solar EUV and SZA variations and found that the first two principle components accounted for over 95% of the variation in density. The first principle component

(PC1) correlated strongly with the solar EUV, while the second principle component (PC2) was

associated with the SZA variation. We made two important observations of these EOF results:

1) cross-correlations between the density and E10.7 and PC1 were greater during solar maximum, when the topside ionosphere was O+ dominated than during solar minimum when it was H+ dominated, and 2) the ionospheric response to SZA changes as represented by PC2 was much stronger during solar maximum than solar minimum. These two observations led to the interpretation that variations in density such as those produced by EUV and SZA variations are amplified when the ionosphere is O+ dominated, rather than H+ dominated, because of the effect of composition on the scale height.

While DE3 propagation and WN4 signatures have been much studied in recent years in the mesosphere-lower thermosphere region, relatively little is known about how WN4 impacts the ionosphere above the F-peak. The second study in this dissertation showed that non-migrating tides are in fact often the dominant source of longitudinal variability in the topside ion densities near the equator. This research addressed science priorities of the National Research Council and the Panel on Atmosphere-Ionosphere-Magnetosphere Interactions (AIMI) and provided important clues to whole-atmosphere coupling. Monthly averages of DMSP ion densities showed tidal patterns present near the magnetic equator throughout the year, and WN4 was most clearly present at September equinox. In May-August, Ni near 180°-270° GLON is increased; in

November-February, Ni in this region is decreased. This is interpreted as the result of the positive geomagnetic field declination in this region. During northern summer in the evening, the field-aligned components of an eastward day-to-night wind and a southward summer-to- winter wind increase Ni in this region, whereas during northern winter in the evening, winds

blow eastward and northward, driving Ni away from this region and leading to a minimum.

Additionally, WN4 moves northward 5-10° in MLAT in May-August; in November-February,

WN4 moves southward 5-10° in MLAT. No solar cycle effects were found in the magnitude of dN/N or in latitude variations. However, the longitude position of the peaks varied by about 10° with F10.7, moving eastward with increasing F10.7. The WN4 pattern changes rapidly near dawn, appears fully developed by 9:30 LT, and is strong and constant throughout the afternoon and evening in terms of dN/N and latitude position of the peaks. The strong afternoon/evening presence of WN4 is consistent with the upward propagation mechanism of the DE3 tide into the

F-region and topside ionosphere proposed by Immel et al. [2006]. Dynamo electric fields in the

E-layer associated with the ExB drift of plasma lifted up into the F-region are present in the morning and daytime, leading to a strong WN4 signature in the topside Ni in the afternoon and evening, but this mechanism weakens or disappears at night, allowing the WN4 signature in the

o topside Ni to dissipate by 6:00 LT. WN4 drifts eastward in the afternoon/evening at about 2

GLON per hour MLT, and there is variation in this rate from year to year. Compared to the afternoon/evening WN4 pattern in September, at 9:30 LT, the 180° GLON peak is further north and strengthened, the 90° and 270° GLON peaks are similar in MLAT, and the 0° GLON peak is further south and much weaker.

The IRI model is the International Standardization Organization (ISO) standard for the ionosphere and an essential resource to space scientists and engineers. As an empirical model, its accuracy relies on incorporating information from all available data sources and validation from the space science community. Accurate models of the topside electron density profiles are

particularly important because of the region’s large contribution to total electron content (TEC)

[Jakowski et al., 2002], which affects GPS measurements. In a special issue of Advances in

Space Research, Reinisch and Bilitza [2017] found that more work is needed to understand and model topside electron density profiles in the equatorial and low latitude region. The third study in this dissertation addressed this need by investigating how the error in the topside profile shape in IRI at low latitudes varies with local time and under difference geophysical conditions. We presented a new data set of over 11,000 magnetic conjunctions between the DMSP and C/NOFS satellites providing two near-simultaneous in situ measurements of topside ionospheric parameters from 2009 through the end of 2015. We found that IRI underestimates the Ne profile slope and plasma scale height in the morning, but matches the profile shape well in the afternoon

(1400-1800 MLT.) % error decreased with increasing effective mass, particularly when F10.7 was >100 SFU. % error increased with increasing Ti and Te in the afternoon, but did not show a linear trend with Ti or Te in the morning. This is consistent with a diffusive equilibrium condition in the afternoon, whereas strong ion flows in the morning produce a more complex relationship between the Ni profile shape and temperatures.

REFERENCES

A, E., D. Zhang, A. J. Ridley, Z. Xiao, and Y. Hao (2012), A global model: Empirical orthogonal function analysis of total electron content 1999–2009 data, J. Geophys. Res., 117, A03328, doi:10.1029/2011JA017238.

Anderson, P. C., and J. M. Hawkins (2016), Topside ionospheric response to solar EUV variability, J. Geophys. Res. Space Physics, 121, 1518–1529, doi:10.1002/2015JA021202.

Anderson, P. C., Characteristics of Spacecraft Charging in Low Earth Orbit, J. Geophys. Res., 117, 10.1029/2011JA016875, 2012.

Araujo-Pradere, E. A., D. Buresova, D. J. Fuller-Rowell, T. J. Fuller-Rowell (2013), Initial results of the evaluation of IRI hmF2 performance for minima 22–23 and 23–24, Adv. Space Res., 51, 630-638, doi:10.1016/j.asr.2012.02.010.

Bailey, G. J., and R. Sellek (1990), A mathematical model of the Earth’s plasmasphere and its application in a study of He+ at L = 3, Ann. Geophys., 8, 171.

Bankov, L., R. Heelis, M. Parrot, J.‐J. Berthelier, P. Marinov, and A. Vassileva (2009), WN4 effect on longitudinal distribution of different ion species in the topside ionosphere at low latitudes by means of DEMETER, DMSP‐F13 and DMSP‐F15 data, Ann. Geophys., 27, 2893–2902, doi:10.5194/angeo-27-2893-2009.

Barth, C. A., K. D. Mankoff, S. M. Bailey, and S. C. Solomon (2003), Global observations of nitric oxide in the thermosphere, J. Geophys. Res., 108(A1), 1027, doi:10.1029/2002JA009458.

Belehaki, A., P. Marinov, I. Kutiev, N. Jakowski, and S. Stankov (2006), Comparison of the topside ionosphere scale height determined by topside sounders model and bottomside digisonde profiles, Adv. Space Res., 37, 963–966.

Bilitza, D (2009), Evaluation of the IRI-2007 model options for the topside electron density, Adv. Space Res., 44, 701-706, doi:10.1016/j.asr.2009.04.036.

Bilitza, D., D. Altadill, V. Truhlik, V. Shubin, I. Galkin, B. Reinisch, and X. Huang (2017), International Reference Ionosphere 2016: From ionospheric climate to real-time weather predictions, Space Weather 15(2), 418-429, doi:10.1002/2016SW001593.

Chamberlin, P. C., T. N. Woods, and F. G. Eparvier, (2008) Flare Irradiance Spectral Model (FISM): Flare component algorithms and results, Space Weather, 6(5), S05001, doi: 10.1029/2007SW000372.

Chapman, S., (1961), Scale Times and Scale Lengths of Variables: with Geomagnetic and Ionospheric Illustrations, Proc. Phys. Soc., 77, 424.

Coley, W. R., and R. A. Heelis (2012), Response of the equatorial topside ionosphere to 27-day variations in solar EUV input during a low solar activity period using C/NOFS, J. Geophys. Res., 117, A03330, doi:10.1029/2011JA017301.

Dudok de Wit, T., Kretzschmar, M., Aboudarham, J., Amblard, P.-O., Auchere, F., and Lilensten, J.: Which solar EUV indices are best for reconstructing the solar EUV irradiance?, Adv. Space Res., in press, 2007.

Emmert, J. T., J. L. Lean, and J. M. Picone (2010), Observations of record low thermospheric densities during the 2008 solar minimum, Geophys. Res. Lett., 37, L12102, doi:10.1029/2010GL043671.

England, S. L. (2012), A Review of the Effects of Non-migrating Atmospheric Tides on the Earth's Low-Latitude Ionosphere, Space Sci Rev, 168:211-236, doi:10.1007/s11214-011-9842-4.

England, S.L., T.J. Immel, J.D. Huba, M.E. Hagan, A. Maute, R. DeMajistre, Modeling of multiple effects of atmospheric tides on the ionosphere: an examination of possible coupling mechanisms responsible for the longitudinal structure of the equatorial ionosphere. J. Geophys. Res., Space Phys. 115, A05308 (2010). doi:10.1029/2009JA014894

Fallen, C. T. and B. J. Watkins (2013) Diurnal and seasonal variation of electron heat flux measured with the Poker Flat Incoherent-Scatter Radar. J. Geophys. Res. 118, 5327-5332 DOI: 10.1002/jgra.50485

Fejer, B. G., J. W. Jensen, and S.-Y. Su (2008), Quiet time equatorial F region vertical plasma drift model derived from ROCSAT-1 observations, J. Geophys. Res., 113, A05304, doi:10.1029/2007JA012801.

Floyd, L., Newmark, J., Cook, J., Herring, L., and McMullin, D., (2005), Solar EUV and UV spectral irradiances and solar indices, J. Atmos. Terr. Phys., 67, 3–15.

Frynberg (Fainberg), E. B., Separation of the geomagnetic field into a normal and anomalous part, Geomagn. Aeron., Engl. Transl., 15, 117, 1975.

Fuller-Rowell, T., Solomon, S., Roble, R., Viereck, R.: 2004, In: Pap, J.M., Fox, P., Frohlich, C., Hudson, H.S., Kuhn, J., McCormack, J., North, G., Sprigg, W., Wu, S.T. (eds.) Solar Variability and Its Effects on Climate, Geophys. Monogr. Ser. 141, AGU, Washington, 341.

Garzon, D. P., C. G. M. Brum, E. Echer, N. Aponte, M. P. Sulzer, S. A. González, R. B. Kerr, and L. Waldrop (2011), Response of the topside ionosphere over Arecibo to a moderate geomagnetic storm, J. Atmos. Sol. Terr. Phys., 73(11-12), 1568–1574.

Golovkov, V.P., Zvereva, T.I., and Chernova, T.A., (2007), Space–time modeling of the main magnetic field by combined methods of spherical harmonic analysis and natural orthogonal components, Geomagn. Aeron., 47, 256–263.

Golovkov, V. P., N. E. Papitashvili, Y. S. Tyupkin, and E. P. Kharin, (1978), Separation of geomagnetic field variations on the quiet and disturbed components by the MNOC, Geomagn. Aeron., Engl. Transl., 18, 511.

Greenspan et al., Characteristics of the thermal plasma monitor (SSIES) for the defense meteorological satellite program (DMSP) spacecraft S8 though S10, , Tech. Rep. AFGL- TR-86- 0227, Air Force Geophys. Lab., Hanscom Air Force Base, Mass., 1986.

Greenspan, M. E., W. J. Burke, F. J. Rich, W. J. Hughes, R. A. Heelis, DMSP F8 observations of the mid-latitude and low-latitude topside ionosphere near solar minimum, J. Geophys. Res., 99, 3817, 1994.

Gulyaeva, T. L., (2009), Proxy for the ionospheric peak plasma density reduced by the solar zenith angle, Earth Planets Space, 61, p. 629-631.

Gustafsson, G., A revised corrected geomagnetic coordinate system, Ark. Geofys., 5, 595-616, 1970.

Hagan, M. E., A. Maute, R. G. Roble, A. D. Richmond, T. J. Immel, and S. L. England (2007), Connections between deep tropical clouds and the Earth's ionosphere, Geophys. Res. Lett., 34, L20109, doi:10.1029/2007GL030142.

Hagan, M., J. Forbes, Migrating and nonmigrating semidiurnal tides in the upper atmosphere excited by tropospheric latent heat release. J. Geophys. Res. 108(A2), 1062 (2003). doi:10.1029/2002JA009, 466.

Hagan, M.E., J.M. Forbes, Migrating and nonmigrating diurnal tides in the middle and upper atmosphere excited by tropospheric latent heat release. J. Geophys. Res.-Atmos. 107, 4754 (2002). doi:10.1029/2001JD001236.

Hannachi, A, I. T. Jolliffe and D. B. Stephenson, (2007), Empirical orthogonal functions and related techniques in atmospheric science: A review, Int. J. Climatol., 27, 1119–1152.

Hanson, W. B. and R. A. Heelis, (1975), Techniques for Measuring Bulk Gas-Motions from Satellites, Space Sci. Instrum., 1, 493.

Hartman, W. A. and Heelis, R. A.: Longitudinal variations in the equatorial vertical drift in the topside ionosphere, J. Geophys. Res., 112, A03305, doi:10.1029/2006JA011773, 2007.

Hawkins, J. M. and P. C. Anderson (2017), WN4 Variability in DMSP Ion Densities Across Season, Solar Cycle, and Local Time, J. Geophys. Res. 122, 8755-8769, doi:10.1002/2017JA024065.

He, M., L. Liu, W. Wan, J. Lei, and B. Zhao (2010), Longitudinal modulation of the O/N2 column density retrieved from TIMED/GUVI measurement, Geophys. Res. Lett., 37, L20108, doi:10.1029/2010GL045105.

Hedin, A. E., and H. G. Mayr, Solar EUV Induced Variations in the Thermosphere, J. Geophys. Res., 92, 869-875, 1987.

Hedin, A. E., Correlations between thermospheric density and temperature, solar EUV flux, and 10.7-cm flux variations, J. Geophys. Res., 89, 9828-9834, 1984.

Heelis, R. A., and W. R. Coley (2007), Variations in the low‐ and middle‐latitude topside ion concentration observed by DMSP during superstorm events, J. Geophys. Res., 112, A08310, doi: 10.1029/2007JA012326.

Heelis, R.A., Coley, W.R., Burrell, A.G., Hairston, M.R., Earle, G.D., Perdue, M.D., Power, R.A., Harmon, L.L., Holt, B.J., Lippincott, C.R., 2009. Behavior of the O+/H+ transition height during the extreme solar minimum of 2008. Geophys. Res. Lett. 36, L00C03. http://dx.doi.org/10.1029/2009GL038652.

Hinteregger, H. E., (1981), Representation of Solar EUV Fluxes for Aeronomical Applications, Adv. Space Res., 1, 39–52.

Hysell, D. L., J. L. Chau, and J. D. Huba (2009), Topside measurements at Jicamarca during solar minimum, Ann. Geophys., 27, 427–439.

Immel, T. J., E. Sagawa, S. L. England, S. B. Henderson, M. E. Hagan, S. B. Mende, H. U. Frey, C. M. Swenson, and L. J. Paxton (2006), Control of equatorial ionospheric morphology by atmospheric tides, Geophys. Res. Lett., 33, L15108, doi:10.1029/2006GL026161.

Jacchia, L. G., and J. Slowey, A study of the variations in the thermosphere related to solar activity, Space Res., 13, 343-348, 1973.

Jakowski, N., A. Wehrenpfennig, S. Heise, Ch. Reigber, H. Lühr, L. Grunwaldt, T. K. Meehan (2002), GPS radio occultation measurements of the ionosphere from CHAMP: Early results, Geophys. Res. Lett., 29, L200210, doi:10.1029/2001GL014364.

Jin, H., Y. Miyoshi, H. Fujiwara, H. Shinagawa, Electrodynamics of the formation of ionospheric wave number 4 longitudinal structure. J. Geophys. Res., Space Phys. 113, 9307 (2008). doi:10.1029/2008JA013301.

Judge, D. L., McMullin, D. R., Ogawa, H. S., Hovestadt, D., Klecker, B., Hilchenbach, M., Mobius, E., Canfield, L. R., Vest, R. E., Watts, R., Tarrio, C., Kuehne, M., and Wurz, P., (1998), First solar EUV irradiances obtained from SOHO by the CELIAS/SEM, Solar Phys., 177, 161– 173.

Kelley, M., The Earth's Ionosphere: Plasma Physics and Electrodynamics, Academic Press, Inc., 1989.

Kendall, M. G. and Stuart, A., (1976), The Advanced Theory of Statistics, Vol. 3, 293– 323, Charles Griffin, London.

Kil, H., E. R. Talaat, S.-J. Oh, L. J. Paxton, S. L. England, and S.-J. Su (2008), Wave structures of the plasma density and vertical E × B drift in low-latitude F region, J. Geophys. Res., 113.

Kil, H., S.-J. Oh, M. C. Kelley, L. J. Paxton, S. L. England, E. Talaat, K.-W. Min, and S.-Y. Su (2007), Longitudinal structure of the vertical E × B drift and ion density seen from ROCSAT-1, GRL 34.

Liu, H., and S. Watanabe (2008), Seasonal variation of the longitude structure of the equatorial ionosphere: Does it reflect tidal influence from below? J. Geophys. Res., 113, A08315, doi:10.1029/2008JA013027.

Liu, L., H. Huang, Y. Chen, H. Le, B. Ning, W. Wan, and H. Zhang (2014), Deriving the effective scale height in the topside ionosphere based on ionosonde and satellite in situ observations, J. Geophys. Res. Space Physics, 119, 8472–8482, doi:10.1002/2014JA020505.

Liu, L., He, M., Wan, W., Zhang, M.-L., 2008. Topside ionospheric scale heights retrieved from Constellation Observing System for Meteorology, Ionosphere, and Climate radio occultation measurements. J. Geophys. Res. 113, A10304. http://dx.doi.org/10.1029/2008JA013490.

Liu, L., W. Wan, X. Yue, B. Zhao, B. Ning, and M.-L. Zhang (2007b), The dependence of plasma density in the topside ionosphere on solar activity level, Ann. Geophys., 25(6), 1337–1343.

Lühr, H., M. Rother, K. Häusler, P. Alken, and S. Maus (2008), The influence of nonmigrating tides on the longitudinal variation of the equatorial electrojet, J. Geophys. Res., 113, A08313, doi:10.1029/2008JA013064.

Lühr, H., and C. Xiong (2010), IRI‐2007 model overestimates electron density during the 23/24 solar minimum, Geophys. Res. Lett., 37, L23101, doi:10.1029/2010GL045430.

MacPherson, B., S. A. González, G. J. Bailey, R. J. Moffett, and M. P. Sulzer (1998), The effects of meridional neutral winds on the O+-H+ transition altitude over Arecibo, J. Geophys. Res., 103, 29,183–29,198, doi:10.1029/98JA02660.

Nigussie, M., S.M. Radicella, B. Damtie, B. Nava, E. Yizengaw, K. Groves (2013), Validation of the NeQuick 2 and IRI-2007 models in East-African equatorial region, J Atmos Sol-Terr Phys, doi:10.1016/j.jastp.2013.04.016.

Nsumei, P., B. W. Reinisch, X. Huang, and D. Bilitza (2012), New Vary-Chap profile of the topside ionosphere electron density distribution for use with the IRI model and the GIRO real time data, Radio Sci., 47, RS0L16, doi:10.1029/2012RS004989.

Oberheide, J., J. M. Forbes, K. Häusler, Q. Wu, and S. L. Bruinsma (2009), Tropospheric tides from 80 to 400 km: Propagation, interannual variability, and solar cycle effects, J. Geophys. Res., 114, D00I05, doi:10.1029/2009JD012388.

Pancheva, D., and P. Mukhtarov (2010), Strong evidence for the tidal control on the longitudinal structure of the ionospheric F-region, Geophys. Res. Lett., 37, L14105, doi:10.1029/2010GL044039.

Park, J., H. Lühr, K.W. Min, Characteristics of F-region dynamo currents deduced from CHAMP magnetic field measurements. J. Geophys. Res., Space Phys. 115, A10302 (2010). doi:10.1029/2010JA015604.

Pawlowski, D.J., Ridley, A.J. (2008), Modeling the thermospheric response to solar flares, J. Geophys. Res. 113, A10309.

Pedatella, N. M., J. M. Forbes, A. Maute, A. D. Richmond, T.-W. Fang, K. M. Larson, and G. Millward (2011), Longitudinal variations in the F region ionosphere and the topside ionosphere- plasmasphere: Observations and model simulations, J. Geophys. Res., 116, A12309, doi:10.1029/2011JA016600.

Pedatella, N. M., J. M. Forbes, and J. Oberheide (2008), Intra-annual variability of the low-latitude ionosphere due to nonmigrating tides, Geophys. Res. Lett., 35, L18104, doi:10.1029/2008GL035332.

Reinisch, B.W., and D. Bilitza (2017), Preface: The International Reference Ionosphere (IRI) at equatorial latitudes, Adv. Space Res., 60, 205-205, doi:10.1016/j.asr.2017.05.006.

Reinisch, B.W., Huang, X., 2001. Deducing topside profiles and total electron content from bottomside ionograms. Adv. Space Res. 27, 23–30. http://dx.doi.org/10.1016/S0273- 1177(00)00136-8.

Reinisch, B.W., Nsumei, P., Huang, X., Bilitza, D.K., 2007. Modeling the F2 topside and plasmasphere for IRI using IMAGE/PRI and ISIS data. Adv. Space Res. 39, 731–738. http://dx.doi.org/10.1016/j.asr.2006.05.032.

Ren, Z., W. Wan, L. Liu, and J. Xiong (2009), Intra-annual variation of wave number 4 structure of vertical E × B drifts in the equatorial ionosphere seen from ROCSAT-1, J. Geophys. Res.

Ren, Z., W. Wan, L. Liu, B. Zhao, Y. Wei, X. Yue, and R. A. Heelis (2008), Longitudinal variations of electron temperature and total ion density in the sunset equatorial topside ionosphere, Geophys. Res. Lett., 35, L05108, doi:10.1029/2007GL032998.

Rich, F. J., P. J. Sultan, and W. J. Burke, The 27-day variations of plasma densities and temperatures in the topside ionosphere, J. Geophys. Res.,108(A7), 1297, doi:10.1029/2002JA009731, 2003.

Richards, P. G., Fennelly, J. A., and Torr, D. G.: EUVAC: A solar EUV flux model for aeronomic calculations, J. Geophys. Res., 99, 8981–8992, 1994.

Rottman, G. (2005), The SORCE Mission, Sol. Phys., 230,1,7–25.

Sagawa, E., T. J. Immel, H. U. Frey, and S. B. Mende (2005), Longitudinal structure of the equatorial anomaly in the nighttime ionosphere observed by IMAGE/FUV, J. Geophys. Res., 110, A11302, doi:10.1029/2004JA010848.

Scherliess, L., D. C. Thompson, and R. W. Schunk (2008), Longitudinal variability of low-latitude total electron content: Tidal influences, J. Geophys. Res., 113, A01311, doi:10.1029/2007JA012480.

Smithtro, C.G., Solomon, S.C. (2008), An improved parameterization of thermal electron heating by photoelectrons, with application to an X17 flare, J. Geophys. Res. 113, A08307.

Solomon, S. C., T. N. Woods, L. V. Didkovsky, J. T. Emmert, and L. Qian (2010), Anomalously low solar extremeultraviolet irradiance and thermospheric density during solar minimum, Geophys. Res. Lett., 37, L16103, doi:10.1029/2010GL044468.

Stankov, S. M., N. Jakowski, S. Heise, P. Muhtarov, I. Kutiev, and R. Warnant, A new method for reconstruction of the vertical electron density distribution in the upper ionosphere and plasmasphere, J. Geophys. Res., 108(A5), 1164, doi:10.1029/2002JA009570, 2003.

Sun, W., W. Y. Xu, and S.-I. Akasofu (1998), Mathematical separation of directly driven and unloading components in the ionospheric equivalent currents during substorms, J. Geophys. Res., 103(A6), 11,695–11,700.

Tobiska, W. K., Chakrabarti, S., Schmidtke, G., and Doll, H., (1993), Comparative solar EUV flux for the San Marco ASSI, Adv. Space Res., 13, 255–259, doi:10.1016/0273- 1177(93)90022-4.

Tobiska, W. and Nusinov, A., (2006), ISO 21348 – Process for determining solar irradiances, in: 36th COSPAR Scientific Assembly, vol. 36 of COSPAR, Plenary Meeting, pp. 2621–2462.

Torr, M. R., D. G. Torr, and R. A. Ong, “Ionization frequencies for major thermospheric constituents as a function of solar cycle” vol. 6, no. 10, pp. 771–774, 1979.

Tsuda, T., and S. Kato, Diurnal non-migrating tides excited by a differential heating due to land-sea distribution, J. Meteorol. Soc. Jpn., 67, 43-54, 1989.

Tulasi Ram, S., C. H. Liu, S.-Y. Su, and R. A. Heelis (2010), A comparison of ionospheric O+/light-ion transition height derived from ion-composition measurements and the topside ion density profiles over equatorial latitudes, Geophys. Res. Lett., 37, L20107, doi:10.1029/2010GL045199.

Tulasi Ram, S., R. Heelis, V. S. Gowtam, K. K. Ajith, and S.-Y. Su (2015), Unique latitudinal shape of ion upper transition height (HT) surface during deep solar minimum (2008–2009), J. Geophys. Res. Space Physics, 120, 1419–1427, doi:10.1002/2014JA020647.

Tulasi Ram, S., S.-Y. Su, C. H. Liu, B. W. Reinisch, and L.-A. McKinnell (2009), Topside ionospheric effective scale heights (HT) derived with ROCSAT-1 and ground-based ionosonde observations at equatorial and midlatitude stations, J. Geophys. Res., 114, A10309, doi:10.1029/2009JA014485.

Venkatesh, K., P. V. S. Rama Rao, P. L. Saranya, D. S. V. V. D. Prasad, and K. Niranjan (2011), Vertical electron density and topside effective scale height (HT) variations over the Indian equatorial and low latitude stations, Ann. Geophys., 29, 1861-1872, doi:10.5194/angeo-29-1861- 2011.

Wan, W., J. Xiong, Z. Ren, L. Liu, M.-L. Zhang, F. Ding, B. Ning, B. Zhao, and X. Yue (2010), Correlation between the ionospheric WN4 signature and the upper atmospheric DE3 tide, J. Geophys. Res., 115, A11303, doi:10.1029/2010JA015527.

Wan, W., L. Liu, X. Pi, M.-L. Zhang, B. Ning, J. Xiong, and F. Ding (2008), Wavenumber-4 patterns of the total electron content over the low latitude ionosphere, Geophys. Res. Lett., 35, L12104, doi:10.1029/2008GL033755.

Wang, S., S. Huang, H. Fang, Y. Wang (2016), Evaluation and correction of the IRI2016 topside ionospheric electron density model, Adv. Space Res., 58, 1229-1241, doi:10.1016/j.asr.2016.06.020.

Warren, H.P., Mariska, J.T., Lean, J. (2001), A new model of solar EUV irradiance variability: 1. Model formulation, J. Geophys. Res. 106, 15745.

100

Webb, P. A., R. F. Benson, and J. M. Grebowsky (2006), Technique for determining midlatitude O+/H+ transition heights from topside ionograms, Radio Sci., 41, RS6S34, doi:10.1029/2005RS003391.

West, K. H., and R. A. Heelis (1996), Longitude variations in ion composition in the morning and evening topside equatorial ionosphere near solar minimum, J. Geophys. Res., 101(A4), 7951– 7960.

Williams, C. R., and S. K. Avery, 1996: Diurnal nonmigrating tidal oscillations forced by deep convective clouds. J. Geophys. Res., 101, 4079–4091.

Woods, T. N. and G. J. Rottman, (1990), Solar EUV Irradiance Derived from a Sounding Rocket Experiment on 10 November 1988, J. Geophys. Res, 95, 6227.

Woods, T.N. and Rottman, G.J. (2002), In: Mendillo, M., Nagy, A., Waite, J.H. Jr. (eds.) Comparative Aeronomy in the Solar System, Geophys. Monogr. Ser. 130, AGU, Washington, 221.

Woods, T. N., F. G. Eparvier, S. M. Bailey, P. C. Chamberlin, J. Lean, G. J. Rottman, S. C. Solomon, W. K. Tobiska, and D. L. Woodraska (2005), Solar EUV Experiment (SEE): Mission overview and first results, J. Geophys. Res., 110, A01312, doi:10.1029/2004JA010765.

Woods, T. N., Prinz, D. K., Rottman, G. J., (1996), Validation of the UARS solar ultraviolet irradiances: Comparison with the ATLAS 1 and 2 measurements, J. Geophys. Res., 101, 9541– 9570, doi: 10.1029/96JD00225.

Woods, T., Eparvier, F., Hock, R., Jones, A., Woodraska, D., Judge, D., Didkovsky, L., Lean, J., Mariska, J., Warren, H., McMullin, D., Chamberlin, P., Berthiaume, G., Bailey, S., Fuller- Rowell, T., Sojka, J., Tobiska, W., Viereck, R.: 2012, Extreme ultraviolet variability experiment (EVE) on the solar dynamics observatory (SDO): overview of science objectives, instrument design, data products, and model developments. Solar Phys. 275, 115–143, doi: 10.1007/s11207- 009-9487-6.

Yang, C., B. Zhao, J. Zhu, X. Yue, and W. Wan. An investigation of ionospheric upper transition height variations at low and equatorial latitudes deduced from combined COSMIC and C/NOFS measurements. Adv. Space Res. (2016), http://dx.doi.org/10.1016/j.asr.2016.11.024.

Yizengaw, E., H. Wei, M. B. Moldwin, D. Galvan, L. Mandrake, A. Mannucci, and X. Pi (2005), The correlation between mid‐latitude trough and the plasmapause, Geophys. Res. Lett., 32, L10102, doi:10.1029/2005GL022954.

Yonezawa, T., (1966), Theory of formation of the ionosphere, Space Sci. Rev., 5, 3–56.

101

Zhang, Y., S. England, L.J. Paxton, Thermospheric composition variations due to nonmigrating tides and their effect on ionosphere. Geophys. Res. Lett. 37, L17103 (2010). doi:10.1029/2010GL044313.

Zhao, B., W. Wan, L. Liu, X. Yue, and S. Venkatraman (2005), Statistical characteristics of the total ion density in the topside ionosphere during the period 1996–2004 using empirical orthogonal function (EOF) analysis, Ann. Geophys., 23, 3615–3631.

102

BIOGRAPHICAL SKETCH

Jessica Hawkins graduated from Plano Senior High School in 2007 as a National Merit Finalist and AP Scholar. She studied Computer Science and Physics at The University of Texas at

Dallas and was a member of the Collegium V honors program. As an undergraduate, she took several electives in diverse technical subjects such as Spacecraft Systems Engineering, Machine

Learning, Artificial Intelligence, and Compiler Design. She worked as an undergraduate teaching assistant in the physics department and enjoyed mentoring students and friends in physics. Her senior year, Jessica worked as a research assistant in the materials science department with teams studying III-V semiconductors and graphene, where she contributed to a paper which showed that exposing samples to atmosphere, rather than keeping samples in a clean ultra-high vacuum (UHV) environment, contributed to interfacial oxide re-growth. She also played violin with the UTD chamber music ensemble Musica Nova.

After graduating with her dual degrees in CS and Physics in 2012, Jessica worked for the summer as an associate engineer for Ball Aerospace in Albuquerque, NM. Her work at Ball

Aerospace involved physics-based modeling in C++ in support of national high-energy laser testing facilities.

Jessica joined the PhD program in physics at The University of Texas at Dallas in Fall 2012, and in 2013 joined the W. B. Hanson Center for Space Sciences, where she applied her dual background in CS and physics to analyzing satellite measurements of the Earth's ionosphere.

She attended several conferences during her graduate career, including the Coupling, Energetics

103

and Dynamics of Atmospheric Regions (CEDAR) conferences, American Geophysical Union

(AGU) conferences, the National Center for Atmospheric Research (NCAR) Advanced Study

Program, and the International Symposium on Recent Observations and Simulations of the Sun-

Earth System (ISROSES) in Varna, Bulgaria. She also traveled to Lima, Peru to visit the

Jicamarca Radio Observatory for an incoherent scatter radar (ISR) summer school in 2014. She published articles on the topside ionospheric response to solar extreme ultraviolet variability and on signatures of the wave number 4 tide in the topside ionosphere.

Jessica is a member of the Dallas Makerspace and of the Other Worlds Writers Workshop for science fiction and fantasy. Her other interests include playing the violin, creative writing, studying science history and scientific subjects outside her specialty, tutoring and mentoring.

104

CURRICULUM VITAE

Jessica Hawkins

EDUCATION

THE UNIVERSITY OF TEXAS AT DALLAS, Richardson, Texas PhD Physics Expected May 2018 MS Physics May 2014, GPA 3.21 BS Computer Science May 2012, GPA 3.31 BS Physics May 2012, GPA 3.31

SAT Scores: 760 Math / 740 Writing / 750 Verbal SAT II Scores: 800 Math Level II / 760 Physics GRE Scores: 760 Quantitative / 720 Verbal

Collegium V Honors Program National Merit Full Scholarship AP Scholar

RESEARCH AND PROFESSIONAL EXPERIENCE

Teaching/Research Assistant, W.B. Hanson Center for Space Sciences May 2013 – present

For my PhD dissertation research, I investigated two important sources of variability in the Earth's ionosphere, the solar extreme-ultraviolet (EUV) irradiance and atmospheric non- migrating tides, as well as the effectiveness of the International Reference Ionosphere (IRI) model at reproducing the shape of the topside ion density profiles. My research addresses objectives in NASA’s Recommended Roadmap for Science and Technology and the National Research Council’s Decadal Survey. I have presented my work at several conferences: • Attended the National Center for Atmospheric Research (NCAR) Advanced Study Program in 2015 and presented a poster. • Presented posters at CEDAR 2014 and 2015, and AGU 2014 and 2015 conferences. • Traveled to Lima, Peru to visit the Jicamarca Radio Observatory for an incoherent scatter radar (ISR) summer school in 2014.

Associate Engineer, Ball Aerospace (Albuquerque, NM) May 2012 – Aug. 2012

Physics-based modeling in support of national high-energy laser (HEL) testing facilities, including atmospheric effects and weather, coordinate systems, materials, and scattering. Original software design in C++ as well as synergy with existing code.

Research Assistant, Materials Science Dept. at UTD May 2011 – Dec. 2011

I worked in a materials characterization lab for III-V semiconductors and graphene. My responsibilities included running tests and maintaining ultra-high vacuum equipment, especially X-ray Photoelectron Spectroscopy (XPS). I contributed to a study which showed that exposing samples to atmosphere (rather than keeping samples in a clean UHV environment) contributes to interfacial oxide re-growth and potentially misleading conclusions.

PUBLICATIONS

Hawkins, J. M. and P. C. Anderson (2017), WN4 Variability in DMSP Ion Densities Across Season, Solar Cycle, and Local Time, J. Geophys. Res. 122, 8755-8769, doi:10.1002/2017JA024065.

Anderson, P. C., and J. M. Hawkins (2015), Topside Ionospheric Response to Solar EUV Variability, J. Geophys. Res. Space Physics , doi:10.1002/2015JA021202.

S. McDonnell, H. Dong, J. M. Hawkins, B. Brennan, M. Milojevic, F. S. Aguirre-Tostado, D. M. Zhernokletov, C. L. Hinkle, J. Kim, and R. M. Wallace, “Interfacial oxide re-growth in thin film metal oxide III-V semiconductor systems”, Appl. Phys. Lett. 100, 141606 (2012).

SKILLS AND PROJECTS

Data analysis methods and modeling for scientific applications, including Fourier analysis and filtering, dimensionality reduction (PCA/EOFs), regression and error analysis, matrix algebra and coordinate transformations.

Programming Languages: Python, C++, C, Java, IDL, MATLAB, Linux shell scripting

Spacecraft Systems Engineering Fall 2008 - Spring 2009 I worked with a team to design Attitude Determination and Control Systems (ADCS) for a picosatellite (CubeSat). This involved algorithm design based on classical mechanics, the spacecraft environment, and coordinate system conversions. I evaluated the effectiveness of design schemes under severe space, mass, and power constraints, and collaborated with student teams responsible for power and telemetry.

TEACHING AND OUTREACH

• Teaching Assistant at UT Dallas (9 semesters) • Volunteer for Sally Ride Science (2013) • Volunteer for Introduce a Girl to Engineering Day (2015)