Equatorial Atmospheric and Ionospheric Modeling at ... · Kwajalein Atoll is located in the Marshall Islands in the Pacific Ocean, approximately four thousand ki- ... period at ALTAIR’s

• HUNT, CLOSE, COSTER, STEVENS, SCHUETT, AND VARDAROEquatorial Atmospheric and Ionospheric Modeling at Kwajalein Missile Range

VOLUME 12, NUMBER 1, 2000 LINCOLN LABORATORY JOURNAL 45

Equatorial Atmospheric andIonospheric Modeling atKwajalein Missile RangeStephen M. Hunt, Sigrid Close, Anthea J. Coster, Eric Stevens,Linda M. Schuett, and Anthony Vardaro

■ During the last peak of the solar activity cycle, in the late 1980s and early1990s, the spatial and temporal variations of the equatorial ionosphere had asignificant impact on the measurement accuracy of the ARPA Long-RangeTracking and Instrumentation Radar (ALTAIR) at the Kwajalein Missile Range(KMR), primarily because of poorly modeled ionospheric range and elevation-angle bias errors. As missile-reentry-vehicle and space-surveillance missions havedeveloped at KMR, improvements in ALTAIR measurement accuracy haveillustrated that unmodeled atmospheric refraction is the principal error sourcelimiting sensor accuracy. Modeling equatorial ionospheric activity and thenremoving its effect on KMR data is particularly challenging, since the radar islocated beneath a complex region of the earth’s ionosphere. To characterize thespatial and temporal electron distribution in the ionosphere, real-timemeasurement data are required to adjust the models to match actual conditions.An effort to better characterize both peak and disturbed ionospheric conditionsin real time has resulted in the ionospheric error-correction model (IECM).This article discusses the KMR atmospheric models, and presents representativeequatorial ionospheric data as viewed by ALTAIR along with an evaluation ofthe IECM at removing these range and elevation-angle bias errors.

T and ballistic-missile reen-try vehicles, high-power radar systems at theKwajalein Missile Range (KMR) transmit sig-

nals at frequencies spanning 160 MHz to 95 GHz.Frequency-dependent signal-propagation effects pro-duced in the atmosphere can degrade the accuracy ofmeasurements collected by these radar systems. Sincethese radars collect position, velocity, and radar cross-section data on objects that traverse the troposphere(the lower atmosphere) and the ionosphere (aboveninety kilometers), we have spent many years devel-oping spatial-dependent and time-dependent radio-frequency (RF) signal-propagation models that miti-gate atmospheric signal degradation. These models

consist of two main components: (1) the spatial-vary-ing and time-varying nature of the index of refractionin each region of the atmosphere, and (2) the corre-sponding RF signal-propagation effects themselves.

Kwajalein Atoll is located in the Marshall Islandsin the Pacific Ocean, approximately four thousand ki-lometers southwest of Hawaii. It consists of approxi-mately one hundred small islands resting on a coralreef formation eleven hundred kilometers north ofthe geographic equator. The two principal islands areKwajalein and Roi-Namur. The KMR radars are lo-cated on Roi-Namur, where they support ballistic-missile-test and space-surveillance missions. The ra-dars at KMR collect highly accurate metric and radar


46 LINCOLN LABORATORY JOURNAL VOLUME 12, NUMBER 1, 2000

cross-section data to characterize missile-reentry ve-hicles as well as help maintain the U.S. Space Com-mand catalog of artificial satellites. The primary radarfor this latter task is the Advanced Research ProjectsAgency (ARPA) Long-Range Tracking and Instru-mentation Radar, or ALTAIR, which is a two-fre-quency (160 and 422 MHz) radar located on Roi-Namur at 4° north geomagnetic latitude.

A notable and significant difference between mis-sile-reentry-vehicle data processing and space-surveil-lance data processing is that reentry-vehicle data arepost-processed to obtain maximum metric accuracy,while space-surveillance data must be accurately cor-rected in real time and immediately transmitted toU.S. Space Command. ALTAIR performs most of thesatellite tracking for U.S. Space Command at KMR,and it is dedicated to this task 128 hours per week.

This article provides an overview of the KMRequatorial atmospheric environment, the first-orderALTAIR atmospheric-correction model, and the newionospheric error-correction model (IECM). The ar-ticle also presents radar calibration data to illustratethe accuracy obtained while correcting metric obser-vations with the IECM and the first-order ALTAIRmodel. We also present an overview of two-frequencytotal electron content (TEC) data reduction and itsapplication, along with a section describing what welearned during years of experimentation with GlobalPositioning System (GPS) receivers for the purpose ofreal-time TEC data collection.

Phenomenology of the Equatorial Ionosphere

As the altitude above the earth increases, the densityof the atmosphere decreases. The first ninety kilome-ters of the atmosphere are basically neutral and con-sist of a homogeneous mixture of several gases withthree primary layers: the troposphere (approximatelyzero to fifteen kilometers in altitude), the stratosphere(fifteen to fifty kilometers in altitude), and the meso-sphere (fifty to ninety kilometers in altitude) [1].These atmospheric layers are generally distinct; theyare well mixed by turbulence, which is primarily theresult of thermal winds coupled with forces from theearth’s rotation. During the daytime, weak ionizationbegins to develop in the mesosphere, which serves as atransition region to the ionosphere.

Above an altitude of approximately ninety kilome-ters the gases are strongly ionized by solar radiation,as shown in Figure 1. At this higher altitude, which isthe region known as the ionosphere, the ionized gasesno longer tend to be uniformly mixed; heavy gas mol-ecules and atoms such as nitrogen and oxygen arefound near the bottom of the ionosphere, and lightergas molecules and atoms such as hydrogen and he-lium are found near the top. Free electrons and ions(both molecular and atomic) are present in nearlyequal numbers in this region, which results in anoverall (net) neutral charge for the particle distribu-tion. The distribution and dynamics of these particlesin the ionosphere affect the propagation of radiowaves and the corresponding radar measurements ofobjects within or above the ionosphere.

A plasma is defined as a collection of positivelycharged atoms and molecules with negatively chargedelectrons in approximately equal numbers, with anoverall net charge of zero. Because the earth has a

Alti

tude

(km

)

800

700

600

500

400

300

200

100

00 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Electron density (1012 electrons/m3)

DaytimeNightime

DE

F1

F2F

(H+, He+, e–)

(N2+, O+, O2

+, NO+, e–)

FIGURE 1. Typical altitude profile of ionospheric electrondensity. Free electrons, molecular ions, and atomic ionsform into distinct layers during the daytime—the D, E, F1,and F2 regions—while the F1 and F2 regions combine dur-ing the diminished solar activity of the evening hours toform the F region. Heavier atoms and molecules (N2, O, O2)tend to reside near the bottom of the ionosphere, whilelighter atoms and molecules (H, He) are found near the top.The net charge of the plasma found in the ionosphere isgenerally neutral. The Kwajalein Missile Range (KMR) atmo-spheric correction models take into account the radio-fre-quency (RF) propagation effects caused by the presence offree electrons.



strong magnetic field relative to any induced mag-netic fields resulting from the motion of charged par-ticles in the ionospheric plasma, the plasma distribu-tion resulting from the dynamical aspects of theionosphere is normally dominated by the earth’s di-pole magnetic field. At the equator and above theionosphere, an enhanced electron density region ex-ists (called the plasmasphere) that can extend to sev-eral earth radii. There is no static boundary betweenthe ionosphere and the plasmasphere; for this articlewe define the region above two thousand kilometersto be the plasmasphere.

During the daytime, several distinct layers areformed in the upper atmosphere, including the D re-gion within the mesosphere (50 to 90 km altitude),and the E region (90 to 120 km altitude) and the F1and F2 regions (120 to 600+ km altitude) within theionosphere. At nighttime the D and E regions be-come neutral and the F1 and F2 layers often mergeinto a single F region, characterized by reduced elec-tron density, as shown in Figure 1. Distinct iono-spheric regions develop for three primary reasons: (1)the solar spectrum deposits its energy at variousheights, depending on the absorption characteristicsof the atmosphere, (2) the physics of recombinationdepends on density, and (3) the composition of theatmosphere changes with height [2].

The location of the peak electron density and theshape of the ionospheric layers depend greatly on anumber of factors and are therefore highly variable. Astandard ionospheric profile can be generated byChapman theory, which provides a closed-form ana-lytic description of ion (and corresponding electron)production in the upper atmosphere, and results in aquasi-exponential distribution of free electrons withaltitude. The terms “ionospheric anomaly” and“ionospheric irregularity” are used to describe condi-tions that are not described by the Chapman ioniza-tion function [3]. (Even in its quiet state, with noanomalies or irregularities, the ionosphere is often notdescribed by a simple Chapman function, which iswhy more sophisticated models are generally re-quired.) An ionospheric anomaly can be thought of asa medium-to-large-scale disturbance in the back-ground ionosphere, such as the mid-latitude troughregions associated with magnetic storms. Ionospheric

irregularities are typically small-to-medium-scaledensity fluctuations in the background ionospherethat are routinely present in the ionosphere, such asgravity waves or small-scale irregularities associatedwith scintillation.

Figure 2 shows GPS TEC data to illustrate the di-urnal variation of the ionosphere. Rapid fluctuationsin TEC during the early evening hours indicate thepresence of ionospheric irregularities, which are moreprevalent during increased solar activity. Figure 3 il-lustrates periodic changes in ionospheric activity thatinclude seasonal and solar-cycle variations, all ofwhich can be thought of as predictable. Figure 3 alsoshows the corresponding VHF range delay, which canbe as great as 1700 m at ALTAIR’s vertical (above50°). When ALTAIR views satellites at the horizon,this VHF range delay is approximately three timesgreater (approximately 5100 m) because of the in-creased amount of atmosphere present along the ra-dar line of sight.

The equatorial region is dominated by the anoma-lies and irregularities mentioned previously and out-lined in Table 1. The equatorial ionosphere is strongly

80

40

Local time of day (hours)

0

20

60

0 126 1218 0 6

Scintillation periodGP

S T

EC (1

016 e

lect

rons

/m2 )

FIGURE 2. Illustration of the diurnal variation of the iono-sphere, using Global Positioning System (GPS) total elec-tron content (TEC) data collected during a thirty-six-hourperiod at ALTAIR’s vertical (above 50° elevation). The rapidfluctuations in TEC shown during the scintillation period inthe early evening hours illustrate the presence of iono-spheric irregularities, which are more prevalent during in-creased solar activity. The increasing and decreasing trendseven at high elevation angle indicate the complex ion distri-bution at ALTAIR’s vertical.



affected by the north-south orientation of magnetic-field lines lying parallel to the surface of the earth atthe equator. At the equator, a strong eastward iono-spheric current flows by day over a narrow latitudinalstrip (a few degrees of latitude wide) along the mag-

netic equator at an altitude near 100 km. This currentis known as the equatorial electrojet. Also present isthe occurrence of the Appleton, or equatorial, anom-aly. This anomaly is the electron-density depletion re-gion that exists at the geomagnetic equator along withregions of enhanced electron density that peak at ap-proximately 15° to 20° north and south latitude, asshown in Figure 4. This distribution is caused by thevertical electrodynamic drift of the equatorial plasmaand its later diffusion along the north-south geomag-netic field lines (B). An east-west–oriented electricfield (E) at the geomagnetic equator perpendicular tothe north-south magnetic field causes this verticalE × B drift. The most disturbed scintillation regionwas observed at one of the crests of the Appletonanomaly during a sunspot maximum period [4].

Also observed on the equator are F-region irregu-larities caused by the existence of gravitational Ray-leigh-Taylor instabilities (GRTs). GRTs result froman interaction between the downward gradient of theearth’s gravitational field and the parallel orientationof the earth’s equatorial electric and magnetic fieldswith the surface of the earth. A detailed treatment onthe formation of this irregularity, which is beyond thescope of this article, is an area of ongoing research inequatorial aeronomy and plasma physics. Data on

FIGURE 3. The eleven-year solar cycle as viewed from thegeomagnetic equator. These data are two-frequency TECmeasurements collected at ALTAIR since 1983. They areused to make a daily adjustment to the first-order ALTAIRionospheric-correction model. One TEC unit is equivalent to15.3 meters of range delay at VHF. Therefore, the magnitudeof the ionospheric range delay reached approximately 1500meters during the last peak in the solar activity cycle.

Table 1. Irregularities and Anomalies Prevalent at Equatorial Locations

Type of Ionosphere Distribution Time most Durationvariation region prevalent

Electrojet E region East-west All times <30 minutesirregularities (stronger in to hours

daytime)

Appleton Above the Enhancement at Late afternoon >4 hoursanomaly F region 10–20o north-south and early

(depletion at equator) evening

Rayleigh-Taylor F region Along After sunset >2.5 hoursinstabilities magnetic field and early

lines (north-south) evening

Magnetic-storm E region Global All times Secondseffects and F region to hours

200

100

Calendar year

0

50

150

84 9288 0096

3060

1530

0

765

2295

VH

F ra

nge

dela

y (m

)

TEC

(1016

ele

ctro

ns/m

2 )



these irregularities were collected at the Jicamarca in-coherent scatter radar in Peru at 12° south latitude[5]. Similar studies have been periodically conductedat ALTAIR since 1978 [6–9]. A continuous long-

term study at Kwajalein would provide a significantcontribution to the understanding of equatorial iono-spheric irregularity phenomena. The outcome of thisscientific research would apply directly to the devel-opment of techniques to mitigate the effects of thesephenomena on RF signal propagation as applied toradar measurements and communications.

Large electron-density depletion regions persist inthe F region during the early evening to early morn-ing hours, and have complex scattering characteristicsthat are a source of intense RF scintillation and fre-quency-spreading (spread-F) conditions. RF scintilla-tion describes the rapid fluctuation of amplitude andphase, while spread-F describes the signal frequencydispersion experienced by high-frequency iono-spheric sounders during disturbed ionospheric condi-tions. Spread-F is a general term used to denote theexistence of an ionospheric irregularity in RF mea-surement data. This term originated from data col-lected with vertical ionospheric sounders, whichtransmit across a broad high-frequency spectrum, be-cause the resulting frequency signatures are spread inrange relative to periods where irregularities are non-existent [10]. As an example, Figure 5 shows 49.92-MHz coherent backscatter data collected by the Jica-marca radar facility in Peru at 12° south latitude and

30

10

150

140

130

120

110

100

90

80

60

70

50

40

Geographic longitude (deg E)

Geo

grap

hic

latit

ude

(deg

N)

TEC

(1016

ele

ctro

ns/m

2 )

–15

–5

20

150 170160 190180

0

–10

5

15

25

FIGURE 4. TEC map from the ionospheric error-correctionmodel (IECM), illustrating the characteristic distribution ofthe Appleton anomaly. The dark red portions near 20° lati-tude are high electron-density regions. The blue section isthe electron-depletion formation at the equator caused bythe upward (E × B) and later north-south drift of electronsalong the earth’s magnetic field lines. The Appleton anom-aly forms at the geomagnetic equator in the late afternoonand persists into the evening hours.

800

1000

1200

400

Local time (hours) on 17 December 1994

Alti

tude

(km

)

0

200

600

20

10Rx (dB)

30–10

2221 23

FIGURE 5. Range-time-intensity plot of received power (Rx) from the Jicamarca radar facility in Peruat 12° south latitude [11]. The Jicamarca operating frequency was 49.92 MHz. These large structuresare most likely due to electron-density depletion regions caused by gravitational Rayleigh-Taylor in-stabilities. Presumably, Jicamarca experiences ionospheric conditions similar to those at ALTAIR,given its close proximity to the equator. Most of ALTAIR’s operating time is spent on space-surveil-lance and missile-reentry-vehicle tests, but its incoherent scatter radar capability is periodically usedto support experimental research to better characterize the equatorial atmosphere. (Image courtesyof Geophysical Research Letters [11], copyright by the American Geophysical Union.)



285° east longitude, which show spread-F conditionscaused by the presence of a large irregularity.

Figure 6 illustrates radar measurements collectedby ALTAIR while it was tracking eight reentering ob-jects, all on the same basic flight-path trajectory, asthey traversed one of these equatorial irregularity re-gions. The trajectory and the variation in radar-returnsignal characteristics (RF scintillation) on these ob-jects are consistent with the size and distribution ofthe irregularity shown in Figure 5. The latitude-longi-tude-height orientation of the ALTAIR measure-ments illustrates evidence of an irregularity regionthat extends to the north in latitude and to the west inlongitude, and spanning at least a hundred kilometersin altitude.

Modeling the Atmosphere

Radar tracking of satellites is complicated by severaleffects introduced by the earth’s atmosphere. Figure 7illustrates how the atmosphere distorts RF measure-ments and causes the position of targets to appear off-set relative to their true location. Apparent positionrefers to the measured range and elevation angle ofthe satellite in the presence of the atmosphere. The

true position refers to the range and elevation angle ofthe satellite with atmospheric effects removed.

During the 1960s and the early 1970s, an effortwas initiated to model the effects of the changing at-mosphere (space weather effects on the ionosphereand terrestrial weather on the neutral atmosphere) onKMR radar measurements. The goal was to improvethe accuracy of the radar data. The models developedfor both the ionosphere and the neutral atmospherewere designed to create a true estimate of target posi-tion by removing the atmospheric effects that pro-duce a distorted apparent target position. RF propa-gation effects include elevation-angle bending, rangedelay, Doppler frequency shifts, polarization rotation,frequency spreading, and rapid amplitude and phasescintillation. The KMR atmospheric models are pri-marily focused on correcting for the effects that areresponsible for radar-range (time delay) and eleva-tion-angle bias errors.

The early atmospheric refraction models devel-oped for KMR sensors were produced by D.W. Bloodand D. Kwan. These models are used to remove theeffects of the neutral atmosphere (below 90 km) onradar measurements. Later, ionospheric correctionmethods were added by T. Pass, using simultaneoustwo-frequency radar range measurements combinedwith an elevation-angle mapping function. As AL-TAIR single-frequency tracking requirements ex-panded for space surveillance, a first-order verticalionospheric range-delay model was added by R.McSheehy [12]. This first-order model maintains acontinuous knowledge of ALTAIR’s vertical iono-spheric range delay, which enabled the operation ofthe radar for extended periods of time without theneed to track at two frequencies to perform iono-spheric correction.

In 1989, a measurement-accuracy shortfall was ex-perienced at ALTAIR during the last peak of the solaractivity cycle because of the limitations of the first-order single-frequency ionospheric model. This accu-racy problem motivated the effort to replace the first-order ionospheric model with the IECM, which wasdeveloped for ALTAIR under a small-business inno-vative research contract by Research Associates ofSyracuse, Inc., Syracuse, New York, and Computa-tional Physics, Inc., Newton, Massachusetts [14, 15].

FIGURE 6. ALTAIR data collected while eight missile-reen-try vehicles traversed an equatorial ionospheric irregularity.The data shown here were measured while the vehicles wereinside the irregularity. The RF return signals exhibited rapidfluctuations in amplitude and phase. As the spatial distribu-tion of the measurement data increases in altitude, it alsoextends to the north in latitude and west in longitude. Thisdistribution is indicative of an equatorial electron-depletionregion resulting from a gravitational Rayleigh-Taylor insta-bility, such as shown in Figure 5.

800

700

Longitude

Latitude

Alti

tude

(km

)

650

750



The focus of the IECM development effort was toincrease the accuracy of the real-time space-surveil-lance measurement data by providing an improvedionospheric modeling capability. The IECM com-bines a climatological low-latitude ionospheric modeldriven by real-time Global Positioning System (GPS)and ALTAIR total electron content (TEC) measure-ments. The early atmospheric-correction model de-veloped by Blood and Kwan is used in the IECM toremove neutral atmospheric effects. The physics andchemistry of the upper atmosphere are based on thelow-latitude ionospheric model developed by D.N.Anderson et al. in 1987 [16] and a plasmasphericmodel developed by D.L. Gallagher [17]. Figure 8 il-lustrates the process of using the IECM to accuratelymeasure satellite range and elevation angle. The en-hanced capability of the IECM resulted in a signifi-cant improvement in the range accuracy of metricdata collected during ALTAIR’s continuous high-vol-ume satellite-tracking operation. The IECM has beenoperational at ALTAIR since the spring of 1998.

The Index of Refraction in the Neutraland Ionized Atmosphere

To support the collection of accurate RF measure-ments (azimuth, elevation angle, range, and range

Plasmasphere

Apparent position

Ray pathDirect path

30 km

90 km

2000 km

Ionosphere

Troposphere

n 1≈

UN

ITE

DS

TAT

ES

15 km

Mesosphere

50 km

Ionized atmosphere

Neutralatmosphere

Stratosphere

n 1≈

n > 1

n < 1

True position

FIGURE 7. Illustration of the two-dimensional ray-traced path of a transmitted RF signal from the earth’s surface through theneutral and ionized atmosphere. The ray-tracing algorithm uses range and elevation-angle corrections derived from the alti-tude-dependent index of refraction n to accurately determine satellite position.

rate) made by KMR radars, we require knowledge ofthe real-time atmospheric index of refraction as afunction of altitude for the neutral atmosphere, iono-sphere, and plasmasphere. The effects on radio signalsin the neutral and ionized regions of the atmosphereare different. The ionosphere and plasmasphere aredispersive mediums with an index of refraction thatis, to first order, a function of electron density and ra-dio-wave frequency. The refraction effects are signifi-cantly larger at lower frequencies (as a result, the ef-fect of the ionosphere at VHF is nearly a hundredtimes the effect at L-band). The index of refraction inthe neutral atmosphere is a function of pressure, tem-perature, and humidity.

When RF signals propagate through the neutral at-mosphere, they are subject to a spatial-varying andtime-varying index of refraction that, to a first ap-proximation, affects all KMR radars equally. The neu-tral atmosphere is thus a nondispersive medium at ra-dio-wave frequencies. The KMR model of the indexof refraction uses a mean refractivity profile to modelthe neutral atmosphere. This mean profile is based onradiosonde data collected at Kwajalein from 1967 to1972. The model has a piecewise exponential formand is used to provide range and elevation-angle biascorrections. Some ionization exists in the upper



stratosphere and lower mesosphere (from fifty toninety kilometers in altitude), even though the KMRmodels define the dominating constituents as neutral.The ionospheric models do not come into effect untilapproximately ninety kilometers in altitude. The ex-ponential refractivity N, which is related to the indexof refraction n by the relation N = (n – 1) × 106, is de-fined as

N e h= −α β , (1)

where α and β are altitude-dependent constants, andh is used to specify seven altitude regions. Table 2 liststhe altitude ranges of the seven regions, and shows theassociated values of the constants α and β in each ofthe regions.

Table 2. Blood and KwanExponential Refractivity Model

Profile Altitude range α βregion (km) (N-units) (km –1)

1 0.0 to 0.3048 378.0 0.21522

2 0.3048 to 0.6096 368.58 0.13255

3 0.6096 to 0.9144 375.57 0.16339

4 0.9144 to 4.1453 374.0 0.15879

5 4.1453 to 8.9916 329.0 0.12795

6 8.9916 to 15.0266 293.0 0.11549

7 above 15.0266 780.0 0.18045

FIGURE 8. Ionospheric measurement and correction of space-surveillance range and elevation-angle data with the ARPALong-Range Tracking and Instrumentation Radar (ALTAIR). Real-time two-frequency (L1 and L2) GPS range data, from a con-stellation of GPS satellites, and ALTAIR two-frequency (UHF and VHF) measured range data are used to support real-timeionospheric correction. These GPS and ALTAIR data are used by the new real-time ionospheric error-correction model (IECM)to compute range and elevation-angle corrections. Real-time corrections are supplied to the ALTAIR satellite-tracking softwareto remove ionosphere-induced range and elevation-angle errors from the ALTAIR measurements. Also shown is the oldALTAIR first-order atmospheric-correction model, which is adjusted by using a single ALTAIR measurement once per day.

GPS1

L1, L2

GPSN

GPS receiver

Corrected range and elevation-angle

measurements(sent to U.S. Space

Command)

Measured rangeand elevation angle(apparent satellite

position)

TECdata

IECM(updated every ten seconds)

UN

ITE

DS

TAT

ES

UN

ITE

DS

TAT

ES

L1, L2

Range andelevation

corrections

First-order ALTAIR correction model

(updated once per day)

Measure andapply corrections

ALTAIR

Two-frequency GPS range data

Two-frequency(UHF and VHF)radar range data



ALTAIR First-Order Single-FrequencyIonospheric Range-Delay Model

In the early 1980s, as the number of orbiting satellitesincreased, ALTAIR single-frequency tracking require-ments expanded dramatically for space surveillance.At this time, a first-order vertical ionospheric range-delay model was created to support the continuoussingle-frequency satellite-tracking and data-collectionoperations [12]. The first-order model was imple-mented as a function of time, azimuth, and elevation.The magnitude of the ionospheric range delay (or,similarly, the value of the TEC) at ALTAIR’s verticalis determined by the time-dependent component ofthe first-order model. TEC is proportional to theionospheric range delay at a particular radio fre-quency. The model is scaled each day by making asingle ALTAIR vertical TEC measurement and thenusing this measurement to compute the ratio of therange delay (or TEC) measurement to the modelvalue at the measurement time. This ratio is then usedto scale the vertical component of the first-order

model estimates for the duration of the tracking ses-sion, as illustrated in Figure 9.

The second component of the ALTAIR first-ordermodel is the radar elevation mapping function usedto extrapolate the vertical TEC value to an arbitraryelevation angle. Figure 10 shows the determination ofTEC at VHF, UHF, and L-band, as a function of el-evation angle, and Figure 11 shows the determinationof elevation-angle bias. During periods of low solaractivity this model provides satisfactory results. Dur-ing periods of moderate to high solar activity themodel generally fails to represent actual ionosphericconditions, resulting in degraded range and elevation-angle measurement accuracy by the radar system.

IECM

In 1989, near the peak of the last solar maximum, itbecame clear that the first-order model failed to meetthe ALTAIR satellite-tracking accuracy requirements.The effects of increased solar activity were revealed bythe examination of calibration data from the ALTAIRsystem. ALTAIR measurements (azimuth, elevation,

FIGURE 9. The ALTAIR first-order ionospheric range-delaymodel. This model, derived from ALTAIR ionospheric back-scatter data and the first-order worldwide ionosphericKlobuchar model [13], is used to maintain continuousknowledge of the ionospheric range delay above ALTAIR.An unscaled analytic model provides a theoretical estimateof the vertical range delay at a given time of day. This modelis then scaled by using a daily ALTAIR two-frequency TECmeasurement that adjusts the model estimate for the re-maining time in the tracking period.

FIGURE 10. Line-of-sight ionospheric range delay deter-mined by using the empirical elevation mapping function.This plot was created for ionospheric conditions with a ver-tical TEC value of 65.3, or a one-way VHF range delay of 1.0km. This function maps the vertical line-of-sight TEC orionospheric range-delay value to an arbitrary elevation. It isderived from historic multifrequency missile data collectedat KMR. The figure also illustrates the corresponding iono-spheric range delay at UHF and L-band.

Time of day (hours)

Ver

tical

TEC

(1016

ele

ctro

ns/m

2 )

Unscaledmodel

ALTAIR vertical TEC measurement

VH

F ra

nge

dela

y (m

)

42

35

21

14

7

28

0

600

500

300

200

100

400

6 12 18 6

Scaledmodel

VHFUHFL-band

Elevation (deg)

Ran

ge d

elay

(m)

90

3000

2500

1500

1000

500

2000

0 30 60



range, and range rate) are compared to precision satel-lite orbits on a daily basis to monitor and later adjustthe calibration parameters of the ALTAIR radar. Amodeling shortfall was evident, and this led to the de-velopment of the IECM.

Figure 12 illustrates the major components of theIECM. The system was integrated at ALTAIR in1998 and couples the use of a climatological iono-spheric model with real-time ionospheric data fromGPS and radar TEC measurements. The IECM con-sists of four major components: (1) the ionosphericcorrection algorithm, which provides range and eleva-tion-angle corrections to the radar tracking softwarefrom the look-up tables provided by the parameter-ized electron content model (the ionospheric correc-tion algorithm is the interface to the radar trackingsoftware for atmospheric correction); (2) the param-eterized electron content model (PECM), which pro-duces the radar look-up tables (tables of true rangeand elevation as a function of apparent range, appar-ent elevation, and azimuth and radar operating fre-quency); (3) the ionospheric specification managerthat is used to manage the communication of thelook-up tables to the ionospheric correction algo-rithm and radar tracking software; and (4) the GPS

and radar TEC data-reduction algorithms used to re-duce and provide the real-time ionospheric data tothe PECM.

The PECM was derived from a global theoreticalionospheric model developed for the U.S. Air Force.This global model, called the parameterized real-timeionospheric specification model (PRISM) was devel-oped by Anderson and R.E. Daniell et al. [18, 19].The PECM consists of a site-specific, theoretical cli-matological model (derived from PRISM) that pro-vides a database of the electron content as a functionof radar range, elevation angle, and azimuth, and adata-driven update algorithm. The update algorithmadjusts the database of electron content values tomatch actual conditions. The output of the PECMconsists of look-up tables in radar coordinates.

A TEC database specified for application at lowlatitude is used at ALTAIR. The database used by thePECM is a function of azimuth, elevation angle, andrange, relative to the ALTAIR radar site for a specificuniversal time. Each database electron content tablealso contains the peak electron-density magnitudeand the range to the peak as a function of azimuthand elevation angle. There are 540 database electroncontent tables covering three seasons (June solstice,December solstice, and equinox), three solar activitylevels (high, moderate, and low), and sixty universaltimes (every thirty minutes for most of the day andevery fifteen minutes near dusk and dawn). Each timethe PECM runs it reads the date, time, and solar ac-tivity index (10-cm solar flux index, F10.7). These val-ues are then used to select the particular electron con-tent table from the database. Then the PECM readsthe available GPS and radar TEC measurements andadjusts the selected electron content table to matchthe measured TEC values.

Once the particular database table is updated, alook-up table is produced for the radar by first com-puting the index of refraction as a function of altitudefrom the KMR mean tropospheric tables for the neu-tral atmosphere and updated TEC table for the ion-ized atmosphere, as shown in Figure 7 [14]. Next, atwo-dimensional ray trace is executed to produce twolook-up tables (one for each operating frequency atALTAIR) in radar coordinates. These look-up tablesare provided to the radar tracking software and are

FIGURE 11. Ionospheric elevation-angle bias computed byusing the empirical elevation mapping function at VHF, UHF,and L-band. This function is used to estimate the iono-spheric elevation-angle bias, given a particular vertical iono-spheric range-delay value. This plot was created for iono-spheric conditions with a vertical TEC value of 65.3, or aone-way VHF range delay of 1.0 km.

20 40 60 80

Elevation (deg)

Elev

atio

n-an

gle

bias

(deg

)

0

0.25

0.20

0.10

0.05

0

0.15

VHFUHFL-band



used in conjunction with the ionospheric correctionalgorithm. The PECM is run at two-minute intervalsto take advantage of the high-quality real-time TECdata used to measure actual ionospheric conditions.

Ionospheric Range Delay and TEC

The TEC measurement data described below are pro-vided to the PECM, which consists of update algo-rithms that use the data to adjust its TEC database tomatch real-time atmospheric conditions. The methodused to deduce line-of-sight ionospheric range delay(or equivalently, time delay) by using two-frequencymeasurements (also expressed in units of range orTEC) is applied in both the GPS and ALTAIR sys-tems. For GPS, the goal is to remove the atmosphericrefraction effects from GPS range measurements tobetter determine the position and velocity solutionfor terrestrial navigation purposes [20]. For ALTAIRthe goal is to determine the position and velocity of asatellite, minus the distorting effects of the atmo-sphere. Ideally, all satellites should be tracked by AL-TAIR at two frequencies to attain the highest attain-able range accuracy. At present, ALTAIR does nothave the capability to track satellites at two frequen-

cies beyond near-earth altitude (about five thousandkilometers). Therefore, the aforementioned atmo-spheric models are applied to correct all data collectedwith single-frequency radar measurements. This sec-tion describes the use of two-frequency range mea-surements to determine line-of-sight TEC and iono-spheric range-delay values for both GPS and thetwo-frequency ALTAIR system.

Because the group delay in the ionosphere issmaller than the speed of light, a time delay betweentwo points S1 and S2 is introduced by the ionosphereat radio frequencies. This delay (in mks units) can beapproximated by

δπ ε

Te

cm fN dS

cfN dS

ee

S

S

e

S

S

= =∫ ∫2

2 28

40 3

1

2

1

2.

, (2)

where c is the speed of light, me is the electron mass, eis the electron charge, f is the transmitted frequency,Ne is the electron density, ε is the permittivity con-stant, and S is the path length of the propagated sig-nal. In the ionosphere, phase delay is advanced by thissame amount. The ionospheric range delay (phase ad-vance) in meters is defined as

Radar lookup tables

Parameterizedelectron content model

(PECM)

Ionospheric correctionalgorithm

Date-time groupRadar operating frequencies

GPS TECRadar TEC

Radar trackingsoftware

Ionospheric model computer Radar tracking computer

Radar lookup tables loaded from the IECM

Radar TECGPS TEC

ClimatologicalTEC database

Two-frequencyGPS range data

IECM

Two-frequencyradar range data

FIGURE 12. Illustration of the major components of the IECM. These components include the climatologicalTEC database that provides a representative ionosphere, the real-time TEC measurement data that are used toadjust the TEC database to match actual ionospheric conditions, and the parameterized electron content model(PECM) that contains the update algorithms and the ionospheric error-correction algorithm. These PECM algo-rithms are used to compute the radar look-up tables, which are then provided to the ionospheric correction algo-rithm in the radar tracking software.



δ δR c T= , (3)

where this δR is the amount that would be added tothe range if the range were calculated by assuming theradar signal traveled at the speed of light. In the caseof a target with radar slant range of R, the TEC alongthe propagation path of the transmitted signal, inunits of 1016 electrons/m2, is

TEC = ∫ N S dSe

R

( ) ,0

(4)

where S1 = 0 and S2 = R. The integrated electron den-sity Ne (or TEC) is the total number of electrons in acolumn with height equal to the radar slant range Rand a cross-sectional area of one square meter [2, 4,21].

When a measurement is made at a second fre-quency, a second equation results, which allows us tosolve for TEC:

TEC = −−

R R f f

f f2 1 1

222

12

2240 3.

. (5)

The ionospheric range delay δR1 (in meters) at one ofthe frequencies is related to the differential range de-lay ∆R = R2 – R1 by

δR Rf

f f1

22

12

22

=−

∆ . (6)

In these equations f1 and f2 are the two measurementfrequencies, and R1 and R2 are the radar slant rangesmeasured at each frequency. The dispersive propertyof the ionosphere that causes these signals to refract ina frequency-dependent manner is used to computethe difference between the total time delay (group de-lay) of the two signals. The total time delay δT is pro-portional to TEC in units of 1016 electrons per unitarea (there are no other significant atmosphericsources of microwave frequency-dependent time de-lay). Once TEC has been calculated, as shown inEquation 5, it is used to estimate the range and eleva-tion-angle corrections needed to determine the truevalues of a single-frequency radar measurement madein the same vicinity.

The frequency-dependent range delay δR due tothe ionosphere and the corresponding TEC correc-tion for ALTAIR’s VHF and UHF operating frequen-cies follow from Equations 5 and 6 as

δ

δ

R R

R R

R

VHF

UHF

TEC

==

=

1 170

0 173

0 0764

. ,

. ,

. ,

∆∆∆

where ∆R = RVHF – RUHF in meters.Table 3 presents the one-way time delay and corre-

sponding range delay for one TEC unit at VHF andUHF. At low elevation angles, the TEC value can beas large as 333 × 1016 electrons/m2, which corre-sponds to range delays of 5100 m (VHF) and 680 m(UHF).

GPS Ionospheric Range Delay

GPS satellites transmit at two L-band frequencies(L1 = 1575.42 MHz and L2 = 1227.6 MHz). TheGPS receiver measures the total time delay at L1 andL2. Equations 2 through 6 can be used to express thedifferential time delay in seconds as

T T Tc

f f

f fL L L

L L

L L

TEC1 22

212

12

22

40 3− = = −

∆ .

. (7)

For a differential time delay δTL equal to one nano-second, the TEC correction is equal to 2.856 × 1016

electrons/m2. Therefore, there are 2.856 TEC unitsper nanosecond of differential time delay at the GPSsatellite operating frequencies. The integrated elec-tron density for GPS frequencies is expressed, in unitsof 1016 electrons/m2, as

TEC L= ×2 856. .∆T (8)

The measured GPS carrier phase data (at L1 andL2) are used to deduce a more precise value of TEC inreal time. At the start of the satellite track, a weightedaverage of the group and phase TEC are computed.After several minutes of averaging, the difference be-tween the average group TEC and the average phaseTEC (∆TEC) is initialized. This difference is thenadded to the subsequent phase TEC measurementsfor the rest of the tracking period, and the result isoutput to the IECM in real time. The phase TEC



provides a more precise measurement of the rate ofchange of TEC, but it is biased by an ambiguousnumber of carrier frequency wavelengths. The differ-ence value ∆TEC removes this bias. Once track hasbeen established on a particular satellite, the differen-tial carrier phase reference is set to zero. If loss of lock(track) is subsequently experienced, the differentialcarrier phase reference is reinitialized, along with the

FIGURE 13. Accurate real-time ionospheric measurement data are critical to the performance of the ionosphericmodels at Kwajalein. This figure shows the real-time reduction of GPS two-frequency data, as illustrated earlierin Figure 8. (a) GPS satellite differential-delay biases in TEC units, computed at the Jet Propulsion Laboratory[22]. (b) Differential-delay bias offset introduced by the GPS receiver, computed by using simultaneous two-fre-quency tracking data collected at ALTAIR. These values change slowly over time, relative to changes in the iono-sphere, and are added to (c) the GPS estimate of the total TEC. The ionosphere affects both the group velocityand phase velocity of the transmitted signals. The estimate of the line-of-sight TEC correction is attained bycombining the phase and group measurements collected by the GPS receiver.

averaging of group and phase TEC described earlier.Figure 13 illustrates representative GPS TEC datacollected at KMR. Each GPS satellite has an associ-ated bias expressed in TEC units. The biases usedhere were provided by the Jet Propulsion Laboratory[5]. The GPS receiver also has a bias that was esti-mated by comparing the GPS TEC data to the AL-TAIR TEC data.

GPS Receiver Installation and Data Quality

Many experiments were conducted during the periodfrom 1989 to 1997 to learn how to obtain TEC datawith sufficient accuracy to support real-time iono-spheric correction. This work was performed at theMillstone Hill radar in Westford, Massachusetts, andat the Kwajalein Missile Range, using several GPS re-ceiver models. The first ever real-time ionosphericmodel based on GPS data was built by Lincoln Labo-ratory in 1991 [23]. Over the years, several different

GPS TEC

Real-time TECoutput to IECM

Phase TEC

Time of day (hours)

TEC

Satellite bias

Satellite number

3

–3

2

1

0

–1

–2

0 5 10 15 20 25 30

300

200

100

0

–10011 14 17 20

(1016

ele

ctro

ns/m

2 )

Receiver bias

Elevation (deg)

26

13

0

–130 30 60 90

(a)

(b)

(c)

∆TEC

Averaging interval

Group TEC

Table 3. Time Delay and Range Delay for One TEC Unit

VHF (160 MHz) UHF (422 MHz)

δT 51.1 nsec 7.45 nsec

δR 15.35 m 2.26 m



GPS receivers and antennas were tested at the two fa-cilities as the technology continued to evolve. Thetype of antenna and placement of the antenna are im-portant to the reduction of multipath, a critical issuefor obtaining accurate real-time TEC data for use atlow elevation angles.

The type of GPS receivers evaluated used three dif-ferent signal processing schemes, which are referred toas precise code, semi-coded, and codeless tracking.While these processing schemes are not describedhere, it suffices to note that codeless tracking providesthe poorest signal-to-noise ratio, while the precise-code signal processing scheme supplies the highestsignal-to-noise ratio. In the end an Ashtech ZY-12 re-ceiver (precise code) was determined to provide high-quality TEC. Currently, this receiver is used for real-time operations at ALTAIR. This section summarizessome of the issues encountered in determining GPSdata quality, and what was learned by resolving theseissues.

The semi-coded GPS receiver provides a 13-dB in-crease in tracking sensitivity over a codeless receiver inthe presence of antispoofing (GPS phase-modulationencryption) by using a proprietary signal processingscheme named Z-mode [24]. With knowledge of thesignal encryption, the same semi-coded receiver wasable to operate with an additional 17 dB in trackingsensitivity. This additional sensitivity resulted in moreprecise measurements, especially at low elevation.

Before antispoofing was implemented, the L1 andL2 carrier phases were both modulated and transmit-ted in the clear with a seven-day (long) pseudo-ran-dom noise code (precise code). In the antispoofingmode of operation, the precise code is encrypted intowhat is called Y-code. In order to take advantage ofthe Y-code and gain back an overall loss of 30 dB intracking sensitivity, the GPS receiver must beequipped with secure hardware and firmware.ALTAIR’s first receiver was developed before anti-spoofing was initiated; it automatically reverted to acodeless signal processing scheme [25]. The codelesstechnique used by the original GPS receiver at AL-TAIR resulted in a 30-dB loss in tracking sensitivityover precise-code signal processing. Figure 14 con-tains representative data to show the increase in thequality of the TEC measurements with codeless,

FIGURE 14. TEC data collected at Kwajalein with codeless,semi-codeless, and precise-code tracking. The variability inthe TEC data below 20° drove the requirement to use pre-cise-code tracking for the acquisition of accurate TEC mea-surements. (a) TEC data, collected down to 20°, from thecodeless receiver were rendered unusable by the implemen-tation of antispoofing. (b) Semi-codeless tracking down to6° offers an improvement of 17 dB in tracking sensitivity butstill does not provide accurate TEC data at low elevationangle for real-time use below 20°. (c) Precise-code trackingprovides a total increase of 30 dB in tracking sensitivity, andhigh-quality TEC data down to very low elevation angles(4°), compared to codeless tracking. These data are assimi-lated by the IECM in real time and are essential to obtain ac-curate real-time TEC corrections for the radar system.

Group TEC

Minimum elevation is 20°(a)

TE

C (

1016

ele

ctro

ns/

m2 )

TE

C (

1016

ele

ctro

ns/

m2 )

TE

C (

1016

ele

ctro

ns/

m2 )

21.6 22.820.4–300

–150

0

150

300

Phase TEC

Group TEC

Minimum elevation is 6°(b)

Minimum elevation is 4°(c)

Phase TEC

14.4 16.8 19.212.0–100

0

100

200

300

Loss of track

Group TEC

Phase TEC

14.4 16.8 19.212.0–100

0

100

200

300

Time of day (hours)



semi-coded, and precise-code signal processing. Theaccuracy and precision of these TEC data are criticalto obtaining usable real-time ionospheric correctionswith the IECM, since these data are assimilated inreal time.

The original location for the GPS receiver antennainstallation on Roi-Namur was subject to signalmultipath from nearby interference sources, as shownin Figure 15. To determine a suitable location for thereceiver antenna, TEC data were collected and exam-ined at two other locations on Roi-Namur. The newlocation for the GPS receiver and antenna was chosenon the basis of a compromise between increased dataquality (less multipath) and adequate infrastructureto house the receiver in a temperature-controlled en-vironment and provide communications to send theTEC data to ALTAIR. The new GPS installation wason top of a building, as shown in Figure 16, with theantenna placed at a height above and distant fromsurrounding obstructions, about three hundredmeters from the ALTAIR antenna.

Results

Operational use has shown that the IECM combinedwith GPS and radar TEC data significantly improves

the range-accuracy performance over the first-orderionospheric model while tracking calibration satel-lites. In great part, this improvement is due to the cli-matology represented in its TEC database. One of thecritical features of the IECM model is that it incorpo-rates the physics of the equatorial region describedearlier. In particular, the Appelton anomaly is repre-sented in the model. The first-order ionosphericmodel originally used at ALTAIR underestimated theazimuthal variation in the TEC distribution. By com-bining the real-time data from the precise-code GPSreceiver, which was installed at ALTAIR in 1998 (andprovides up to twelve TEC measurements every tenseconds), and the more complex ionospheric model(the IECM), ALTAIR has a far more accurate repre-sentation of the local ionosphere. A precise-code GPSreceiver, which has been used to obtain TEC mea-surements at ALTAIR since 1998, provides a set of upto twelve TEC measurements every ten seconds.

Figure 17 illustrates ALTAIR measurement residu-als, on precision reference orbits, collected during1998 while using both the IECM model and theolder first-order ALTAIR model. These data illustratea significant improvement in the mean range error,

FIGURE 15. GPS TEC data collected near the ALTAIR an-tenna and at an alternate location where few interferencesources were present. These data were collected by using asemi-coded receiver on a GPS satellite, which provided theworse-case multipath data and resulted in a deterministiccharacterization of multipath sources. The relocation of theGPS antenna to an alternate location eliminated nearly allmultipath effects. This GPS receiver was later replaced witha precise-code receiver, further improving the quality of theGPS data by increasing the tracking sensitivity an additional17 dB.

FIGURE 16. A GPS receiver antenna installed near the Kwaj-alein Atoll lagoon. The ALTAIR antenna shown in the back-ground is approximately three hundred meters from the GPSantenna. The position of this GPS antenna installation,which is well away from and above potential scatterers,eliminated multipath effects from corrupting the TEC cor-rection data at low elevation angles. The yellow boom on theright side of the photo is part of the crane used to lift thephotographer to obtain this photograph, and is not a perma-nent structure that could act as an interference source.

Group TEC

Time of day (hours GMT)126

50

75

100

125

Multipath

TEC

(1016

ele

ctro

ns/m

2 )



while an improvement in the elevation-angle mea-surements was not yet realized. This lack of improve-ment in elevation-angle measurements is possibly dueto model error, since the large elevation-angle errorsare correlated with the increase in solar activity. Asimilar system for mid-latitude ionospheric modelingbased on GPS was developed at the Millstone HillRadar in the early 1990s [23].

Future Work

The Kwajalein atmospheric models have been underdevelopment since the 1960s, and further study is re-quired to meet KMR’s increasing accuracy require-ments and national scientific objectives.

In particular, the KMR neutral atmospheric modelsupplies a mean estimate for the effects of the atmo-sphere below ninety kilometers in altitude. While themodel is useful for real-time application, significantimprovements have been accomplished since the time

of its implementation. Newer models for estimatingtropospheric refraction were developed in the 1980sand 1990s as a consequence of extended research invery long baseline interferometry, satellite laser rang-ing systems, GPS, and other space-based electromag-netic ranging techniques. The majority of these newmodels separate out the neutral atmospheric delayinto two components: a hydrostatic (“dry”) compo-nent, and a non-hydrostatic (“wet”) component dueto the water vapor in the atmosphere. An excellentreview of these models can be found elsewhere [26].The next logical area of research and application in-volves the incorporation of one of these models intothe IECM, for example, the atmospheric model de-veloped for the FAA’s Wide Area Augmentation Sys-tem [27].

At the time IECM was developed, the capacity ofthe available computers forced compromises in thesize of the climatological TEC database that repre-

FIGURE 17. Comparison of metric accuracy with the older first-order ALTAIR model and theIECM during 1998. These residual data, which were collected on calibration satellites, resultfrom weekly numerical orbit solutions created with highly accurate laser-ranging data col-lected by NASA. The tracking data acquired by using the first-order ALTAIR model span ap-proximately 250 days, while the tracking data acquired by using the IECM span nearly sixtydays. ALTAIR attained a significant improvement in range accuracy with the IECM, althoughsimilar improvements were not apparent in elevation-angle accuracy.

Mean –1.3Standard dev. 16.8

IECM300

200

100

–100

–200

–300

0

Mean 4.9Standard dev. 12.0

60

40

20

–20

–40

–60

0

250 260 270 280

280

290 300

Mean 18.0Standard dev. 102.9

First-order ALTAIR model300

200

100

–100

–200

–300

0

0

Mean –5.6Standard dev. 10.5

60

40

20

–20

–40

–60

0

0

50 100 150 200

50 100 150 200 250

Day of year (GMT 1998)

Mea

n re

sidu

al e

rror

(sin

gle

freq

uenc

y U

HF)

250 260 270 290 300

Elev

atio

n (m

deg)

Ran

ge (m

)

250

Day of year (GMT 1998)



sented the ionosphere, thereby affecting model reso-lution. At present, these tables are implemented as afunction of season and solar activity; there is a charac-terization of three seasons and solar activity levels. Abetter characterization of the representative iono-sphere would result from the creation and use of TECmodels for each month and solar activity period. Thisexpansion would vastly increase the number of tablesand increase the granularity of the representativeionosphere by merely taking advantage of the greatercapabilities in present computer technology.

The IECM alone does not provide an accurate rep-resentation of the actual ionospheric conditions, butwhen the IECM is combined with real-time GPSTEC measurements, it provides a significant im-provement over the first-order ALTAIR model. WhileGPS TEC measurements are significant, on their ownthey provide little or no real-time information on theionospheric distribution and structure, which is espe-cially important in the equatorial region. To obtaininformation on the ionospheric profile, measure-ments such as those obtained from a digital high-fre-quency ionospheric sounder (ionosonde) or incoher-ent scatter are needed. Another approach wouldinvolve the implementation of tomography receiversand software to reconstruct the ionosphere throughtomographic imaging techniques combined withTEC measurements [28]. The best approach wouldbe to combine many different systems of measure-ment collection and then provide the data to IECM.

Summary

Background on the equatorial ionosphere has beensummarized to provide support for the emphasis ofcurrent research and model application. The proper-ties of the equatorial ionosphere drive the need for arepresentative model to accompany real-time mea-surement data when attempts are made to model thestate of the ionosphere at any given time. This needfor a model is due to a limited amount of measure-ment data relative to the complex structure of theequatorial ionosphere. To rectify this shortfall, theIECM—a climatological model that provides a repre-sentative ionosphere—has been in operational use atKMR since 1998.

The collection of real-time ionospheric measure-

ments to adjust the IECM to match actual conditionshas proven to be the best technique to resolve modelaccuracy issues. The improvement in radar range ac-curacy brought about by the use of IECM combinedwith real-time GPS TEC is clearly illustrated. Furtherimprovement in accuracy would be realized if addi-tional real-time ionospheric measurement data (suchas that provided from an on-site ionosonde) could beprovided to the IECM. Proving the accuracy of theionospheric models relative to a fundamental truth isa strength that the KMR radar systems bring to bearon the problem. Their precise calibration and accessto “ground truth” using accurate calibration satelliteorbits is the basis for the error analysis presented here.

Acknowledgments

Art Lewis of Raytheon has contributed to the design,development, and integration of the KMR models forover twenty years. Mike Austin of the Field Systemsgroup at Lincoln Laboratory submitted the requestfor and managed the Small Business Innovative Re-search contract that led to the IECM. Greg Hogan ofthe Field Systems group contributed to the contentsof this document. George Millman and Jeff Lamicelaof Research Associates of Syracuse, Inc., Syracuse,New York, and Rob Daniell and Lincoln Brown ofComputational Physics, Inc., in Newton, Massachu-setts, developed the IECM for its application at AL-TAIR under the Small Business Innovative Researchcontract. Lorraine Thornton of the Surveillance Tech-niques group at Lincoln Laboratory provided the pre-cise calibration data used to evaluate the accuracy ofthe IECM. Dale Sponseller from Raytheon atKwajalein provided technical support in the integra-tion of the GPS receiver equipment at ALTAIR. An-thony Manucci, Brian Wilson, and Garth Franklin ofthe Jet Propulsion Laboratory provided guidance inthe installation of the GPS receiver equipment andsupplied the GPS satellite calibration biases appliedin the IECM.



R E F E R E N C E S1. MIT Haystack Observatory Web Site, <http://www.haystack

.edu/ysp/atmosphere/earth.html>, June 1, 1999.2. J.W. Chamberlain and D.M. Hunten, Theory of Planetary At-

mospheres: An Introduction to Their Physics and Chemistry (Aca-demic Press, Orlando, Fla., 1987).

3. T.I. Gambosi, Physics of the Space Environment, Cambridge At-mospheric and Space Science Series (Cambridge UniversityPress, Cambridge, U.K., 1998), pp. 177–189.

4. A. Jursa, ed., Handbook of Geophysics and the Space Environ-ment (Air Force Geophysics Laboratory, Air Force SystemsCommand, U.S. Air Force, 1985).

5. M.C. Kelley, The Earth’s Ionosphere: Plasma Physics and Electro-dynamics (Academic Press, San Diego, 1989).

6. R.J. Moffet, “The Equatorial Anomaly in the Electron Distri-bution of the Terrestrial F-Region,” Fundam. Cosmic Phys. 4,1979, pp. 313–391.

7. M. Mendillo, J. Baumgardner, X. Pi, P.J. Sultan and R.Tsunoda, “Onset Conditions for Equatorial Spread F,” J.Geophys. Res. 97 (A9), 1992, pp. 13,865–13,876.

8. R.T. Tsunoda, “Time Evolution and Dynamics of EquatorialBackscatter Plumes. 1. Growth Phase,” J. Geophys. Res. 86(A1), 1981, pp. 139–149.

9. R.T. Tsunoda, “Magnetic-Field-Aligned Characteristics ofPlasma Bubbles in the Nighttime Equatorial Ionosphere,” J.Atmos. Terr. Phys. 42 (8), 1980, pp. 743–752.

10. J.K. Hargreaves, The Solar-Terrestrial Environment: An Intro-duction to Geospace, Cambridge Atmospheric and SpaceScience Series (Cambridge University Press, Cambridge, U.K.,1992).

11. W.E. Swartz and R.F. Woodman, “Same Night Observationsof Spread-F by the Jicamarca Radio Observatory in Peru andCUPRI in Alcantara, Brazil,” Geophys. Res. Lett. 25 (1), 1998,pp. 17–20.

12. R. McSheehy, private communication, Group 33, LincolnLaboratory, 1992.

13. J. Klobuchar and R. Allen, “A First Order, Worldwide, Iono-spheric Time Delay Algorithm,” AFCRL-TR-75-0502,AFCRL(LI), Hanscom AFB (25 Sept 1975), DTIC #AD-A018862.

14. R.E. Daniell, Jr., L.D. Brown, D.N. Anderson, M.W. Fox,P.H. Doherty, D.T. Decker, J.J. Sojka, and R.W. Schunk,“Parametrical Ionospheric Model: A Global ParameterizationBased on First Principles Models,” Radio Sci. 30 (5), 1995,pp. 1499–1510.

15. G. Millman, Research Associates of Syracuse, Inc., Syracuse,New York, private communication, June 1998.

16. D.N. Anderson, M. Mendillo, and B. Herniter, “A Semi-Empirical Low-Latitude Ionospheric Model,” Radio Sci. 22(2), 1987, pp. 292–306.

17. D.L. Gallagher, P.D. Craven, and R.H. Comfort, “An Em-pirical Model of the Earth’s Plasmasphere,” Adv. Space Res. 8,1988, pp. (8)15–(8)24.

18. D.N. Anderson, “The Development of Global, Semi-Empiri-cal Ionospheric Specification Models,” Proc. Ionospheric EffectsSymp., Alexandria, Va., 4–6 May 1993, pp. 4B/3/1–3.

19. R.E. Daniell, Jr., W.G. Whartenby, and D.N. Anderson,“PRISM Validation,” Proc. Ionospheric Effects Symp., Alexan-dria, Va., 4–6 May 1993, pp. 4B/2/1–11.

20. J.J. Spilker, Jr., “GPS Signal Structure and Performance Char-acteristics,” Navig. J. Inst. Navig. 25 (2), 1978, pp. 121–146.

21. D. Coco, “GPS—Satellites of Opportunity for IonosphericModeling,” GPS World 2 (10), 1991, pp. 47–50.

22. A. Mannuci and B. Wilson, Jet Propulsion Laboratory, privatecommunication, June 1997.

23. A.J. Coster, E.M. Gaposchkin, and L.E. Thornton, “Real-Time Ionospheric Monitoring System Using GPS,” Navig. J.Inst. Navig. 39 (2), 1992, pp. 191–204.

24. Ashtech Z-12 GPS Operating Manual, Ashtech DocumentNumber 600224, May 1994.

25. Technical Manual for the Model ICS-4Z Mini-Rogue, AllenOsborne Assoc., June 1992.

26. V. de Brito Mendes, “Modeling the Neutral-AtmosphericDelay in Radiometric Space Techniques,” Technical ReportNo. 199, University of New Brunswick, Geodesy andGeomatics Engineering, Apr. 1999.

27. P. Collins and R.B. Langley, “Tropospheric Delay—Predictionfor the WAAS User,” GPS World 10 (7), 1999, pp. 52–58.

28. G. Bust, T.L. Gaussiran II, and D.S. Coco, “Ionospheric Ob-servations of the November 1993 Storm,” J. Geophys. Res. 102(A7), 1997, pp. 14,293–14,304.



is an associate staff member atthe Kwajalein Missile Rangefield site, where her primaryduty is in space surveillance atALTAIR, including orbitalanalysis for operational space-launch tracking support,routine spacecraft monitoring,and unplanned space-surveil-lance events. Her interests alsoinclude ionospheric modelingand analysis, as well as meteorresearch. Previously sheworked for the AdvancedElectromagnetic Systemsgroup at Lincoln Laboratoryin the area of radar phenom-enology. She has a B.S. degreein physics and astronomy fromthe University of Rochester,and an M.S. degree in physicsfrom the University of Texas atAustin, where her graduatethesis involved the study ofgravity waves in plasmas.

. is an associate staff member inthe Field Systems group. He isthe system analyst of theKwajalein Space SurveillanceCenter, which is used to con-duct the remote operation andmeasurement-data reductionfor the radar systems atKwajalein. He received a B.S.degree in physics fromWorcester Polytechnic Insti-tute and is currently pursuinga graduate degree in spacephysics at Boston University.Prior to his current position heworked in the Space Surveil-lance Techniques group, in-cluding several years at theKwajalein Missile Range. AtKwajalein he was a space-surveillance analyst, and heparticipated in a collaborativeeffort to characterize meteorsby using the ARPA LongRange Tracking and Instru-mentation Radar (ALTAIR).He was also responsible for thedevelopment and integrationof a real-time ionosphericmodeling system for ALTAIR.

. is a staff member in the Sur-veillance Techniques group.She received a B.A. degree inmathematics from the Univer-sity of Texas at Austin, andM.S. and Ph.D. degrees inspace physics and astronomyfrom Rice University, whereshe was involved with iono-spheric heating experiments atthe Arecibo Observatory inPuerto Rico. At Lincoln Labo-ratory she has been responsiblefor the development of soft-ware and hardware used tocompensate for atmosphericrefraction at the Millstone Hillradar. Together with her co-workers, she developed thefirst real-time ionosphericmonitoring system based onGPS. This system is now anintegral part of the MIT radarcalibration system. In 1995,she was awarded an AdvancedConcepts Program grant toconduct the Westford WaterVapor experiment, whichverified that GPS could beused to obtain accurate, near-real-time information on theamount of water vapor presentin the atmosphere. She is amember of the InternationalUnion of Radio Science(URSI), the Institute of Navi-gation (ION), and the Ameri-can Geophysical Union. She isthe current vice-chair of com-mission G (ionosphere), U.S.chapter of URSI, a member ofthe GPS Meteorology Inter-agency Working Group, and arecent technical chair for theSeptember 2000 ION-GPSmeeting.



. is a systems engineer forRaytheon Range SystemsEngineering at KwajaleinMissile Range. Her primaryduty is systems support at theALTAIR radar, includingmission planning and director-ship, radar upgrade analysisand testing, and engineeringsupport for the radar iono-spheric correction model. Herprimary research interestsinclude radar signal process-ing, space surveillance, andradar ionospheric correction.Previously she was at theGeorgia Tech Research Insti-tute (GTRI), working in areasof electronic warfare modelingand analysis, and teachingsessions of a radar course.Previous to GTRI she was atthe Westinghouse DefenseCenter, working in electronicwarfare, advanced systemsdevelopment, and system testsupport. She received a B.S. inelectrical engineering fromValparaiso University, and anM.S. in electrical engineeringfrom the Georgia Institute ofTechnology.

is a software engineer withRaytheon Corporation at theKwajalein Missile Range.Previously he was a LincolnLaboratory contractor for nineyears, working with the LaserRadar group and the AirTraffic Automation group. Hereceived B.S.E.E. andM.S.E.E. degrees from North-eastern University.

is an orbital analyst in theField Systems group atKwajalein Missile Range. Heattended the University ofNew Hampshire, where hemajored in mechanical engi-neering before switching tocomputer science. His work atKwajalein has focused onadding new capabilities to theALTAIR deep-space trackingsystem, including ionosphericcorrection. Prior to coming toALTAIR he was leader of thesoftware section of the Aero-space Engineering group. Herecently retired after thirty-twoyears at Lincoln Laboratory.

Documents

Equatorial Atmospheric and Ionospheric Modeling at ... · Kwajalein Atoll is located in the Marshall Islands in the Pacific Ocean, approximately four thousand ki- ... period at ALTAIR’s