A COMPARATIVE STUDY OF THE HARRIS-PRIESTER, JACCHIA-ROBERTS, AND MSIS ATMOSPHERIC DENSITY MODELS IN THE CONTEXT OF SATELLITE ORBIT DETERMINATION*

R. E. Shanklin, Jr., T. Lee, M. K. Mallick, R. A. Kuseski, and J. O. Cappellari, Jr. Computer Sciences Corporation ABSTRACT Extensive comparlsons of the Harris-Priester, JacchiaRoberts, and MSIS (Mass Spectrometer/Incoherent Scatter) atmospheric density models as used in satellite orbit determination are summarized. The quantities compared include Bayesian weighted least squares differential correction statistics and orbit solution consistency and accuracy.

*This work was supported by the Operations Analysis Section, Operational Orbit Support Branch, Goddard Space Flight Center, National Aeronautics and Space Administration, under Contract NAS 5-24300.

i-i

SECTION

i -

INTRODUCTION

Atmospheric ellite eters. orbits The

drag with

is a significant perigee of by heights

perturbation of less than

of

Earth

sat-

i000 due

kilomto

acceleration drag is given

a spherical the equation

satellite

atmospheric

Adrag

im 2
density velocity reference drag mass of at the position to the of the satel-

where

D = atmospheric lite V = satellite

relative

atmosphere area

A = satellite C D = satellite m = satellite Therefore, knowledge tion This three context are the and calculation of the

cross-sectional

coefficient

the

drag density

acceleration as

requires of posi-

atmospheric

a function

time. presents the results atmospheric of a comparative density The model, three the models models study of

paper

different of orbit

global

in the compared

determination. (H-P) Mass

Harris-Priester and the

Jacchia-Roberts Scatter

(J-R) (MSIS) The ture

model, model.

Spectrometer!Incoherent

Harris-Priester profile

model of

is based the heat

on

theoretical

temperaunder two and

solutions equilibrium solar heat from

conduction The model (EUV) the

equation assumes heating

hydrostatic heat an sources:

conditions. extreme

ultraviolet produces

artificial deduced

source satellite model

that

diurnal In

variathe mod-

tion ified

drag used

calculations. for this study,

Harrls-Priester

the

EUV

1-2

heating

level

is selected profile flux, and

by

choosing

among

i0

different

altitude-density levels by of solar

tables the

representing

i0 different is modeled

diurnal

variation of

a correction

calculated 2). model an upper

using

a power

a cosine

(References The

1 and

Jacchia-Roberts scaled Analytic of by

is based boundary

on

empirical exospheric

temperature temperature through in-

profiles (T).

density

calculation

is accomplished The modeling

integration cludes

thermodynamic for EUV

equations. heating, solar

corrections

particle

flux sea2

(so-called

geomagnetic) and

heating, the diurnal

semiannual variation

variations, (References

sonal variations, and 3). The MSIS model

is based

on

fitting the

spherical

surface

har-

monic

expansions

to match and

angular

dependence scatter that

exhibited

by mass The ing, nual MSIS

spectrometer formulation particle

incoherent

measurements. model EUV heatsemian-

includes flux

sections annual

solar

heating, variations,

variations,

variations,

diurnal

semidiurnal from

variations, equi-

terdiurnal librium. GTDS load

variations, MSIS modeling Dr.

and has Hedin

departures been and who and

diffusive

implemented his

in a special at model the {Refsub-

module. Flight

associates the their

Goddard erence routines Table pheric

Space 4),

Center, advice GTDS

developed some of

contributed the sample with activity MSIS

program

during

implementation. profiles solar for EUV the three atmosone

1 shows models

density two

different

levels the

and

geomagnetic Roberts density, imum table, and as

level.

Figure relative altitude. but, are as

1 shows to The the

Jacchia-

densities, of 2.0

Harris-Priester shows from max-

a function as three high as

figure

ratios the

is apparent similar

the shape.

profiles

quite

in overall

1-3

TABLE

1. ATMOSPHERIC DENSITIES COMPUTED USING HARRIS_RIESTER, JACCHIA-ROBERTS, AND MSlS MODELS

DENSITY ALTITUDE (kin) F10.7 = 125.0 F10.7 = 150.0 HARRISPRIESTER

(kg/km 3) MSIS F10.7 = 116.2 F10.7 = 135.1 .203 E + 1 ;274E 0 F10.7 = 140.0 F10.7 = 165.3 .204 E + 1 .313E 0

JACCHIA-ROBERTS F10.7 = 116.2 -F10.7 = 135.1 .193 E + 1 .228E 0 F10.7 = 140.0 F10.7 = 165.3 .210 E + 1 .270E 0

150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000

.205 E + 1 .224E 0

.206 E + 1 .255E 0

.459 E - 1 .129 E - 1 A25 E -- 2 .155 E -2 .521 E -- 3 .218 E -- 3 .963 E - 4 .451 E - 4 .227 E - 4 .112E-4 .691 E - 5 .464 E - 5 .316 E -- 5 .245 E 5

.583 E -- 1 .178 E -- 1 .631 E -- 2 .247 E -- 2 .879 E -- 3 .392 E -- 3 .182 E -- 3 .851 E -- 4 .451 E -- 4 .217 E--4 .127 E --4 .804 E --5 .462 E -- 5 .301 E -- 5 .201 E-- 5 .141 E--5

.559 E -- 1 .177 E -- 1 .637 E - 2 .246 E -- 2 .835 E -- 3 .353 E -- 3 .155 E - 3 .706 E - 4 .339 E - 4 .154E--4 .878 E - 5 .548 E - 5 .348 E - 5 .258 E -- 5 .201 E - 5 .155E--5

.721 E - 1 .249 E - 1 .977 E - 2 .4i3 E -- 2 .157 E -- 2 .724 E - 3 .344 E3

.636 E -- 1 .187 E1

.802 E - 1 .255 E -- 1 .926 E -- 2 .368 E - 2 .131 E - 2 .582 E - 3 .266 E - 3 .125 E -- 3 .600 E -- 4 .259E-4 .134 E -4 .728 E -5 .378 E - 5 .236 E - 5 .158 E - 5 .107 E-5
co

.633 E -- 2 .236 E -- 2 .780 E -- 3 .324 E - 3 .139 E 3

.169 E - 3 .851 E - 4 .394E-4 .219 E -4 .128 E -4 .737 E - 5 .500 E - 5 .361 E - 5 .262E-5

.619 E --4 .285 E -- 4 .120E-4 .623 E -5 .352 E -5 .200 E - 5 .137 E 5

.198 E-- 5 .163E 5

.102 E - 5 .761E--6

NOTES:

1. Kp = 3.3 FOR JACCHIA-ROBERTS

DENSITY AND Ap = 33 FOR MSIS DENSITY

ARE USED. OF 205 ,

2. THESE PROFILES ARE FOR AUGUST 30, 1978, AT A LATITUDE AND A LOCAL SOLAR TIME OF 1:40 P.M.

OF 46 N, AN EAST LONGITUDE

2.0

F10.7 _, 120 1.8-

1.6-

X _ X X X X
"_
CL

1.41.2 X

"_ ; r_

X
X
J--R

o_

_. 1.0
Z ,,, 0.8

X X

=-

X X X X
MSlS X

0.6
0.4

0.2

I 200

1 400 ALTITUDE

I 600 (kilometer=)

I 800

/ 1000

2.0 F10.7 _- 145 1.8

J-R

1.6

"_ 1.4-

X X X X X X

_
0

_:

x 1_
X

x
X
X

1.2

< cc >I.LM

1.0

X 0.8
0.6

x x

MSIS

0.40.2 I 200

q_ O

4_0 ALTITUDE

6{_]0 (kilometers)

800 L

10_0 ==

FIGURE

1. ATMOSPHERIC

MODEL DENSITY

RATIOS

1-5

SECTION 2 - COMPARATIVE STUDY STRUCTURE All the results presented in Section 3 of this paper are based on Goddard Tra3ectory Determination System (GTDS) Bayeslan weighted least squares differential correction solutions. Nine different series of six GTDS Differential Correction (DC) Program runs were made for each of the three atmospheric models. Three different satellites, with perigee heights between 310 and 560 kilometers, were studied; other orbital parameters for these satellites are given in Table 2. The nine series of orbit determination arcs are listed in Table 3. Each series contains six 30-hour-arc solutions. The solu-

tions are used to generate 30-hour ephemerides that overlap adjacent ephemerides by 6 hours. The ephemerides are then compared in order to determine the maximum position differences (in the orbital reference frame) during the overlap periods. The 162 DC Program solutions produce 135 maximum overlap position differences. These differences are used to evaluate the consistency and accuracy obtained when each of the three atmospheric models isused. Each differential correction solution is made up of seven numbers: three position coordinates, three velocity coor-

dinates, and the drag variation parameter (pl), which is a scaling factor in the drag acceleration equation, i.e.,

-_drag -

i CD A pll + 01) 2 m

--_ V IVl -_

This scaling factor is applied during generation of the ephemeris that uses the differential correction solution.

1-6

TABLE 2. SATELLITE ORBITAL ELEMENTS

SATELLITE

DATE

PERIGEE HEIGHT (kilometers)

APOGEE HEIGHT (kilometers)

INCLINAT)ON (degrees)

AE-3 AUGUST 1, 978 1 _1 MAGSAT OCTOBER,1,1979 352 MARCH 1, 9_0 1 _2_

SAGE FEBRUARY 19, 1979 560

341 581 471

655 "

97 97
55
o3

TABLE 3. COMPARATIVE STUDY SERIES

SERIES NUMBER 1 2 3

SATELLITE

TIMESPAN AUGUST 1-6, 1978

AE-3

AUGUST 14--19, 1978 SEPTEMBER2--8, 1978

4 5 6 7 8 MAGSAT

OCTOBER31--NOVEMBER 5, 1979 DECEMBER1--6, 1979 JANUARY 1--6, 1980 FEBRUARY 1-6_ 1980 MARCH 1--6, 1980

SAGE

FEBRUARY 19--25, 1979

co

.].-7

Spacecraft model (e.g.,

attitude

is not

considered, no

since aerodynamic

a spherical forces spherical and it is

is employed. lift) other

Furthermore, than drag for are

modeled.

The

approximation possible However, have that it

is crude other

all

three

satellites, are

aerodynamic to on

forces expect

nonnegligible. assumptions study, of the ob-

is reasonable effect are models Simply the

that

both of

a negligible the results

the

results by same

this

because three

obtained to the

applying arcs

each the

atmospheric sets.

with

same

servation should models

stated, for

unmodeled all three

aerodynamic atmospheric

forces

perturb

solutions manner.

in a similar

1-8

SECTION

3 - COMPARATIVE

STUDY

RESULTS

This study

section of

summarizes

the

results models

of In

this the

comparative context of run-

atmospheric (30-hour)

density orbit of

short-arc by-run erence First, ison merits cialized average Second, parison ferences residuals. such The large average of

determination. results are be

A detailed,

presentation 5. Two

these

is available appropriate. as the highly

in Ref-

cautionary should models; must

remarks not

these

results

interpreted about to this

a comparrelative spean

atmospheric of the models

conclusions be limited orbit

context--short-arc drag any runs and scaling series may of factor orbit

determination for. and large

in which

is solved

determination sporadic

ephemeris overlap large

comdifRMS

contain

a few

a few Some of

differential the runs and/or RMSs three and AE-3

corrections included high the in

with this

study

show

differences weighted for the

RMSs. average are maximum glven position in Table along and the 4.

differences The with solar the cent and averages the

series are

over of

all the

three EUV

series heating (Kp).

also

given,

ranges

index The

(FI0.7)

particle

flux

index overlap

averages are

show about

that 11.5 per-

Jacchia-Roberts (24 meters) that the MSIS the

differences the

smaller averages

than are

Harris-Priester 19 percent The

averages (38 meters) 62-meter difcannot

about

larger

than

Harris-Priester

averages.

ference between the Jacchia-Roberts and MSIS be considered either large or significant. The study that are same information five is given series of for Magsat The

averages

in Table Magsat

5. results

This show

includes both the

arcs. and

Jacchia-Roberts larger

MSIS

average

differences average

about

9 percent

than

the

Harris-Priester

1-9

TABLE 4. COMPARATIVE ATMOSPHERIC DENSITY MODEL STUDY FOR AE-3 (AUGUST AND SEPTEMBER 1978)

RESULTS

HARRIS-PRI SERIES I I'_ i._= C) AUGUST AUGUST 1-6 14-19 (1 RANGE OF F10'7 ) RANGI_ OF Kp 0-6 0-6 0-6 AVERAGE WEIGHTED RMS 4.9 7.3 7.3

ESTER MODEL MAXIMUM DIFFERENCE POSITION (meters) 191 225 209

JACCHIA-ROBERTS AVERAGE WEIGHTED RMS 5.2 7.8 8.4

MODEL

MSIS MODEL AVERAGE WEIGHTED RMS 8.4 8.5 7.2 MAXIMUM DIFFERENCE POSITION (meters) 265 324 164

MAXIMUM DIFFERENCE POSITION (meters) 175 217 163

0 -22 watt/(m 2 -Hz)

106.0-117.6 115.6-134.9 159.8-181.1

SEPTEMBER 2-8

AVERAGES

6.5

208

7.2

184

8.0

251

co

TABLE 5. COMPARATIVE ATMOSPHERIC (NOVEMBER AND DECEMBER

DENSITY MODEL STUDY RESULTS FOR MAGSAT 1979; JANUARY, FEBRUARY, AND MARCH 1980)

RANGE PE HIO(_) OF F10.7 VARIATION (t0 " 22 watts m 2 Hz 1) RANGE OF Kp VARIATION

HARRIS

PRIESTER

RESULTS MAXIMUM

JACCHIA-

ROBEHTS

RESULTS MAXIMUM

MSIS RESULTS MAXIMUM WEIGHTED RMS POSITION DIFFERENCE (m_=urs)

WEIGHTED RMS

POSITION DIFFERENCE (meters)

WEIGHTED" RMS

POSITION DIFFERENCE (meters)

t-'
I OCT. 31 NOV. 5, 1979 207.5-214.9 0.- 4 8.3 204 7.8 1 16 8.0 190

DEC. 1 -6,

1979

152.2-

223.4

0 -4

12.4

204

11.5

175

12.8

25b

JAN. 1

6, 19_0

181:1.9 212.4

1--.5

9.4

213

9.5

166

11.3

28S

FEB. 1 - 6, 1980

212,6 _ 231.7

0-4

12.7

326

12.5

298

13.8

313

MAR.

E, 1980

170.2

-176.7

O. 3

9.8

.161

13.4

396

10.0

159

AVEHAGES

"

30.6

222

10.9

242

11.2

243

"=

differences. demonstrate comparable The average of

As that in the

zn the the

case

of

AE-3,

the

Magsat

results models are

three

atmospheric of this study.

density

context and

RMSs SAGE

overlap are given agree

position in Table to within

differences 6. Both

for the

the and

serzes the

arcs

RMSs all

overlap

differences models

3 percent; equivalent

three

atmospheric

produce

essentially

errors.

TABLE 6. COMPARATIVE ATMOSPHERIC FOR SAGE (FEBRUARY 19-25,

DENSITY 1979)

MODEL STUDY

RESULTS

ATMOSPHERIC DENSITY MODEL USED

AVERAGE WEIGHTED RMS

AVERAGE MAXIMUM POSITION DIFFERENCE (meters}

HARR IS--PR I ESTER

10.9

108

JACCHIA--ROBERTS

11.2

! 14

MSIS

11.0

112

g
o0

NOTE: DURING THIS PERIOD, F10.7 VARIED FROM196.0 TO

237.7 X 10-22WATTS METER -2 VARIED FROM 1 TO 7. HERTZ -1 AND Kp

1-12

SECTION 4 - CONCLUSION
The results for presented satellites in above this 300 and paper support the the conclusion Harrisdensity and es-

that,

kilometers, MSIS

Priester, models all

Jacchia-Roberts, produce comparable roughly orbit

atmospheric density

similar

profiles results

sentially drag

determination is solved for

when

the is

variation by

parameter arc of

and

orbit It will orbit

quality is

measured sible best tion pheric to fit arc.

adjacent which

overlap the

comparisons. models given arcs, solution

imposthe

predict or best

three for any

produce

predictions for result some

determinaatmos-

However, may

problem

switching

models

in marked

improvements.

1-13

REFERENCES

i.

I. Harris and W. Priester, "Atmospheric Structure and Its Variations in the Region From 120 to 800 Km," COSPAR International Reference Atmosphere (CIRA) 1965, Space Research IV, North Holland Publishing Co., Amsterdam Goddard Space Flight Center, X-582-76-77, Mathematical Theory of the Goddard Trajectory Determination System, J. O. Cappellari, Jr., C. E. Velez, and A. J. Fuchs (editors), April 1976 L. G. Jacchia, Revised Static Models of the Thermosphere and Exosphere With Empirical Temperature Profiles, Smithsonian Astrophysical Observatory Special Report No. 332, Cambridge, Massachusetts, May 1971 A. E. Hedin et al., "A Global Tropospheric Model Based on Mass Spectrometer and Incoherent Scatter Data: MSIS i. N 2 Density and Temperature," Journal of Geophysical Research, 1977, vol. 82, pp. 2139-2147 Computer parative Context T. Lee, Sciences Corporation, CSC/TM-81/6166, ComStudy of Atmospheric Density Models int_ of Spacecraft Orbit Determination, R. Kuseski, M. Mallick, and R. Shanklin, Jr., September 1981

2.

3.

4.

5.

1-14