ADA17 16 P k I(N Ut U 4 LLRN (N[N S U THE /INENTONA RF NC IONOSPE AR F ORCE

NEPLSC A HANSCOM AB MA L F NANAA 2 0 SEP 83ACSI D AG -R8 2 FIG 4/1 NLE

JfIO'" fWo

l .1,.6

Lao112.0

11I.25 VII11[4 11.6

MICROCOPY RESOLUTION TEST CHARTNATIONAL BUREAU Of STANDARD. IRAJA

AFGL-TR-83-0239ENVIRONMENTAL RESEARCH PAPERS, NO. 863

Prediction of Total Electron Content Using theInternational Reference Ionosphere

4* LEO F. McNAMARA

20 September 1983

App..d f, publo ,.fa ..: disivbution u m..i.

DTICELO 7 f

SPACE PHYSICS DIVISION PROJECT 4643 JN2418

Q. AIR FORCE GEOPHYSICS LABORATORYC31 "."Scm Are. MASIACmUSuETT 01731

AIR FORCE SYSTEMS COMMAND, USAF

84 01 23 038

This report has tken reviewed by the LI) h>, lic Af'fliir., hI'fi .e ([A) Fuid isr] ',.ti-hhl e.- if Ueii rh r 1,:a. . ',n 1 ' . )

This technical report has been reviewed aruin is approvei Ir publieatoi.

>I

,4ERBERT C. CAMON, Chief RITA .X irecto:lonospheric Physics Division ,vi

Qualified requestors may obtain additional ,opie f !',. Vh I'fense Tt'oluii.Information Center. All others should apply t,, the Nnt i.,al Technica.Information Service.

If your address has changed, or If you wish to be removed from the mailin.list, or if the addressee is no longer employed by your organization, pleasenotify AFGL/DAA, Hanscom AFB, MA 01731. This will assist us in maintaininga current mailing list.

Do not return copies of this report unless contractual oblip.,aations or notice."on a specific document requires that It be returned.

ta

UnclassifiedSECURITY CLASSIFICATION OF THIS PAGE 1 111- Ok, 0uEn.,.d)

REPOT DCUMNTANON AGEREAD INSTRUCTIONSREPOT DCUMNTATON AGEBEFORE COMPLETING FORM

-REPORT NUMSER 2.GOVT ACCESSION NO. 3- RECIPIENT'S CATALOG NUMISER

AFGL-TR-83-0239 ____ _______

4. TITLE (Rid SilibitiI.) S. TYPE OF REPORT A PERIOD COVERED

PREDICTION OF TOTAL ELECTRONCONTENT USING THEINTRATOA Scientific. Final/'REFERENCE IONOSPHERE G. PERFORMING OA6. REPORT NUMDER

_____________________________ ERP. No. 8537. AUTHOR(.) S. CONTRACT OR GRANT MNMBERfC)

Leo F. McNamara

9 PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM EL EMENT. PROJECT. T ASK

Air Force Geophysics Laboratory (PHY) AREA & WORK UNIT NUMIIERS

Hanscom AFB3 62101FMassachusetts 01731 46430901

1 1 CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE

Air Force Geophysics Laboratory (PHY) 20 September 1983Hanscom AFB 13 NUMBER OF PAGES

Maq uP tts 0173 4214 ONTOIN AENCY NAME & AODRESS(II dIffere, from, Co,f,oting Otf(,c., IS SECURITY CLASS. tot ,t,,. r.po1,

Unclassified

I.DECLASSIFICATION DOWNGRADINGSCHEDULE

IS. DISTRIBUTION STATEMENT (.1 11,10 Re.or)

Approved for public release; distribution unlimited.

17. DISTRIBUTION STATEMENT (.1 heb. ,r 1 1.1 t- I. 0lock 20. fI difI.,.I I- .P., epr)

IL. SUPPLEMENTARY NOTES

*NRC/AFS Senior Research Associate on leave from theAustralian Ionospheric Predirtion Service, P.O. Boc 702,Darlinghurst 20 10, Australia

19, KEY WORDS (Cantinu. on recers. stdo if necesf-ry end identify by block "tboj

Total electron content predictionInternational reference ionosphere( . TEC prediction by IRI

20 STRACT 'Contilnu on re.... e side If 1,ecbsorI end idonity bi biork numberJ

The International Refe.,'en.ce Ionosphere (IHZ) is a model of the ionosphere,based ,,)i experimenta! data, that has been proposed as a standard ionosphericmode'.. As s-ich. it should be tested ex~tensively to determine its range ofvalidity. Ore.' *f the wavs in which the electron density profile given by the IRIespecially ubve the peak of the F lav'er, can be tested is to compare calculatejand '),)servedi values of total electron content (TECi. We have therefore studiethe discrepancies betwecoi ,al, ulato-d anid observed values of TEC recorded at

DD ~I 1473 EDITION OF I NOV 65 IS OBSOLETE Unclassified b4~SECJ*ITY CLASSIFICATION OF THIS PACE fWA., Do.. Eo~d)

It . I.

UnclassifiedSECUHITY CLASSIFICATION or THIS PAOE(*W at. CS.fted)

20. (Contd)

-1 northern hemisphere, and mainly for high levels of solar activity. Wehave found that the IRT produces reasonably accurate values of TEC atmid- and high-latitudes, but that it greatly underestimates the daytimevalues of TEC at low latitudes. We conclude therefore that the daytimeelectron density profile given by the IRI is reasonably accurate at mid-and high-latitudes: at least above the peak of the F2 layer. The situationat low latitudes clearly requires more work, and we have suggested Iwopossible lines of study. The generally low discrepancies at night indicatethat the nighttime electron density profiles given by the IRI correspondfairly closely to the actual profiles.

'

UnclassifiedSICURITY CLASSIFICATION OF THIS PAGItWH" b.0. E.,wO)

1tig .

Accession For

NTIS GRA&IDTIC TABUnannounc edJustificati o

Distrihft i on/

Avail: Militv Codes

Av: 1 ind/orDist~ Special

1, Preface

I would like to thank R. Leitinger, J.A. Klobuchar, and D.N. Anderson for

many useful discussions throughout the course of this study. The large amount

of data organization would not have been possible without the cheerful support of

Patricia Doherty. A large fraction of the TEC and foF2 data was supplied by the

World Data Center A for Solar Terrestrial Physics, Boulder, Colorado.

4.

i3

t

* I I

II

Contents

1. INTRODUCTION 9

2. METHOD OF ANALYSIS 11

3. "A" STATIONS 13

3. 1 Hamilton/Wallops Island 133.2 Goose Bay/St. Johns 16

4. "B" STATIONS 16

4. 1 Narssarssuaq/Goose Bay 164.2 Manila/Manila 214.3 Lunping/Chung Li 214.4 Palehua/Maui 21

5. "C" STATIONS 265. 1 Ascension Island 265.2 Osan, Korea 265.3 Ramey 26

5.4 Athens 305.5 La P-sta 305.6 Patrick 305.7 Shemya 305.8 Boulder 305.9 Anchorage 36

6. DISCUSSION OF RESULTS 36

6. 1 Mid-Latitude Stations 366.2 High-Latitude Stations 386.3 Low-Latitude Stations 386.4 Nighttime Discrepancies 39

7. CONCLUSIONS 42

REFERENCES 43

5

Illustrations

1. The IRI Electron Density Profile 10

2. Predicted and Observed Values of TEC at Hamilton for Four MonthsWith Low Solar Activity (1976) 14

3. Predicted and Observed Values of TEC at Hamilton for Four MonthsWith High Solar Activity (R = 100) 15

4. Predicted and Observed Values of TEC at Goose Bay for Four MonthsWith Low Solar Activity (1975, 1976) 17

5. Predicted and Observed Values of TEC at Goose Bay for Four MonthsWith High Solar Activity (R -- 100) 18

6. Predicted and Observed Values of TEC at Narssarssuaq forFour Months With Low Solar Activity (1974) 19

7. Predicted and Observed Values of TEC at Narssar'suaq F)rFour Months With Medium Solar Activity (1972) 20

8. Predicted and Observed Values of TEC at Manila for Four MonthsWith High Solar Activity (1981. 1982) 22

9. Predicted and Observed Values of foF2 at Manila for Four MonthsWith High Solar Activity (1981, 1982) 23

10. Predicted and Observed Values of TEC at Lunping (Taiwan) r~rFour Months With High Solar Activity (1981, 1982. 1983) 24

11. Predicted and Observed Values -af TEC at Palehua (Hawaii) forFour Months With High Solar Activity (1981) 25

12. Predicted and Observed Values of TEC at Ascension Island forFour Months With High Solar Activity (1980. 19811 27

13. Predicted and Observed Values of TEC at Osan (Korea) forFour Months With High Solar Activity (1981) 28

14. Predicted and Observed Values of TEC at Ramey (Puerto Rico) forFour Months With High Solar Activity (1981) 29

15. Predicted and Observed Values of TEC at Athens for Four MonthsWith High Solar Activity (1981, 1982) 31

16. Predicted and Observed Values of TEC at La Posts (California) forFour Months With High Solar Activity (1981) 32

17. Predicted and Observed Values of TEC at Patrick AFB (Florida) forFour Months With High Solar Activity (1981) 33

18. Predicted and Observed Values of TEC at Shenia (Alaska) forFour Months With High Solar Activity (1978. 1979) 34

19. Predicted and Observed Values of TEC at Boulder for Four MonthsWith High Solar Activity (1981) 35

20. Predicted and Observed Values of TEC at Anchorage for Four MonthsWith High Solar Activity (1980, 1981, 1982) 37

21. Calculated and Observed Values of TEC for January at a Station at13 ° Dip, for R = 137 40

22. Electron Density Profiles at Noon in January, R 137, for a Stationat 130 Dip, as Given by the Anderson and IRI-BENT Programs 41

6

-~~ I-.NOW-- ---

Tables

1. Geographic Locations of the Observing Stations, Sub-ionosphericPoints and lonosonde Stations for the 15 Cases Considered 12

7

i4

Prediction of Total Electron Content Using the

International Reference Ionosphere

1. INTRODUCTION

The International Reference Ionosphere (IRI) is an empirical model of the

ionosphere based on experimental observations. It is the outcome of work by the

U HSI Working Group G4. which is identical to the COSPAR task group on the Inter-

national Reference Ionosphere. The goals and status of the IRI have been discussed

by Rawer et al. IThe aim of the IRI is to establish a compendium of height profiles

through the ionosphere for the four main parameters, namely plasma density,

temperature of ions arid electrons, and ion composition. These parameters are

generated from a descriptive model containing reliable data that can be used to

obtain average profiles.

The IRI working group is well aware that the present model is inadequate in

some areas and has encouraged tests of the model's validity. We describe here a

test of the model's ability to reproduce observations of total electron content (TEC)over a wide range of conditions. If the tests are successful, they will confirm the

validity of the electron density profile N(h), especially in the region above the peak

of the F2 layer which contains approximately 2/3 of the TEC.

(Received for publication 15 September 1983)

1. Rawer, K. (1978) Goals and status of the International Reference Ionosphere,Res. Geophys. Space Phys. 16:77.

9

Haw2 has described the construction of the IRI Nih) profile, which is illus-

.rated in !gur'e I. [he key vAIue is the peak electron density NIdF2, which is

derived isrig the ('('ll model. The bottomside profiles are mainly based upon

ionog rams rd.iced t, true height profiles. The peak height HAlt12 is obta ned froM4

the p: r;zir tev r' 1 30001 2 using ;in empirical relationship dLe :" 1 3i Za. 4 ,e top-

side profile shape is deduced trom i sue iiarlv of topside ionosordc. rineasure-u cii 'nts5

(mainly A[.()IL L:TTEL as given by lient and Li.e%,ellyn. 5lowever, because of limita-

tions in the presentation of these data, the lt~l uses a modified form of the ort-ginal

data. 2

I: Topside (I)

HMF21

F2(2)

HMF1F 1 (3)

H ZHST -- -- Interrd 1:4)-. Figure 1. The 1lW Electron

I IA Density Profile

HEF

-- E-Volley (5) - - ORHAIR

HMEm~Dx : E/D(6)HD

HA

NME NMF2

Log N

2. Rawer, K. (1981) Introduction to IRI 1979, In "International Reference lonos-sphere - 1RI 79". J. V. Lincoln and R. 0. Conkright (Eds.) Report UAG-82,World Data Center A for Solar Terrestrial Physics, Boulder. Colorado.

3. CUIR (1974) Atlas of Ionospheric Characteristics, C(IR Report 340-2,International Union for Telecommunications, Geneva.

4. Bilitza, D. (1977) Modell zur Darstellung der Hohe des F2 - Maximums mitIfilfe des M-3000 - Wertes des (CIR. Kleinheubacher Ber. 20:4.

5. Bent, I. B. . and Llewellyn, S. K. (1970) Description of the 1965-71 IonosphericModel . sed in the Definitive Systems (DODS1, DBA-SYvstems, Melbourne. FL.

10

.. raw

A test of the 1RI, as far as its ability to calculate TEC is concerned, has been6

published by McNamara and Wilkinson. These authors found that at 31 ° S, the IRI

yielded discrepancies in TEC sometimes exceeding 30%, but usually less than 20%.

We present here a much more extensive analysis based on data from 15 stations,

covering a range of magnetic dip angles 14 to 77 ° .

2. METHOD OF ANALYSIS

The IRI profiles are essentially monthly median profiles and are referenc ' to

monthly median values of foF2 and M(3000)1'2. These parameters are calcul I

using the CCIR coefficients for a given month and linear interpolation in the -ionth

smoothed sunspot number R. t Consequently the most appropriate compariso,

calculated values of TEC is with observed values which have been linearly re I

to H. However, this necessitates a long sequence of data covering a solar c'

and only a few such data sets are available.

Since it is the IRI N(h) profile which is of most interest to us here, we would

like to test it under conditions for which it is the major source of error. The other

main potential source of error is in the calculation of foF2, so it is advantageous to

use observed values of foF2 at the sub-ionospheric point wherever possible. Again,

the values of foF2 should be linearly related to R.

We have, therefore, adopted a three-pronged approach in which we first of all

aim for the ideal test and then suucessively lift the restrictions. We consider three

types of stations/data:

1) ''A" Stations are those for which a set of TEC data and a set of observed

foF2 data are available for a solar cycle,

(2) "B" stations are those for which corresponding TEC and observed foF2

data are available on a monthly basis,

(3) 'C" stations are those for which the CCIR method must be used to

calculate foF2.

"rable I lists all the stations considered, arranged in increasing order of dip

angle. For each station we have considered where possible, data for for March,

June, September, and December at low and high levels of solar activity. When par-

ticular years must be used, we have used 1976 (R - 13) and 1981 (R - 140) if the

corresponding data were available. The exact values of R are not important.

6. McNamara, L.. F. , and Wilkinson, P. J. (1983) Prediction of total electron contentusing the International Reference lonosphere, J. Atmos. Terr. Phys. 45:169.

tFor convenience in this report, we shall use the symbol R instead of the usualR12"

11

* _ _ _ _ _ _

V'able i. (;egraphic, o ,zti s .i tne Observing Station,, ;-: -io s;Jhe 1 ilsand lon, .nde .tL-tikons t ' thte 1 1 ,,ses unrlslcIered

St:tiin '0" ft - _Point- _ ... -7 tT . .. .

,. , ii n2 ,'1 , 1 . i21 .

C'. Ascension

Island -3. 0 34.5.6 430. 4 -. 4 345.6

B. Lunping 25. 2 122. 2 35. 2 2:3.0 121.8 Oiung 1.i 25. 9 1>1. 2

B. Palehua 20. 7 203. T 38. 7 19. 8 202.6 Alaui 20. 2 203. 5

C. Osan 37.1 127.0 51.0 34.0 127.4

. arney 18.5 292.8 51.3 17. 1 292.6

C. Athens 38.0 23.6 54.5 34. 1 18.4

C. LaPosta 32.5 242.9 58.0 30.0 245.0

C. Patrick 28. 2 279.4 60. 0 26. 3 280.4

C. Shemya 52. 5 174. 1 63.0 46. 8 181.7

C. Boulder 40.0 254. 68. 3 36.3 259.6

A. Hamilton 42.6 289. 2 72. 9 38. 7 289. 3 \\allops I. 3i. 9 284.5

C. Anchorage 61.2 210.2 73.7 52.0 224.0

A. Goose Bay 53.5 299.5 76. 9 47.5 297. 8 St. Johns 47.6 307. 3

B. Narssarssuaq 61.2 314.6 77.5 53. 1 307.8 (,ooset Bay 53.3 [299.5

The 1RI has been made available as a F(ORTIAN computer program by \World

Data C'enter A for Solar Terrestrial Phvsics. We have modified the %%DC-A ver-

sion to permit the optional use of the C'hiu 8 model of foF 2 and H1\I-2 and of the

original Bent topside profile above 15I\I.2. Listings of the tihiu model are given

by Lincoln and Conkright. The modifications to include the original Bent topside

profile were taken from the program SLANTEC with some extra coding required

because of different systems of units.

It would be expected that, in general, the IRI would yield values of TEC some-

what less than the observed values, since the IRI sets the electron density to zero

above 1000 km. The TEC observations were obtained using the Faraday rotation

4 technique, which yields the TEC out to about 2000 ki. Tests using the Bent

ionospheric model indicate that the altitude range 1000 to 2000 km contributes

< 5% of the TEC up to 2000 km. This discrepancy would not be noticeable in the

present context.

(Due to the large number of references cited above, they will not be listed here.See References, page 43.

12

! !.- . j

3. "A" STATIONS

These are the stations for which a long sequence of TEC data covering a solar

cycle is available, in conjunction with a sequence of foF2 data at the sub-ionospheric

point, also covering a solar cycle. There are in fact only two such stations,

Hamilton and Goose Bay.

For each of these stations, we have correlated the observed monthly mean

values of 'FI.' at each hour with the smoothed sunspot number R. The least squares

fit regression lines have then been used to determine the values of TEC corres-

ponding to E = 0 and Ht : 100. A similar technique was applied to foF'2 at the sub-

ionospheric point.

It was found that the values of TEC for { = 0 were unreliable at night when the

values were very low. Consequently, we have been forced to use actual observa-

tions for low solar activity in lieu of It I 0 results.

3.1 I liimillhoi/Wallop. s Islald

TFhe Ilamilton '1EC that was used covered the period 1967 to 1980. The sub-

ionospheric point lies ver c Close to Wallops Island (see Table 1) for which fol"2

covering the period 1975 to 1981 were used. Fhe correlation coefficients of TEC

and fol2 with It were usually - 0. 95.

F.'igures 2 and 3 illustrate the observed and calculated diurnal variations of

TI.(' for I Iamilton at solar minimum (1976, It - 131 and maximum (iu - 100 exactly),

for Ma rtch, .June, September and l)ecembe r. "

It can be seen that there are no major discrepancies between the observed values

of 'E' and those calculated using the 11I. 1 Note that for 1976, ilamilton must be

considered a "B" station. I In general, the discr'epancies for I = 100 are least

when the It - 100 values of foV2 at Wallops Island are used. They are greatest in

l)ecember, It - 100, between about 08-12 LT. The figures also include the results

for a profile in which the II topside profile is replaced by the Bent profile.

t in all dragrans, the TEC is plotted as a function of local time in units of-2

10el , and the following code is used for the data points:x Observed monthly mean value (OBS)0 Value given by the original IRI (1111)o Value given by the I1I bottoniside and Bent topside models. This may be

the value obtained using the CCIR value of foF2 (IRI-BENT) or using theobserved value of fol"2 at the sub-ionospheric point (1I-BENT-OBS)

+ Value given by the IHI using the observed value of foF2 (IRI-OBS)Where necessary, the data points for the predicted values have been slightly offsetfrom the corresponding hour.

13

HAMILTONo 2 4 6 8 10 12 4 16 is 20 22 2 14

0 ~MAR 197600 0 +R- 13: rS75* # RI

00~ . RI-OHBS

4P 4PJUNE 1976

0 4Pgo +R- 13

0 0 0 0SEP 1976

0 0 DEC 19760 R. 13

TEC 5'5

2-2

togthe plottedintsd isgnd absre a afootnote in Setio 3.ilo 1 oorMnh

14

HAMILTONi I or IF' IS4 16 -21

400

R2000

300 oo0 0 0 IN-B T

200 0 IN - 8

1000 0

400TJUN

Cs 10R-00300

200

400

00 2 4 0 R- 100 i 0 2 2

200 ~ ~ LCA -TiME 00

Figur 3. Prdce n bevd\auso0 E tHmlo orMnh

0 115

I~0 _ _ _ _

3.2 Goose Bay/St. Johns

The Goose Bay TEC data covered the period 1973 to 1980, while the St. Johns'

foF2 data covered the period 1970 to 1978. Figures 4 and 5 show the calculated and

observed diurnal variations of TEC for Goose Bay for solar minimum (R - 13) and

solar maximum (R = 100) respectively. The greatest discrepancies occur for

December, I = 100, between about 08 and 12 LT (as with Hamilton). The accuracy

of the CCIR predictions of foF2 for St. Johns is such that the use of "observed"

R = 100 values of foF2 does not lead to a significant reduction in the discrepancies

(Figure 5).

For the low solar activity data, Goose Bay may be considered to be a "B" station

and comparisons made using the monthly median values of foF2 observed at St. Johns.

Figure 4 shows that when this is done, the discrepancies between the observed and

predicted values tend to increase. This is possibly because St. Johns is not quite

at the sub-ionospheric point for Goose Bay, but it is also possible that the better

results given using the predicted values of foF2 are the result of fortuitous cancella-

tion of errors in foF2 and in the N(h) profile.

4. "B" STATIONS

"B" stations are those for which corresponding values of TEC and foF2 at the

sub-ionospheric point are available, but the data do not cover a solar cycle. There

are four such stations, Manila, Lunping (Taiwain), Palehua, and Narssarssuaq,

which have sub-ionospheric points near the ionospheric stations at Manila, Chung Li,

and Goose Bay (see Table 1). We consider Narssarssuaq first because it causes the

least trouble.

4.1 Narssarssuaq/Goose Bay

Narssarssuaq data are available for only 1972- 1974, so we consider data for

"low" activity (1974. R - 30, Figure 6) and "medium" activity (1972, H = -60,

Figure 7). Figure 6 and 7 show that the values of TEC given by the IRJ are quite

accurate in larch, April, June and September, especially when the observed

Goose Bay foF2 values are introduced. There is a tendency for the diurnal varia-

tions of the TEC based on the observed values of foF2 to be somewhat irregular,

corresponding to irregular variations in foF2 due to the low sample sizes and the

complexities of the Goose Bay ionograms. A discrepancy between the observed

and calculated values of TEC remains for December in both 1972 and 1974, between

08 and 14 LT, even when the observed foF2 values are used. This indicates an

error in the N(h) profile given by the 1I1I.

16

-" -- -------

GOOSE BAY0 2 4 6 6 10 62 14 66 to 20 22 24

100 MAR t9760 0ft* 12

75 - 065* IRI

+ 0 IftI-09S

50 -

25+

0

100 JUNE 1976R-* 12

7500+4

5

00 SEP 1976

75

50

T 0 EEC17

17 1

I _ _ _ __ _ __0

GOOSE BAY0 2 4 6 S 1 2p l4 1 S 2 2 2I T I 7 7

40 - 0

0~ ~ 0 0

~~A*100,

I~~~~~R 100012 1 6 S 2

00 2 4 + 2 2

10018

20 -

NARSSARSSUAQ0 2 4 6 8 10 12 14 16 Is 20 22 Z4

+

100 0 +MAR 1974

* + O RB375 6 nB

0~ IRI-OHS

50

0 0 +025 + + 0

+ + JSE 1974

100 SEC 974

75

0 2E 419 0 2 14 1 I 042 2

LOCAL. TIM

Fiu 5 6. PeitdadOsre aus fT Ca asasuqfrFu

Mot sWt o w Soa ciiy(94

rE19

A~ ~ ~ ~2 ________________________ ___I..~~~~~~~~~ +__ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __

NARSSARSSUAQ

? 4 6p lr0 2 4 I? I

AP" 1972

200 - + I - 73 S

00

0 - 0

IDOS

0

200 DUE 972+ R. 51

1000

TEC :5

0

ISO0 2 4 I 0 22

02

-0

4.2 Manila/Manila

It is in the calculation of the Manila TEC that we encounter the first major dis-

crepancies between the calculated and observed values of TEC. Manila TEC data

were available for 1980-1982, apart from missing months. Figure 8 shows the

observed and predicted values of TEC for April 1982 (H = 124). June 1981 (H = 155),

October 1981 (R = 142), and December 1981 (H= 138). (March and September data

were not available.)

in general, the IRI predicts daytime values of TEC which are only about one-

half of the observed values, in all months. The errors at night are somewhat

smaller. The source of the error does not lie in the prediction of foF2, since as

Figure 9 shows, the error in the predicted foF2 is typically only about 10%. Use

of the original Bent profiles 9 improves the situation but still leads to prediction

errors reaching 300 Y 1015 el m - 2 and incorrect diurnal variations.

4.3 Lunping/Chung Li

The Lunping TL'(" data were available for 1980-March 1983, with some mssing

months. We have compared the predicted and observed values of TEC for March

1980 (R = 80), June 1981 (H = 142), September 1982 (R = 109), and December 1981

tR= 138).Figure 10 shows that, as with Manila, the IR] greatly underestimates the TEC

in some months, that the prediction of fot 2 is not the problem, and that the use of

the Bent topside profile leads to more accurate results. However, the discrepancy

between the predicted and observed TEC values still exceeds 300 10 15 el m- 2

during the day in September and December.

4.4 Palelua/Mawi

The Palehua TEC data were available for 1980-1982 and we have compared the

predicted and observed TEIL values for March, June, September, and December

(H - 140). Figure II shows that l1alehua follows l.unping and Manila in the way the

IRI greatly underestimates the TEL during the day during some months. The dis-1 255-2

crepancy is least for June, when it is of the order of 80 > 1015 el m or about 15V.

The daytime discrepancies remain e.ssentially unchanged when Maui observations

of foF2 are used. lHowever, the discrepancies at night tend to decrease.

Use of the original Bent profiles above the peak of the layer yields values in

better agreement with the observations, but large discrepancies still remain in

December. The calculated diurnal variation agrees well with the observations in

September, but not in the other months.

21

I. .

MANILA0 2 4 6 a 10 12 14 16 is 20 22 24

APR IM0Rt. 124

00000

41 0 b000.

200

JU9"I 191ft. 142

OC 116+

o. 4

600

400

LOA OCTE 1"I 0 222

200

0

Figur 8. Prdite anIbevdVle fT a aiafrFu o tsI

W 0h Hig Soa Acivt (181 192

22

f- 3

+ 0 +

MANILA0 2 4 6 8 10 12 14 16 1B 20 22 24

APR 1982R- 124

I- ' OBS

* ~ S RI

JUNE 1981

'-514

15

OCT 1981R ,142

'5

00O

DEC 1981

04-0 1 ILI 18 A0 22 24

Figure 9). Predicted and Observed Values of foF2 at Manila for Four MonthsWith High Solar Activity (1981, 1,18?). No observations were available forOectcf!.' 1981

23

LUNPING0 2 4 6 a 10 2 14 6 Is 20 22 24

0 0 MARCH j983

600- 0R-80

400-~~ 0 RI

200~~ - RI-BENT

4p(no Wof2 085)

0 1 -I IL1 .I I

00

0 JUNE 1981900 0 0R-142

400 -+ IRI-OBS

200

800 0 0SEPT I9R2R- 109

+ R .138

2024

.

I. C, 0

PALEHI.A

4 C~I 22

0 0 o R. ,

40*

" .o8s

0 - . - I00' 000I

0

0 0 JUNE

2000

0 0

SEPT r981

900 R; 43

400 +4RI

400 At

EC 198,X o rR 138

0 - o

00

I I

i LOCAL TIME

" l'igure 11. 1'redictedl and Observed \alue. f :I ' at Palehua (Hawaii) fr FourMlonths \With Higlh Solar Ar'tivi tv (1981)

25

LEC _ _ __0+_ _0

0 115e +

5. -C" STATIONS

These are the stations for which the calculation of foF2 relies on the CCIR

method, with no real means of checking the calculated values. However, the re-

suits for the "A" and "B' stations suggests that the major source of any discrepan-

cies will lie in the topside NIh) profile. We consider th 'C" stations in order of

increasing dip latitude.

5.1 Aseension Island

For Ascension Island we have considered the months March 1981 (H z 143).

June 1980 M1 = 153), Septenber 1980 (1 = 150), and December 1980 (H = 143). In

general, the daytime TEC values predicted by the 1IR3 are too low by at least1 - 2 15 -,

200 \ 1015 el , the discrepancy reaching 600 , 1015 el m -, or -50'%, in

March 1981 (see Figure 12). The original Bent topside profile yields somewhat

more accurate results, but for June 1980. retains the peculiar diurnal variation

given by the 11I.

[he post-sunset peak in the observed values of Tt' is missing from both sets

of predicted values, although there is a plateau from 1800-2400 ILT in December I980.

5.2 Osan. Korea

TEt" data for ()san w ere av.ilable for 1980-1982, %%ith sonc mlissing months,

noticeably the equinoxes. \\ e hatve considered the dita for April, lune, (October

and December 11081 R - 140) and the results art shown in I. igure 13.

The daytime values of TE(' calculated using the I t re generally quite good,

although the errors reach - 140 e 1015 L- m- for ()ctober 1981. The nighttime

values given by the I1I, on the other hand, tend to be too high by a factor of -"with a typical absolite error of -100 10 e m -2

5.3 Ra,,,e

Haev 'n.TI' obsvvations were available for 1980-1982 and we hazve conside red

the data for March, ,June, September, and )ecember 1911 Il - 140). the predi, td

values, shown in I igure 14, tend to underestimate the observtcd "IIC by up to,

-200 - 10 el ill-2 - 25'.) during the day, except in June. [here is no consistent

discrepancy at night, with only a small discrepancy in .1une and lDeoember, but :in

underestimite by 80 - 015 e I 2 1-30 0 during Alarh and .teptenhber.

26

I

ASCENSION ISLAND

0 2 4 6 a 10 1200 140 6s Is 20 22 24

0 % 0 REFLECTED

Bo8%00- - 1 MAR 190100 \ / R -143

*00 * 0 a Des00 0 0 IRI

60 0 0 Ot0 RI-BENT

400 r

200 00

0 I I

Boo 0JUNE 1980

600 00

600 0

400

2000

0 SE 1980

0 0R *150800

400

200

800 C

600

0 0 1 14 16 Is 20 22 2

Month With HihSla98iit0180 1

R274

OSAN0 2 4 6 a 10 12 14 16 Is 20 22 24

800 APR 1981

R- 140

600 .OBs0 IRI

400

800 JUNE 1981R- 140

600

OCT 1981

0 0

R 140

40C 0

0

800 DEC 1981

600V

400- *O 0

TEC

0 2 4 6 8 10 12 14 16 18 20 22 24

LOCAL TIME

Figure 13. Predicted and Observed Values of TEC at Osan (Korea) for FourMlonths With High Solar Activity (1981'

28

RAMEY0 2 4 6 8 10 12 14 16 Is 20 22 24

MAR 1981Bo00

R-* 140

600 - ais

400

0

800 JUNE 1961R- 140

600

400

800 SEP 1981R~ 140

600 0

200

00

DEC 1961

"00R *140

600

TEC 004

200

00

L. 2 4 6 8 10 12 14 16 Is 20 22 24

LOCAL TIME

Figure 14. Predicted and Observed Values of TEC at Ramey (Puerto Rico) forFour Months With High Solar Activity (1981)

29

5.4 Athens

Athens TEC data were available for 1980-1982. with some missing months.We have considered the data for March 1982 (R= 129). June 1981 (R= 142).

September 1982 (h = 117), and December 1982 (R z 138). The results are shown in

Figure 15.

The IRI TEC values during the day tend to be too low by up to -200 X 1015 el m- 2

in March and December. while the values during the night tend to be too high by up

to - 100 x 1015 el m -2 in September and December.

5.5 La Posta

La Po~ta TEC data were available from 1980 to 1983 and we have considered

the four months March, June, September. and December 1981 (R - 140) (see Fig-

ure 16). During the day, the discrepancies between observed and predicted values

reach - 150 ), 1015 el m -2 in March and December, with the IRI values being too

low. At night, the predicted TEC values tend tc be too high by a factor - 2. Fig-

ure 16 also shows the TLC values calculated using the original Bent topside profiles.

In this case, the results are very similar to the 1111 results.

5.6 Patrick

TEC data for Patrick were available for 1980-1983 and we have considered

March, June, September, and December 1981 (Rl - 140) tsee Figure 17). In general,

the IRI underestimates the TE( during the day by up to -200 K 1015 el m - 2 except

in June when the discrepancies are very loA. rhe discrepancies at night do not

yield a consistent picture.

5.7 SltemVa

For Sheyma, TEC data were available for September 1977 to April 1979, and

for 1980-1982. We have considered the months March 1979 (R = 137), June 1978

(1 = 89), September 1978 R= 108), and December 1.78 (f = 118) (see Figure 181,

In general, the discrepancies are reasonably small, except in June 1978, when the

calculated and observed diurnal variations of TEC do not follow each other very

closely.

5.8 Boulder

Boulder TEC data were available for 1980-1982 and we have considered the

months llarch, June, September, and December 1981 (H- 140). In general, theagreement between the observed and predicted values of TEC is quite good during

the day, but the predicted nighttime values are a factor of -2 too high (see Figure 19).

30

ATHENS0 2 4 6 a 10 12 14 16 is 2 22 24I

MAR 1962Boo R. 129

600 o 0iRs

400

20 0

So0 JUNE 1981R- 142

600

400

200

Boo SEP 1982

600

40

B00 DEC 1982R. 136

600

400TEC

1. 10,5 , -21 0 0

200L

0 0 2 4 6 8 10 12 14 I6 Is 20 22 24

LOCAL TIME

Figure 15. Predicted and Observed Values of TEC at Athens for Four MonthsWith High Solar Activity (1981,* 1982)

31

LA POSTA0 2 4 6 6 I0 12 14 16 Is 20 22 24

MAN MI6goo0 R- 140

600t b 6 in I6

0 0 iNi-IBNT

400-

Boo -JUNE 1961R - 140

600

0

Soo~ SEP 1961

Rt 140

600

400 0b6to...

200 0b 0 0 0 b

2000

400 2 S 10 1 4 1 S 2 2 2

0LOCALLTIME

Figure 16. Predicted and Observed Values of TEC at L-a llosta (Californial forFour Months With High Solar Activity (1981)

32

PATRICK0 2 4 6 8 10 12 14 is IS 20 22 24

MAR 1981

800 R - J40

400 08

200 0

800 JUNE 1981Ft - 140

600

400

20

800 SEP 1981pt 140

600

400 0

20

(So D0 EC 1981ft a 140

0 2 4 6 8 10 12 14 16 is 20 22 24

LOCAL TIME

Figure 17. Predicted and Observed Values of TEC at Patrick AFB (Florida) forFour Months With High Solar Activity (1981)

33

SHEMYA

o 2 4 6 8 10 12 14 16 8 20 22 241

MAR 1979800 -R- 137

0 0* RI0" 0 00 iRI-BENT

000

800 SE -

200 0 o

00 0

(800JUE 197880C R= R 8

T~ EC1,0 .T -

0 2 40 0 1 4 1 8 2 2 2

Month With 19hSla7ciit8178 99

,4(34

2BOULDER r p 2

AR196

Bo0 R.1140

60D

Boo0 JUNE 1981R-140

600

400

200 0

0 1

SEPT 1901

So0 R-140

600

400 )

20) 0

00

200-

8 02 16 IS 20 :

Figure 19. Predicted and Observed \alues of TEC at Boulder for Four MonthsWith High Solar Activity (1981)i

35

5.9 Anchorage

Anchorage TEC data were available for 1980-1983, with some missing months,

and we have considered March 1981 (R = 143), June 1982 (R = 117), September 1981

(R = 143), and December 1980 (R = 143) (see Figure 20). The discrepancies between

calculated and observed values of TEC during the day show no consistent trend, the

IRI values being too low in March, too high in December and fairly minor in June

and September. The discrepancies at night are also fairly minor. In this case,

the IRI values match the observations better than do the values calculated using the

original Bent topside profile.

6. DISCUSSION OF RESULTS

Since we have found that errors in the calculated values of foF2 are not the

major cause of the discrepancies between calculated and observed values of TEC,

and that the discrepancies are significantly greater at low latitudes, we shall discuss

the results under three latitude headings-mid, high, and low.

6.1 Mid-Latitude Stations

We consider here Osan, Ramey, Athens, La Posta, Patrick, Shemya, Boulder.

and Hamilton. The low latitude limit was set by the fact that Palehua TEC is

affected by the equatorial anomaly, while the upper latitude limit was set to 1. = 4.

The Boulder, Hamilton, and Osan results show no major discrepancies between

the observed and calculated values of TEC during the day, but the 1WI values tend

to be too high at night by a factor - 2. The 1111 results for Athens, I.a Posta,

Patrick, and Ramey are too low during the day in March, September, and December.

but have only minor errors in June (summer). There is no consistent discrepancy

at night. The discrepancies at Shemya are similar, except for June, when the cal-

culated values of TEC vary substantially both in absolute value and diurnal variation

from the observed values. The cause of these discrepancies is not known, but the

double-peaked diurnal variation of the TEC (which is also found for June 1981)

indicates that in June the ionosphere at Shemya differs from the normal mid-latitude

ionosphere. The discrepancies at the other stations in the other seasons may in fact

be partly due to errors in the predicted values of foFl2, but the consistency of the

errors over a wide range of longitudes also suggests that the profile shape may be

at fault as well.

36

ANCHORAGE0 a 4 6 a 10 12 14 16 Is 20 22 24

MAP 1961P- t43

0 0 0 A::-DENT

400000000

000

0 00

40

- SUE 1902

400-

Bo0 - DEC 19806 *R- 143

600 0

400 -6

400 0~

200- 0 ,-

0 ~ ~ 0 0

6 6 0 4 6 i 20 22 24

Wt ihSola AcivtyU90.191.l92

600 -0 0 0

P- 17

40 _ _0_

TEC 9 0

6.2 1igh-Latitude Stations

The results for Anchorage. Goose Bay, and Narssarssuaq are similar to those

for the mid-latitude stations, with no major discrepancies, especially when ob-

served values of foF2 are available. The largest discrepancy occurs for about

08-14 LT in December when the 1111 values are too high.

6.3 IA)w-Latitude Stations

The four low latitude stations, Manila, Ascension Island, Lunping. and Palehua

all yield considerable discrepancies between the observed and calculated values of

PEC. During the day, the calculated values can be a factor of 2 too low (absolute

error < 500 > 1015 el m-2), with an incorrect diurnal va riation. The IRI also

does not reproduce the observed post-sunset increase of TI'E at Ascension Island.

As indicated by the results for Manila, Lunping, and l'alehaa, the discrepancies

are not due to errors in the predicted values of fo[2 and must therefore be due to

the NOh) profile shape. Larger daytime values of TEL, which agree better with the

observed values, are obtained by replacing the Ifi topside profiles by the origina!

Bent profile. While this procedure was not found necessary at mid- and high-latitude

stations, it is at least a step in the right direction for low latitude stations. In some

months, the modified 1(1I profiles yield good agreement with the observations. Even

so. the calculated diurnal variations of TEC often do not match the observed day-

time variations, nor' the post-sunset increases at Ascension Island.

The IRI-BENT (that is, with the original Bent topside profile) results for Manila

indicate a "crisis of confidence" which occurs at around 08 LT-the calculated TE'

flattens out rather than continuing the post-sunrise increase. Consideration of the

diurnal variation of fo1F2 at Manila, with foF2 going through 10 Mllz at around 07 [.,

led us to consider a possible modification of the Bent topside profile, in particular

the semi-thickness of the topside parabolic YT. The coding for I'ROFIL2 (l.lewellyn

& and Bent; 9 especially page 181) shows that if fot'2 exceeds 10.5 MHz, YT is given

by the formula:

YT = YM X 10. 133 X(foF2 - 10.5) + 1

where YM is the half thickness of the bottomqide parabola. Since the TEU increases

when YT is increased, the value of TEC given by the IRI-BENT model can be in-

creased when foF2 > 10.5 MHz simply by increasing the factor 0. 133. Increasing

it to 0. 5 yields TEC values more in agreement with the observed daytime values.

The predicted daytime TEC values for the IRI-BENT model can also be in-

creased (by a lesser amount) by relaxing the upper limit on the 10. 7 cm flux set by

Llewellyn and Bent. 9 Subroutine REFRAC (Report page 172) limits the flux to 130.

38

MOON.~

Rtenoving this limit allows the exponent of the lower topside exponential to achieve

lower values, yielding a more slowly decaying profile. (See Figure 4 of Llewellyn

aind Bent. 1 The exponents nust, however, be constrained to stay positive, which

.,s thei n reason for the limit on the flux value.

Better igr'eemient with the observ'ed daytime UL( vailues at low latitude stations

ntv thus he obtained by jidicious manipulation of the parameters of the Bent top-

side p)rofile. \We ha ve not pursued this point. All results given in this report were

obtained using the unniodified Bent profiles.

"hie [iLU at Ascension Island has been modeled theoretically by Anderson and12

Kh lbuchar, who were able to reproduce the observed TE' vairiations including

the lite afternoon decrease and post-sunset enhancements. In :an (as vet) incomplete

study of the effects of IE \ B drifts at equatorial stations, Anderson and NIcNamara

calcul:ted the Nh) profiles and TEC for l'almyra (5.9N. 197.8E, 130 dip) for

J:nuary 1949 (i= 137). The calculated TEc values are shown in Figure 21, which

arlso shows the observed TEC for Manila for January 1982 (II R 137), the IHI TEC

values and the daytime IMII-BENT 'rt values. The theoretical values (A) are

clearrlh the best match to the observed data, although some discrepancies still

renain at aibout 18-22 I.T. These mav in fact be due to different EC B drifts

existing at Manila and lPalnyra, and are not our present concern. We are con-

cerned rather with the considerably different N(h) profiles given by the Il(I-IINT

program and by the theoretical model. Figure 22 shows that the two profiles for

12 I. F. January, i - 137, bear little resemblance to each other, the theoretical

profile being nuch broader than the II-BENT profile (and consequently yielding a

greater Tit'). This discrepancy points up the need for further studies of the pro-

file shape at low latitudes. The fact that at least half of the "missing' TEL would

appevr to c,,ue from below the peak of the layer suggests caution in trying to match

the observe, iles of TEC simply by adjusting parameters of the topside profile.

6.4 Nighltfite IDiscrepanc'ie.i

We have tended so fat to discuss mainly the discrepancies which occur during

the day, since these have been found to have the largest absolute values. However,

the discrepancies might tend to reach higher relative values, which may be of more

concern in some applications than the absolute discrepancies. There seems to be

no general trend in the discrepancies, overestimates and underestimates being

about equally likely. Some of the "A" and "B" stations, (for example. Hamilton and

12. Anderson, D. N., and Klobuchar, J. A. (1983) Modeling the total electronelectron content observations above Ascension Island, J. Geophys. Res.,88:8020.

39

* 1 _ _ _

Goose Bay) shows very small discrepancies when observed values of foF2 are used.

For Manila and Narssarssuaq. on the other hand, the 1111 systematically under-

estimates the TLC, even when the observed foF2 values are used.

TEC900

A A

A M A

800 / NA

604.oo\ AM \

500 A11/ gRED

400 -i

A \ MA

300M M0

M MANILA JAN 1982 R 137200 A 1OOSE RVEO)

I t I PALMYRA JAN 1949 R- 137

\\I A PALMYRA JAN 1949 A-137(I. lo"., M 1j 7 ANDERSON)

1 0 PALMYRA JAN 1949100 Ilk / UIRI- 8ENTI

I AT LON DIP

MANILA )4.7 '21 14 5PALMYRA 5 9 198 13 0

00 12 14 16 .B 20 22 24

LOCAL TIME

Figure 21. Calculated and Observed Values of TEC forJanuary at a Station at 13 ° Dip, for H = 137. The ob-served values are for Manila, while calculated valuesare given for the IIi, the modified IRI-BENT and theAnderson theoretical models. The Anderson valueswere calculated for Palmyra

40

OD It-(

in 0 >

44a 00

.jZ -- 0 cz--'

0L.a"

z w-41-

co

z-

Z C

-0

cr o0

0

0Lii

0cJo

Mc

S0 0 0 0 0 0 -

0 0 0 0 0 0

41

| .0.. . . . ., I.

-

7. CONCLUSIONS

We have found that, in general, the Iil yields reasonable, and often quite

accurate, values of TEC at mid- and high-latitude stations, but that it tends to

underestimate severely the daytime values of TEC at low latitude stations. Better

agreement with the observed daytime values of "'EC at low latitudes can be obtained

by replacing the IRI topside profile by the original Bent profile, but theoretical

model studies indicate that a substantial revision of both tne llIM and ItI-IBENT

profiles is required at low latitudes. In those areas for which agreement has been

found between observed and predicted values of TE, the IR1 electron densit. pro-

files are presumably good representations of the actual profiles. lrom an ope ,-

tional point of view, this means that the lIil may be used to calculate reliable values

of TEC in those same areas.

4

References

I. H ,tc,, K. 1978) Goals and status of the International Reference Ionosphere,Res. ( eoph's. Space Phys. 16: 177.

2. Rawer, K. (1980 Introduction to IRI 1979, In "International Reference lonos-phere - I II 79', J.\ . Iincoln and It.(). (onkright (Eds. ) Report UAG-82,Wkorld Data tCenter A for Solar Terrestrial Physics, Boulder, Colorado.

3. ('Vl I (1974, Atlas of Ionospheric Characteristics, CCIR R/eport 340-2,interna tion5a Union for Telecommunications, Geneva.

4. Bilitza, 1). (1977) Mlodell zur I)arstellung der Hohe des F2 - Maximums mitIlilfe des lI-3000 - Wertes des (CIR, Kleinheubacher Ber. 20:4.

5. Bent, It. B. . and Ilewellyn, S. K. ( 19701 Description of the 1965-71 IonosphericModel I sed in the Iefinitive S%'stem (DODS), DBA-Systems, Melbourne, FL.

6. Mcm\a t'ar, I.. F. , and \ ilkinson, 11..J. ( 19831 Prediction of total electron con-tent using the International Reference Ionosphere, J. Atmos. Terr. Phys.45: 169.

Lincoln. .1.\, and Conkight, R. 0. ( 1981) International Reference Ionosphere -

IHf-79, lipor-t I.A(;-82, World Data center A for Solar Terrestrial Physics,T7- -r, C'().

8. thiu, Y. 'r. (19751 An improved phenomenological model of ionospheric density,.1. :ktmos. [rerr. Ih Ls. 1563.

9. Ilewellyn, S. K. , and Bent, It. 11. (1973) Documentation and description of theBent Ionospheric Model, AFCHL-TR-73-0657, AD 772733.

10. McNamara, I.. F. (1983) The Conversion of Off-vertical Observations of TotalElectron Content into Equivalent Vertical Incidence Values, AFGL-TR-83-U092.

I I. Titheridge, J. E. (19721 Determination of ionospheric electron content from theFaraday rotation of geostationary satellite signals, Planete .v and Space

Science, 20:353.

12. Anderson, I). N.. and Klobuchar, J.A. (1983) Modeling the total electron

electron content observations above Ascension Island, J. Geophys. Res.,88:8020.

43

&...........