Pos si ble Spa tial Ex tent of Ion o spheric GPS-TEC and NmF2Anom a lies Re lated to the 1999 Chi-Chi and Chia-Yi Earth quakes

in Taiwan

Masahide Nishihashi1, Katsumi Hattori 2, *, Hau-Kun Jhuang 3, and Jann-Yenq Liu 3, 4

1 Grad u ate School of Sci ence and Tech nol ogy, Chiba Uni ver sity, Chiba, Ja pan2 Grad u ate School of Sci ence, Chiba Uni ver sity, Chiba, Ja pan

3 In sti tute of Space Sci ence, Na tional Cen tral Uni ver sity, Jhongli, Tai wan, ROC4 Cen ter for Space and Re mote Sens ing Re search, Na tional Cen tral Uni ver sity, Jhongli, Tai wan, ROC

Re ceived 17 October 2008, ac cepted 22 January 2009

AB STRACT

Sci en tists an a lyzed tem po ral vari a tions of lo cal data re corded in Tai wan and re ported anom a lous re duc tions in the

ion o spheric to tal elec tron con tent (TEC) on days 4 and 3 be fore (17 and 18 Sep tem ber) the Chi-Chi earth quake (Mw 7.6,

ML 7.3), and days 3 and 1 prior to (19 and 21 Oc to ber) the Chia-Yi earth quake (ML 6.4) in Tai wan. In this pa per, we pro pose a

spa tial anal y sis which com pares the data re corded in side and out side of the earth quake area around Tai wan to dis crim i nate

whether the anom a lies are lo cal (earth quake re lated) or global (non-earth quake re lated) phe nom ena. Re sults sug gest that the

anom a lies ap pear ing day 3 be fore the Chi-Chi earth quake, and days 3 and 1 be fore the Chia-Yi earth quake are lo cal

(earth quake re lated) phe nom ena and the anom aly ap pear ing on 17 Sep tem ber 1999 (day 4 be fore) might be the re sult of both

global phe nom ena (i.e., a geomagnetic storm) and the local effect of the Chi-Chi earthquake.

Key words: To tal Elec tron Con tent (TEC), NmF2, Anom a lous re duc tion, Spa tial anal y sis, The 1999 Chi-Chi and Chia-Yi earth quakes

Ci ta tion: Nishihashi, M., K. Hattori, H. K. Jhuang, and J. Y. Liu, 2009: Pos si ble spa tial ex tent of ion o spheric GPS-TEC and NmF2 anom a lies re lated to the1999 Chi-Chi and Chia-Yi earth quakes in Taiwan. Terr. Atmos. Ocean. Sci., 20, 779-789, doi: 10.3319/TAO.2009.01.22.01(T)

1. IN TRO DUC TION

Many anom a lous elec tro mag netic phe nom ena pos si bly

as so ci ated with large earth quakes have been re ported (e.g.,

Hayakawa and Fujinawa 1994; Hayakawa 1999; Hayakawa

and Molchanov 2002; Pulinets and Boyarchuk 2004). Ion o -

spheric anom a lies be fore large earth quakes were re ported

first in the be gin ning of the 1970s (Datchenko et al. 1972;

Birfeld 1974; Nestorov 1979). Sci en tists (Pulinets et al.

1994; Pulinets 1998; Liu et al. 2000, 2006; Chuo et al. 2001,

2002) ob served that the ion o spheric plasma fre quency (or

cor re spond ing elec tron den sity) de creased a few days be fore

some large earth quakes oc curred. Ionosondes have been the

most pop u lar in stru ments for prob ing the ion o spheric elec -

tron den sity for more than seven de cades (Hunscucker 1991).

They pro vide ex cel lent op por tu ni ties for study ing pre- earth -

quake ion o spheric anom a lies. Gokhberg et al. (1989) re -

vealed the ex is tence of anom a lies of the crit i cal/plasma fre-

quency at the ion o spheric F2-peak (foF2) a few days be fore

sev eral earth quakes. Cur rently, there are more than 200

ionosondes avail able world wide; how ever, only a frac tion

of them is rou tinely op er ated. There fore, spa tial and tem -

poral cov er age of ionosonde ob ser va tions are lim ited and

difficult to use sys tem at i cally for cor re la tion with seis mic

ac tiv i ties. More over, be cause of the use of me dian and high

fre quen cies (MF and HF) of 1 - 20 MHz, ionosondes of ten

suf fer from short-wave fadeout and in turn data gaps (Davies

1990).

On the other hand, due to de vel op ments in space tech -

nol ogy, to day there are thou sands of ground-based re ceiv ers

of the global po si tion ing sys tem (GPS) de ployed to mon i tor

the Earth’s sur face de for ma tion rates (see the pa pers listed in

Calais and Amarjargal 2000). Ow ing to the use of two ul -

tra-high fre quency (UHF, f1 = 1575.42 MHz and f2 =

1227.60 MHz) waves, GPS sig nals are gen er ally free from

short-wave fadeouts. Be sides the Earth’s sur face de for ma -

tion ob ser va tions, the same net work of GPS re ceiv ers can

Terr. Atmos. Ocean. Sci., Vol. 20, No. 6, 779-789, December 2009 doi: 10.3319/TAO.2009.01.22.01(T)

* Cor re spond ing au thorE-mail: hattori@earth.s.chiba-u.ac.jp

also be used to si mul ta neously and con tin u ously mon i tor the

ion o spheric to tal elec tron con tent (TEC) (e.g., Sardon et al.

1994; Leick 1995; Liu et al. 1996). Re search ers have re -

ported anom a lous re duc tions in GPS-TEC (e.g., Liu et al.

2001, 2004a) and/or foF2 (Liu et al. 2000, 2006; Chuo et al.

2001, 2002) that ap peared within 1 - 5 days prior to M ³ 6.0

earth quakes in Tai wan. Liu et al. (2004b) re ported the sta tis -

ti cal re sult that the GPS-TEC and foF2 de creased be fore 144

M ³ 5.0 earth quakes in Tai wan dur ing 1997 - 1999. More -

over, Chen et al. (2004) dem on strated the sta tis ti cal sig nif i -

cance of ob served foF2 anom a lies within 1 - 5 days prior to

M ³ 5.0 earth quakes be ing re lated to such earth quakes in the

Tai wan area for the pe riod 1994 - 1999. Liu et al. (2006)

defined the cri te ria for de tect ing pre-seis mic events in the

ionosphere and then ap plied those cri te ria with foF2 and

184 M ³ 5.0 earth quakes in Tai wan be tween 1994 and 1999.

Their study showed both pos i tive and neg a tive anom a lies

but that only the neg a tive ones yielded sta tis ti cal sig nif i -

cance. They found, af ter re mov ing geo mag netic storm ef -

fects, that ion o spheric foF2 sig nif i cantly de creased dur ing

the af ter noon pe riod 1 - 5 days be fore the earth quakes.

Based on the sta tis ti cal re sults, sci en tists (Liu et al. 2000,

2001, 2002, 2004a, b; Chuo et al. 2002) claimed that the

pre-earth quake anom a lies in the ion o spheric foF2 or GPS-

TEC ap peared on 3 and 4 days be fore the Chi-Chi earth -

quake (Mw 7.6, ML 7.3), and 1 and 3 days prior to the Chia-Yi

earth quake (ML 6.4). How ever, these stud ies have not been

crosschecked by si mul ta neous data sets ob served at other

places to con firm that the ob served ion o spheric anom a lies

were re lated to the earth quakes.

In this pa per, we con duct a spa tial anal y sis by cross-

examining si mul ta neous data of NmF2 and GPS-TEC in

Taiwan, Ja pan, and other places to find whether the anom a -

lies ob served in Tai wan dur ing three pe ri ods of the Chi-Chi

earth quake, the Chia-Yi earth quake, and be tween the two

are lo cal or global phe nom ena. The NmF2 and TEC are de -

rived by lo cal ionosondes and GPS re ceiv ers in Tai wan and

Ja pan, while GPS-TEC at other places are re trieved from the

global ion o sphere maps (GIM) pub lished by the Cen ter for

Or bit De ter mi na tion in Eu rope (CODE; ftp://ftp.unibe.ch

/aiub/CODE/).

2. DATA ANAL Y SIS

The Mw 7.6 (ML 7.3) Chi-Chi earth quake oc curred at

0147 LT (120 EMT), 21 Sep tem ber 1999 (1747 UT, 20 Sep -

tem ber) in cen tral Tai wan. This large earth quake re sults in

sur face rup tures ex tend ing over about 85 km along the

Chelungpu fault with ver ti cal thrust and left lat eral strike-

slip off sets (Ma et al. 2000). About a month later, at 1018 LT

(120 EMT), 22 Oc to ber 1999 (0218 UT, 22 Oc to ber), the

ML 6.4 Chia-Yi earth quake oc curred at about 55 km south -

west of the epi cen ter of the Chi-Chi earth quake. The Chi-

Chi earth quake is the larg est event in Tai wan in the last cen -

tury (Chang et al. 2000; Shin et al. 2000) and the Chia-Yi

earth quake was an other very large event that oc curred soon

af ter the Chi-Chi earth quake. The or i gin time, lo ca tion, and

mag ni tude of earth quakes in the Tai wan area dur ing the

study pe riod can be re trieved from Cen tral Weather Bu reau

Seis mo log i cal Net work (CWBSN).

To con firm the ion o spheric anom a lies be ing re lated to

the two earth quakes as re ported pre vi ously (Liu et al. 2000,

2001, 2004a; Chuo et al. 2001, 2002), we an a lyze the F2-

peak elec tron den sity (NmF2) de rived from the foF2 not

only at Chung-Li (CHL; 25.0°N, 121.2°E) in Tai wan but

also at Kokubunji (KKB; 35.7°N, 139.5°E) in Ja pan (Fig. 1).

Note that CHL is close to the two epi cen ters while KKB is

far from them al low ing for mon i tor ing both in side and out -

side the two earth quake ar eas, re spec tively.

Tai wan is lo cated in a seis mi cally ac tive zone of the rim

en cir cling the Pa cific Ocean and, there fore, earth quakes

occur fre quently. For in stance, the re cur rence in ter val of

an M ³ 5.0 earth quake be tween 1991 and 1999 was about

13 - 15 days (Liu et al. 2004a). To min i mize pos si ble con -

founding ef fects of con sec u tive earth quakes and prop erly

iden tify the ab nor mal sig nals, we com puted the mean NmF2

(NmF2_mean) val ues for the pre vi ous 15 days, and the as so ci -

ated stan dard de vi a tion (s) as a ref er ence at spe cific times.

Then, we de rived the nor mal ized NmF2 (NmF2*) val ues by

the fol low ing equa tion:

(1)

The sam pling in ter val at CHL is 15 and 60 min. at KKB. If

the NmF2* falls out of -2s (t), we then de clare that the ab -

nor mal sig nal is de tected.

Mean while, by us ing the GAMIT soft ware, widely

adopted in ge od esy and earth sci ences, two co-lo cated TECs

de rived from GPS re ceiv ers S105 (23.0°N, 121.1°E) in

Taiwan and TSKB (36.1°N, 140.1°E) in Ja pan (Fig. 1) are

calculated. GAMIT is a com pre hen sive GPS anal y sis pack -

age de vel oped at the Mas sa chu setts In sti tute of Tech nol ogy

(MIT) and the Scripps In sti tu tion of Ocean og ra phy (SIO)

for the es ti ma tion of three-di men sional rel a tive po si tions of

ground-based re ceiv ers and sat el lite or bits (King and Bock

2004). Note that the GPS-TEC de rived by the GAMIT is

con sid ered as an ob ser va tion with out any fur ther smooth ing

or in ter pre ta tion pro cess. Sim i lar to Eq. (1), the nor mal ized

TEC (TEC*) is given as:

(2)

where the TEC* is de rived ev ery 30 sec and av er aged ev ery

15 min. More over, for es ti mat ing the spa tial ex tent of TEC

780 Nishihashi et al.

anom a lies in de tail, we com puted TEC data in the Oki -

nawa- Kyushu area us ing GPS data ob served by the GPS

Earth Ob ser va tion Net work (GEONET) (Miyazaki et al.

1997). The Oki nawa area is lo cated at al most the same lat i -

tude as Tai wan, 24 - 27°N but slightly shifted in the lon gi -

tude, 123 - 131°E, while the Kyushu area is lo cated in 27 -

34°N and 129 - 132°E.

To ob tain GPS-TEC val ues at any lo ca tion, the GIM is

ide ally used. The GIM is gen er ated on a daily ba sis at CODE

us ing data from GPS sites of the In ter na tional GNSS Ser vice

(IGS) and other in sti tu tions. The ver ti cal TEC is mod eled in

a so lar-geo mag netic ref er ence frame us ing a spher i cal har -

mon ics ex pan sion. In 1999, the de gree and or der of the

spher i cal har mon ics ex pan sion were 12 and 8, re spec tively.

Piece-wise lin ear func tions are used for rep re sen ta tion in the

time do main. The spa tial res o lu tion is 2.5-de gree in lat i tude,

and 5-de gree in lon gi tude and the tem po ral res o lu tion is 2

hours. Schaer (1999) de scribes the com pu ta tional al go rithm

in de tail. For sim plic ity of cross com par i son among the ob -

ser va tions, we lin early in ter po late yield ing a 15-min ute re -

solution at a cer tain lo ca tion. Re mem ber that the GIM-TEC

is ob tained via the spher i cal har mon ics ex pan sion and the

lin ear in ter po la tion. Sim i larly, the nor mal ized GIM-TEC

(GIM-TEC*) can be given as:

(3)

Tai wan is also right un der the well-known equa to rial ion -

iza tion anom aly (EIA) (Ratcliffe 1972) re gion. To have a suit -

able cross com par i son, GIM-TEC at four EIA re gions, f30

(14.3°N, 30.0°E), f120 (22.1°N, 120.0°E), f210 (10.1°N,

150.0°W), and f300 (2.0°N, 60.0°W) have been iso lated

(Fig. 1). Note that the mag netic lat i tudes of these four EIA re -

gions are 12.0°N. In ad di tion, we ex tract GIM-TEC* co- lo -

cated at CHL and KKB (de noted, CHLg and KKBg) for fur -

ther ref er ences. Ta ble 1 sum ma rizes lo ca tions of the iono -

sondes, the GPS re ceiv ers, and the GIM points in de tail. Mean -

while, at low lat i tudes such as Tai wan, the ion o spheric NmF2

and/or TEC also could be de pressed about a few hours to 2

days af ter geo mag netic storm onsets (Kelley 1989; Davies

1990). To en sure that an ob served NmF2 and/or TEC anom a -

lous de pres sion is earth quake re lated, we re quire that the re -

duc tion of the geo mag netic in dex Dst does not ex ceed -70 nT,

oth er wise it is con sid ered to be geo mag netic storm re lated.

3. RE SULT AND IN TER PRE TA TION

Fig ure 2 dis plays that vari a tions of NmF2 at CHL and

Spa tial Ex tent of Ion o spheric Anom a lies of Tai wan Earth quakes 781

Fig. 1. Lo ca tions of ionosondes, GPS re ceiv ers, GIM-TEC points, and the epi cen ters of the Chi-Chi, Chia-Yi earth quakes, and M 5 - 6 earth quakes in

Tai wan from 14 Sep tem ber to 28 Oc to ber 1999.

782 Nishihashi et al.

Fig. 2. The NmF2, TEC, GIM-TEC, Kp in dex, Dst in dex, and M ³ 5.0 seis mic ac tiv ity dur ing the Chi-Chi earth quake pe riod from 14 to 26 Sep tem ber

1999. The gray, red, and blue shaded re gions de note the geo mag netic dis turbed, global (pos si ble mag netic storm), and anom aly fea tures, re spec tively.

TEC at S105 yield low val ues on 17 and 18 Sep tem ber,

which are days 4 and 3 be fore the Chi-Chi earth quake. The

vari a tions of TEC at Minami-daitoh (0497; 25.8°N, 131.2°E)

in the Oki nawa area show sim i lar ten den cies with S105. In

or der to de tect anom a lies, we cal cu lated the run ning stan -

dard de vi a tion s of the pre vi ous 1 - 15 days and nor mal ized

the ob served value on the 16th day. Fig ure 3 il lus trates vari a -

tions of the nor mal ized quan ti ties of NmF2*, TEC*, and

GIM-TEC*. It may be seen that the NmF2* at CHL and

TEC* at S105 ex ceed the lower thresh old of -2s on 17 and

18 Sep tem ber, which are 4 - 3 days be fore the Chi-Chi earth -

quake. It is sur pris ing to see that the GIM-TEC* at CHLg

de creases, slightly ex ceed ing -2s on 17 Sep tem ber, but no

clear re duc tion is ob served on 18 Sep tem ber. By con trast,

the NmF2* at KKB, TEC* at TSKB, and GIM-TEC* at

KKBg de crease, ex ceed ing the -2s lower thresh old on 15

and 16 Sep tem ber (days 6 and 5 be fore the earth quake). Al -

though TEC* at TSKB and GIM-TEC* at KKBg de crease

slightly ex ceed ing -2s on 17 Sep tem ber, no de crease fea ture

is found on 18 Sep tem ber. Note that the NmF2* at KKB is

about -2s on 17 but not on 18 Sep tem ber. The TEC* at 0497

is about -3s on 17 Sep tem ber. In gen eral, the trends of

GIM-TEC* are sim i lar to those of TEC* and NmF2* in

Taiwan and in Ja pan. More over, GIM-TEC* at f30, f120,

f210, and f300 de crease ap proach ing -2s on 17 and those

on 18 Sep tem ber stay within ±2s (within nor mal vari a tion).

This sug gests a global ef fect ap pear ing on 17 Sep tem ber.

After the geo mag netic storm sud den com mence ment (SSC)

on 22 Sep tem ber, the re lated re duc tion anom aly of TEC* at

S105 is de tected on 23 Sep tem ber. In this pe riod, due to an

elec tric power fail ure caused by the Chi-Chi earth quake,

NmF2 was not con stantly re corded at CHL.

Spa tial Ex tent of Ion o spheric Anom a lies of Tai wan Earth quakes 783

Fig. 3. The nor mal ized quan ti ties NmF2*, TEC*, and GIM-TEC*, as well as the geo mag netic Kp in dex, Dst in dex, and M ³ 5.0 seis mic ac tiv ity dur -

ing the Chi-Chi earth quake pe riod from 14 to 26 Sep tem ber 1999.

Fig ures 4 and 5 re veal vari a tions of the NmF2, TEC,

GIM-TEC, Kp in dex, Dst in dex, and seis mic ac tiv ity as well

as the as so ci ated nor mal ized val ues dur ing the pe riod be -

tween the Chi-Chi earth quake and the Chia-Yi earth quake,

from 27 Sep tem ber to 15 Oc to ber, re spec tively. Again due to

the power fail ure af ter the Chi-Chi earth quake, the CHL

ionosonde yields many data gaps in the NmF2* data (Fig. 5).

No tice that the value of TEC* at S105 de creased be yond -2s

on 1 Oc to ber.

Vari a tions of the NmF2, TEC, GIM-TEC, Kp in dex,

Dst in dex, and the seis mic ac tiv ity from 16 to 28 Oc to ber

are il lus trated in Fig. 6. The geo mag netic con di tion is rel a -

tively quiet un til 21 Oc to ber, which is one day be fore the

Chia-Yi earth quake. In Fig. 6, it may be seen that the daily

max i mum val ues of NmF2 at CHL and TEC at S105 on 21

Oc to ber are small in com par i son with the other days. In

fact, on 19 and 21 Oc to ber, days 3 and 1 be fore the Chia-Yi

earth quake, the NmF2* at CHL de creases ex ceed ing the

-2s thresh old (Fig. 7). How ever, the GIM-TEC* at CHLg

does not show any anom a lous be hav ior on those days.

More over, the NmF2* at KKB, the TEC* at TSKB and

0497, and the GIM-TEC* at KKBg do not de crease ex -

ceeding the -2s thresh old on 19 and 21 Oc to ber ei ther.

Similarly, the GIM-TEC* val ues at f30, f120, f210, and

f300 do not de crease ex ceed ing -2s on days 3 and 1 be fore

the Chia-Yi earth quake. Al though the causal re la tion is not

un der stood, the GIM-TEC* at f300 de creased significantly

ex ceed ing -2s on 18 Oc to ber.

Ta ble 2 dis plays the cor re la tion co ef fi cients be tween

TEC* com puted by GAMIT and co-lo cated NmF2* re -

corded by CHL ionosonde as well as TEC* and co-lo cated

GIM-TEC* in Tai wan and Ja pan dur ing the three pe ri ods.

For all three pe ri ods, the cor re la tion co ef fi cients be tween

TEC* and NmF2* (or TEC* and GIM-TEC*) in Ja pan are

784 Nishihashi et al.

Fig. 4. The NmF2, TEC, GIM-TEC, Kp in dex, Dst in dex, and seis mic ac tiv ity of M ³ 5.0 earth quakes dur ing the pe riod be tween the Chi-Chi and

Chia-Yi earth quakes from 27 Sep tem ber to 15 Oc to ber 1999.

greater than those in Tai wan, re spec tively. In fact, each cor -

re la tion co ef fi cient be tween TEC* and NmF2* is greater

than that be tween TEC* and GIM-TEC*, cor re spond ingly.

Mean while, it is in ter est ing to find that the cor re la tion co -

ef fi cients be tween TEC* and GIM-TEC* yield the two

small est val ues dur ing the Chi-Chi earth quake and the Chia-

Yi earth quake pe ri ods.

4. DIS CUS SION AND CON CLU SION

Shiokawa et al. (2002) re ported that three se quen tial

mag netic storms on set at 0900 UT on 12 Sep tem ber, 2000 UT

on 13 Sep tem ber, and 2200 UT on 14 Sep tem ber. Since ion -

o spheric elec tron den sity might sig nif i cantly de crease from

a few hours to two days af ter a SSC or storm on set (Kelley

1989; Davies 1990), the de crease anom a lies of TEC*, NmF2*,

and GIM-TEC* ob served in Ja pan on days 6 and 5 be fore

the Chi-Chi earth quake (15 and 16 Sep tem ber) shown in

Fig. 3 are most likely re lated to the re ported storms. Fol low -

ing the three storms dur ing 12 - 14 Sep tem ber, two geo mag -

netic SSCs oc curred at 0753 and 2025 UT on 15 Sep tem ber

(http://isgi.cetp.ipsl.fr/). Fig ure 3 re veals that all the obser -

vables, (ex cept GIM-TEC* at f30) sig nif i cantly de crease on

17 Sep tem ber (day 4 be fore the Chi-Chi earth quake). TEC*

val ues on 17 Sep tem ber of all sta tions in the Oki nawa-

Kyushu area also ex ceed -2s. These re sults strongly sug gest

that the de crease anom a lies are re lated to the storm on 15

Sep tem ber. By con trast, the re duc tion anom a lies ap pear only

at CHL and S105 in the Tai wan area on 18 Sep tem ber, day 3

be fore the Chi-Chi earth quake. It is sur pris ing to find that

GIM-TEC* in Tai wan does not sig nif i cantly de crease on the

day. Re call that to con struct a GIM, iso lated ground-based

Spa tial Ex tent of Ion o spheric Anom a lies of Tai wan Earth quakes 785

Fig. 5. The nor mal ized quan ti ties NmF2*, TEC*, and GIM-TEC*, as well as the geo mag netic Kp in dex, Dst in dex, and M ³ 5.0 seis mic ac tiv ity dur -

ing the pe riod be tween the Chi-Chi and Chia-Yi earth quakes from 27 Sep tem ber to 15 Oc to ber 1999.

GPS re ceiv ers scat tered world wide are needed. The TECs

de rived from data recorded at the iso lated re ceiv ers are fur -

ther in ter po lated by means of the spher i cal har mon ics to

ob tain the GIM. There fore, in a place with out iso lated GPS

re ceiver, the GIM-TEC is ob tained by in ter po la tion, which

might in tro duce in ac cu racy. The list (ftp://ftp.unibe.ch/aiub

/CODE/1999/P1P29909_ALL.DCB.Z) in Sep tem ber 1999

shows that the iso lated re ceiv ers near Tai wan are at PIMO

(14.6°N, 121.1°E), GUAM (13.6°N, 144.9°E), WUHN

(30.5°N, 114.4°E), DAEJ (36.4°N, 127.4°E), and TSKB,

which, in fact, are all rather far from Tai wan. Nev er the less,

sig nif i cant re duc tions ap peared in NmF2* at CHL and TEC*

at S105, in side Tai wan but not in TEC* at TSKB and 0497,

GIM-TEC* at KKBg, and NmF2* at KKB. On the other

hand, in the Oki nawa-Kyushu area, TEC* at Haterumajima

(0751; 24.1°N, 123.8°E) ex ceeds the -2s thresh old. The

TEC* val ues at the other sta tions in the west ern Oki nawa

area also de crease to about -2s. This in di cates that the

anom a lies on 18 Sep tem ber (day 3 be fore) may be as so ci -

ated with the Chi-Chi earth quake.

Fig ure 5 in di cates that the TEC* de creased at S105 on 1

Oc to ber. Ac cord ing to Liu et al. (2006), ion o spheric elec tron

den sity should sig nif i cantly de crease within days 1 - 5 be -

fore M ³ 5.0 earth quakes in Tai wan. There fore the anom aly

at S105 on 1 Oc to ber is un likely re lated to the Chia-Yi earth -

quake on 22 Oc to ber 1999. How ever it is in ter est ing to find

that four mod er ate earth quakes, M 5.12, M 5.3, M 5.05, and

M 5.01 oc curred on 1, 3, 4, and 5 Oc to ber 1999, which are 0,

2, 3, and 4 days af ter the anom aly, re spec tively. More over,

for these days, the de crease in TEC* at TSKB, NmF2* at

KKB, and GIM-TEC* at all ob ser va tion points did not ex -

ceed the -2s thresh old. Since geo mag netic con di tions are

786 Nishihashi et al.

Fig. 6. The NmF2, TEC, GIM-TEC, Kp in dex, Dst in dex, and M ³ 5.0 seis mic ac tiv ity dur ing the Chia-Yi earth quake pe riod from 16 to 28 Oc to ber

1999.

rel a tively quiet dur ing this pe riod, the anom aly ap pear ing on

1 Oc to ber is a lo cal phenomenon and possibly related to the

four earthquakes.

Dur ing the Chia-Yi earth quake pe riod, the re duc tion

anom a lies on 19 and 21 Oc to ber are reg is tered by NmF2* at

CHL and TEC* at S105 and in the west ern Oki nawa area,

but not by NmF2* at KKB, TEC* at TSKB nor in the east ern

Oki nawa and Kyushu area, nor for GIM-TEC* at six points

Spa tial Ex tent of Ion o spheric Anom a lies of Tai wan Earth quakes 787

Fig. 7. The nor mal ized quan ti ties NmF2*, TEC*, and GIM-TEC*, as well as the geo mag netic Kp in dex, Dst in dex, and M ³ 5.0 seis mic ac tiv ity dur -

ing the Chia-Yi earth quake pe riod from 16 Oc to ber to 28 Oc to ber 1999.

(CHLg, KKBg, f30, f120, f210, and f300). Once again, no

anom aly in the GIM-TEC* at f120 might have re sulted from

the fact that no GPS re ceiver in Tai wan was in cluded in the

list of GPS re ceiv ers for the com pu ta tion of the GIM in

Oc to ber 1999 (ftp://ftp.unibe.ch/aiub/CODE/1999/P1P29910_

ALL.DCB.Z). Nev er the less, the spa tial anal y sis in di cates

that the anom a lies in Tai wan on days 3 and 1 be fore the

Chia-Yi earth quake (19 and 21 Oc to ber) are lo cal phe no -

mena around Tai wan.

The dis tance be tween Tai wan and To kyo Ja pan is about

2200 km. The anom a lies of the Chi-Chi, Chia-Yi, and the

M 5 earth quakes can only be ob served in the Tai wan and

west ern Oki nawa area, which sug gests that the ex tent of the

reach of the dis tur bances due to these earth quakes is shorter

than 2200 km. Dobrovolsky et al. (1979) pro pose that the

radius R of earth quake prep a ra tion area, in which pre cur -

sory sig nals are ob served, can be given as R = 100.43M, where

M is the earth quake mag ni tude. For the larg est event in

this study, the Chi-Chi Mw 7.6 earth quake, we ob tain R =

1854 km. There fore, both the ob ser va tion and the o ret i cal

estimation in di cate that a spa tial com par i son of data si -

multaneously re corded at Tai wan and To kyo may be used

to iden tify ion o spheric anom a lies as so ci ated with Tai wan

earth quakes.

Ta ble 2 sum ma rizes the cor re la tions be tween the TEC*,

the GIM-TEC*, and the NmF2*. It is found in com par i son

with Ja pan that there is a lack of good cor re la tion be tween

TEC* and GIM-TEC* in Tai wan dur ing the pe ri ods of the

Chi-Chi and the Chia-Yi earth quakes. The poor cor re la tion

is cer tainly be cause there were no GPS sta tions in the Tai -

wan area for con struct ing the GIM in 1999. How ever, the

cor re la tion should be come even worse, when the anom a lies

also ap pear dur ing the two earth quake pe ri ods. In fact, the

poor cor re la tion of Tai wan in Ta ble 2 vice versa sug gests

that the ion o sphere was highly dis turbed prior to the 1999

Chi-Chi and Chia-Yi earth quakes and that such disturbances

were local phenomena.

The spa tial anal y sis in this study in di cates that the re -

duc tion anom a lies on day 3 be fore the Chi-Chi earth quake,

and days 3 and 1 be fore the Chia-Yi earth quake may be seis -

mi cally re lated. This find ing is in agree ment with the re sults

re ported by Liu et al. (2000, 2001, 2004a) and Chuo et al.

(2001, 2002). The spa tial ex tents of these anom a lies did not

reach to the mid-lat i tude (i.e., the To kyo and Kyushu area),

ap peared only in the Tai wan and west ern Oki nawa area in

the EIA re gion. This re sult sup ports that the EIA crest moves

equatorward as re ported by Liu et al. (2001). On the other

hand, the sig nif i cant NmF2 and TEC re duc tions on 17 Sep -

tem ber (4 days be fore the Chi-Chi earth quake) ap peared not

only in side the Tai wan and west ern Oki nawa area but also

out side (i.e., the east ern Oki nawa, Kyushu, and To kyo area),

in di cat ing that the re duc tion anom a lies were pos si bly af -

fected by a geo mag netic storm on 15 Sep tem ber (i.e., global

ef fect or non-earth quake re lated). In con clu sion, this study

shows that spa tial anal y sis could be use ful in iden ti fy ing

earth quake re lated anom a lies.

Ac knowl edge ments The au thors wish to thank Koichi C.

H. Chen in In sti tute of Space Sci ence, Na tional Cen tral

Uni ver sity for sup port ing this work and Prof. W. M.

Boerner of the Uni ver sity of Il li nois at Chi cago for use ful

com ments. The GPS data at S105 were ob tained from the

Min is try of In te rior of Tai wan, the GPS data at TSKB were

ob tained from the “Scripps Or bit and Per ma nent Ar ray

Cen ter” (SOPAC), and GEONET GPS data were pro vided

by the Geo graph ical Sur vey In sti tute of Ja pan. The NmF2

datasets at CHL were re trieved from the Chung-Li ion o -

spheric ob ser va tory and the NmF2 data at KKB were pro -

vided by WDC for Ion o sphere, To kyo, Na tional In sti tute of

In for ma tion and Com mu ni ca tions Tech nol ogy. The GIM

data were iso lated from the Cen ter for Or bit De ter mi na tion

in Eu rope (CODE). The Dst in dex and the Kp in dex were

pro vided by world data cen ter (WDC) for Geo mag ne tism,

Kyoto. This re search has partly been sup ported by a Grant-

in-Aid for Sci en tific Re search of the Ja pan So ci ety for the

Pro mo tion of Sci ence (JSPS) (16560360) and by the Ja -

pan- Tai wan Joint Re search Pro gram of the In ter change As -

so ci a tion, Ja pan.

REF ER ENCES

Birfeld, Y. G., 1974: Seismo-ion o spheric cou pling. Ap pli ca tion

to ion o spheric earth quake pre dic tion. In: Osimi, M. (Ed.),

Search for Earth quake Pre cur sors at Prog nos tic Net works,

Nauka, Mos cow, 200-202.

Calais, E. and S. Amarjargal, 2000: New con straints on cur rent

de for ma tion in Asia from con tin u ous GPS mea sure ments

at Ulan Baatar, Mon go lia. Geophys. Res. Lett., 27, 1527-

1530, doi: 10.1029/1999GL005444. [Link]

Chang, C. H., Y. M. Wu, T. C. Shin, and C. Y. Wang, 2000: Re -

lo ca tion of the 1999 Chi-Chi earth quake in Tai wan. Terr.

Atmos. Ocean. Sci., 11, 581-590.

Chen, Y. I., J. Y. Liu, Y. B. Tsai, and C. S. Chen, 2004: Sta tis ti cal

tests for pre-earth quake ion o spheric anom aly. Terr. Atmos.

Ocean. Sci., 15, 385-396.

Chuo, Y. J., Y. I. Chen, J. Y. Liu, and S. A. Pulinets, 2001: Ion o -

spheric foF2 vari a tions prior to strong earth quakes in Tai -

wan area. Adv. Space Res., 27, 1305-1310, doi: 10.1016/

S0273-1177(01)00209-5. [Link]

Chuo, Y. J., J. Y. Liu, S. A. Pulinets, and Y. I. Chen, 2002: The

ion o spheric per tur ba tions prior to the Chi-Chi and Chia-Yi

earth quakes. J. Geodyn., 33, 509-517, doi: 10.1016/

S0264- 3707(02)00011-X. [Link]

Datchenko, E. A., V. I. Ulomov, and C. P. Chernyshova, 1972:

Elec tron den sity anom a lies as the pos si ble pre cur sor of

Tashkent earth quake. Dokl. Uz bek. Acad. Sci., 12, 30-32.

Davies, K., 1990: Ion o spheric Ra dio, Pe ter Peregrinus Ltd.,

Lon don, 580 pp.

Dobrovolsky, I. P., S. I. Zubkov, and V. I. Miachkin, 1979: Es ti -

788 Nishihashi et al.

ma tion of the size of earth quake prep a ra tion zones. Pure

Appl. Geophys., 117, 1025-1044, doi: 10.1007/BF0087

6083. [Link]

Gokhberg, M. B., A. V. Ustov, V. A. Liperovaskiy, R. K. H.

Liperovskaya, E. P. Kharin, and S. L. Shalimov, 1989: Per -

tur ba tions of the ion o spheric F-layer prior to strong earth -

quakes. Izvestiya, Earth Phys., 24, 254-260.

Hayakawa, M. (Ed.), 1999: At mo spheric and Ion o spheric Elec -

tro mag netic Phe nom ena with Earth quakes, Terra Sci. Pub.

Co., To kyo, 996 pp.

Hayakawa, M. and Y. Fujinawa (Eds.), 1994: Elec tro mag netic

Phe nom ena Re lated to Earth quake Pred i ca tion, Terra Sci.

Pub. Co., To kyo, 667 pp.

Hayakawa, M. and O. A. Molchanov (Eds.), 2002: Seismo

Elec tromagnetics: Litho sphere-At mo sphere-Ion o sphere

Cou pling, TERRAPUB, To kyo, 477 pp.

Hunscucker, R. D., 1991: Ra dio Tech niques for Prob ing the

Ter res trial Ion o sphere, Springer, Berlin, Ger many.

Kelley, M. C., 1989: The Earth’s Ion o sphere, Ac a demic Press,

San Diego, 487 pp.

King, R. W. and Y. Bock, 2004: Doc u men ta tion for the GAMIT

GPS Anal y sis Soft ware, Re lease 10.2, Mas sa chu setts In -

sti tute of Tech nol ogy and Uni ver sity of Cal i for nia at San

Diego, Cam bridge.

Leick, A., 1995: GPS Sat el lite Sur vey ing, John Wiley, New

York, 560 pp.

Liu, J. Y., H. F. Tsai, and T. K. Jung, 1996: To tal elec tron con -

tent ob tained by us ing the global po si tion ing sys tem. Terr.

Atmos. Ocean. Sci., 7, 107-117.

Liu, J. Y., Y. I. Chen, S. A. Pulinets, Y. B. Tsai, and Y. J. Chuo,

2000: Seismo-ion o spheric sig na tures prior to M ³ 6.0 Tai -

wan earth quakes. Geophys. Res. Lett., 27, 3113-3116, doi:

10.1029/2000GL011395. [Link]

Liu, J. Y., Y. I. Chen, Y. J. Chuo, and H. F. Tsai, 2001: Vari a tions

of ion o spheric to tal elec tron con tent dur ing the Chi-Chi

earth quake. Geophys. Res. Lett., 28, 1383-1386, doi:

10.1029/2000GL012511. [Link]

Liu, J. Y., Y. J. Chuo, S. A. Pulinets, H. F. Tsai, and X. Zeng,

2002: A study on the TEC per tur ba tions prior to the Rei-Li,

Chi-Chi and Chia-Yi earth quakes. In: Hayakawa, M. and

O. A. Molchanov (Eds.), Seismo Electromagnetics: Litho -

sphere-At mo sphere-Ion o sphere Cou pling, TERRAPUB,

To kyo, 297-301.

Liu, J. Y., Y. J. Chuo, S. J. Shan, Y. B. Tsai, Y. I. Chen, S. A.

Pulinets, and S. B. Yu, 2004a: Pre-earth quake ion o spheric

anom a lies reg is tered by con tin u ous GPS TEC mea sure -

ments. Ann. Geophys., 22, 1585-1593.

Liu, J. Y., Y. I. Chen, H. K. Jhuang, and Y. H. Lin, 2004b: Ion o -

spheric foF2 and TEC anom a lous days as so ci ated with M

³ 5.0 earth quakes in Tai wan dur ing 1997-1999. Terr. Atmos.

Ocean. Sci., 15, 371-383.

Liu, J. Y., Y. I. Chen, Y. J. Chuo, and C. S. Chen, 2006: A sta tis -

ti cal in ves ti ga tion of pre-earth quake ion o spheric anom aly.

J. Geophys. Res., 111, A05304, doi: 10.1029/2005JA

011333. [Link]

Ma, K. F., T. R. A. Song, S. J. Lee, and H. I. Wu, 2000: Spa tial

slip dis tri bu tion of the Sep tem ber 20, 1999, Chi-Chi, Tai -

wan, earth quake (MW7.6) - in verted from teleseismic data.

Geophys. Res. Lett., 27, 3417-3420, doi: 10.1029/2000

GL011393. [Link]

Miyazaki, S., T. Saito, M. Sasaki, Y. Hatanaka, and Y. Iimura,

1997: Ex pan sion of GSI’s na tion wide GPS ar ray. Bull.

Geogr. Surv. Inst., 43, 23-34.

Nestorov, G. T., 1979: A pos si ble ion o spheric pres age of the

Vrancha earth quake of March 4, 1977. Doklady (Trans ac -

tions) Bul gar ian Acad. Sci., 32, 443-446.

Pulinets, S. A., 1998: Seis mic ac tiv ity as a source of the ion o -

spheric vari abil ity. Adv. Space Res., 22, 903-906, doi:

10.1016/S0273-1177(98)00121-5. [Link]

Pulinets, S. A. and K. A. Boyarchuk, 2004: Ion o spheric Pre cur -

sors of Earth quakes, Springer, Berlin, Ger many, 316 pp.

Pulinets, S. A., A. D. Legen’ka, and V. A. Alekseev, 1994: Pre-

earth quake ion o spheric ef fects and their pos si ble mech a -

nisms. In: Kikuchi, H. (Ed.), Dusty and Dirty Plas mas,

Noise, and Chaos in Space and in the Lab o ra tory, Ple num

Pub lish ing, New York, 545-557.

Ratcliffe, J. A., 1972: An In tro duc tion to the Ion o sphere and

Mag neto sphere, Cam bridge Uni ver sity Press, Lon don.

Sardon, E., A. Rius, and N. Zarraoa, 1994: Es ti ma tion of the

trans mit ter and re ceiver dif fer en tial bi ases and the ion o -

spheric to tal elec tron con tent from global po si tion ing sys -

tem ob ser va tions. Ra dio Sci., 29, 577-586, doi: 10.1029/

94RS00449. [Link]

Schaer, S., 1999: Map ping and pre dict ing the Earth’s ion o -

sphere us ing the Global Po si tion ing Sys tem. Ph.D. The sis,

As tro nom i cal In sti tute, Uni ver sity of Bern, Bern, Swit zer -

land.

Shin, T. C., K. W. Kuo, W. H. K. Lee, T. L. Teng, and Y. B. Tsai,

2000: A pre lim i nary re port on the 1999 Chi-Chi (Tai wan)

earth quake. Seismol. Res. Lett., 71, 24-30.

Shiokawa, K., Y. Otsuka, T. Ogawa, N. Balan, K. Igarashi, A. J.

Ridley, D. J. Knipp, A. Saito, and K. Yumoto, 2002: A

large-scale trav el ing ion o spheric dis tur bance dur ing the

mag netic storm of 15 Sep tem ber 1999. J. Geophys. Res.,

107, 1088, doi: 10.1029/2001JA000245. [Link]

Spa tial Ex tent of Ion o spheric Anom a lies of Tai wan Earth quakes 789