478 MONTHLY WEATHER REVIEW VOl. 99, No. 6
TROPICAL CYCLONE EYE RE-FORMATION AFTER A 30-HOUR .MOVEMENT
UDC 551.516.23(287) (691)
OVER THE MALAGASY REPUBLIC
CADESMAN POPE, JR.
Analysis and Forecast Division, National Weather Service, NOAA, Suitland, Md.
ABSTRACT
An analysis of the life cycle of a tropical cyclone in the Southern Hemisphere is presented. The recurving cyclone
of hurricane intensity moved over the Malagasy Republic, and its well-developed eye disappeared. After moving back
over warm oceanic water in which the temperature was 2So-29OC, a large cloud-free vortex center was observed by
satellite within 30 hr. This center or eye became increasingly well-formed until the storm became extratropical. Some
possible conclusions are discussed.
1. INTRODUCTION
A quasi-steady-state tropical cyclone of hurricane in-
tensity born in the Indian Ocean moved west-southwest-
ward through the Ocean Islands, making landfall just
south of Mahanoro, Malagasy Republic (19.8'S, 48.8'E).
Recurvature to the south took the center of the storm over
the rough east-central interior for approximately 30 hr,
during which interval the eye (well-formed before landfall
as seen by satellite photographs) disappeared. The
storm completed recurvature; and within 30 hr after
moving back over sea water in which the temperature was
28'-29'C, the storm formed a large cloudless center.
Satellite photographs during the next 48 hr showed
the clouds (surrounding the cloud-free center) exhibiting
more of the characteristics the storm had before it moved
onshore. However, the cyclone never did quite attain
its earlier appearance before it became extratropical.
Available upper air data suggest the cyclone was main-
tained aloft in the mid-troposphere while over land and
redeveloped its lower structure over a warm-surface
oceanic heat source and under anticyclonic conditions in
the upper troposphere.
begun. By 1200 GMT on February 10, the cyclone was
located near 17.5'S, 76'E. Satellite photographs indicated
that it had reached hurricane intensity. A satellite
weather bulletin for 1022 GMT on February 10 described
the storm-stage X, diameter of 2.5', category 4, well-
defined eye. This diameter and category gave an estimated
maximum wind speed (MWS) of 85-90 k t (Hubert and
Timchalk 1969).
A generally west-southwest movement persisted until
recurvature on February 23-24. No less than five other
tropical cyclones of varying intensity were in evidence
in the tropical Indian and Pacific (Coral Sea) Oceans
south of the Equator at the birth and during the life
of the cyclone under study. The high water mark of trop-
ical cyclone activity was reached contemporaneously
with this storm. A satellite photograph (fig. 1) taken
above the Indian Ocean at 0929 GMT on February 17
shows the unnamed cyclone near IS'S, 67.5'E, and cyclone
9. DEVELOPMENT AND MOVEMENT
OF THE CYCLONE
The tropical cyclone was first observed by satellite
photographs in the Indian Ocean near 16'S, 81°E, at
0000 GMT on Feb. 8, 1970. It was so indicated on the trop-
ical strip surface maps (2OoN-2O0S, O0-18O0EW) now
being analyzed routinely for 0000 and 1200 GMT by the
Southern Hemisphere and Tropical Analysis Section of the
Surface Analysis Branch of the Analysis and Forecast
Division (A&FD), National Meteorological Center
The cyclone formed over the warm Indian Ocean in
easterlies poleward of the intertropical convergence zone
(ITCZ) located near and parallel to 9's. During the period
of formation and deepening, the cyclone remained es-
sentially stationary; but by 1200 GMT on February 9.
satellite data showed that a westward movement had
(fiMC).
FIGURE 1.-ESSA 9 photograph of the unnamed cyclone (to the
the left) and cyclone Judy (to the right) at 0929 QMT on Feb. la,
1970. 1 After re-formatlon, the eye is defined in this paper as a cloud-free vortex center.
June 1971 Cadesman Pope, Jr. 479
Judy near 15'S, 88'E. The ITCZ was near 9's (to the
north of the two cyclones). The satellite classification
system indicated that the cyclone had maximum winds
of 95-100 kt and that Judy had maximum winds of 95 k t
(Hubert and Timchalk 1969).
The cyclone (near 18OS, 56.5'E, a t 1200 GMT on Febru-
\I .. . . . . . . . I .. . . . .?. . . I
I O i .6 . . .IO16 . . . . . . . . .
I : :: :::::: 1 : ::::::::I __ CIAO
FIGURE 2.-Surface chart for 1200 GMT on Feb. 20, 1970; pressure
(mb), temperature ("C), wind (kt), and ITCZ (1-1).
ary 20) was moving through the Ocean Islands in a
west-southwest direction a t 11 kt. Peripheral ship data
(fig. 2) indicated that the storm was quite small, with the
area of hurricane-force winds probably not more than 50
to 70 n.mi. in diameter. An ESSA 9 photograph (fig. 3)
taken at 1128 GMT on February 21 showed the cyclone
(centered near 19'S, 52.5OE) still moving west-southwest
at 11 kt. Only a slight increase in size and intensity had
occurred in 72 hr, indicating that a quasi-steady state
had been attained. The satellite weather bulletin for 1128
GMT on February 21 described the intense cyclone-stage
X, diameter 4, category 4. These criteria indicated a
maximum wind speed of 105 kt. The eye (fig. 3), with a
diameter of approximately 30 n.mi., was clearly discernible.
3. LANDFALL, RECURVATURE, AND
DISAPPEARANCE OF THE EYE
From 1200 GMT on February 21 until landfall between
0000 and 1200 GMT on February 23, the cyclone exhibited
a steady deceleration in its foIward movement. The eye
of the storm (fig. 4) was still clearly visible a t 1012 GMT
on the 22d when the center was near 2OoS, 50'E, or ap-
proximately 75 n.mi. east of Mahanoro. The weathei sta-
tion there was reporting winds (fig. 5 ) of only 25 k t with
the center of the storm in close proximity, indicating again
the smallness of the area of hurricane winds. An estimate
of maximum winds would be 80-85 kt, n slight decrease in
24 hr, possibly due to the interference of the rough
Malagasy landmass with the western edge of the storm.
The satellite photograph (fig. 6) at 1108 GMT on Febru-
ary 23 no longer showed an eye. An apparent center of
cyclonic circulation and the densest cloud mass was near
2OoS, 48'E, or 30 n.mi. west of Mahanoro. The cyclone
had moved onshore and had weakened. No surface 01
satellite information was available for 0000 GMT on Feb-
FIGURE 3.-ESSA 9 view of the unnamed cyclone near the Malagasy
Republic at 1128 GMT on Feb. 21, 1970; MWS, 105 kt (Hubert
and Timchalk 1969).
t
&13-%10 0-71-3
FIGURE 4.-ESSA 9 photograph of the unnamed cyclone near the
Malagasy Republic at 1012 GMT on Feb. 22, 1970; MWS, 80-85
k t (Hubert and Timchalk 1969).
480 MONTHLY WEATHER REVIEW VOI. 99, No. 6
......... .........
......... .........
I 30"
FIGURE s.-surface chart for 1200 GMT on Feb. 22, 1970.
......... .........
1 . . .8. .. ............
FIGURE 7.Surface chart for 1200 GMT on Feb. 24, 19719.
FIGURE 6.-ESSA 9 view of the unnamed cyclone over the Malagasy
Republic at 1108 GMT on Feb. 23, 1970.
ruary 24. Ey 1200 GMT on the 24th, surface and satellite
data (figs. 7 and 8) located the center near 25'S, 45'E, or
approximately 50 n.mi. northwest of Faux-Cap (25.5OS,
45.6'E). Recurvature to the south had begun after 1200
GMT on the 23d, as depicted by the track (fig. 9). The storm,
after traversing the rough east-central interior of the island
FIGURE 8.-ESSb 9 photograph of the unnamed cyclone over the
Malagasy Republic at1207 GMT on Feb. 24,1970. No eye is visible;
and the cloud band, extending southeastward from the center of
the cyclonic circulation, may be frontal in nature but could
be associated with a high-level jet.
(Madagascar) was beginning to emerge from the landmass.
Cyclonic cloud circulation was apparent, but no eye could
be detected. Surface data (fip. 7) showed a weakened but
June 1971 Cadesman Pope, Jr. 481
lmbl
100 -19 A 6 1 --J9 -681 1108 .I) 1614
-0
1 5 0 -6 5 1421 2' -621 I4181 -591 zo
200 -521 I216
'g
-411 12411 6>
-361 1113 L L I'
-34 1110 6 -3 6 1 1 0 1
-41 r I -36 1 250 -4 0 1 1160
FIQURE lO.-Vertical time section at Tananarive, Malagasy Repub-
lic; height [dekameters (dam)], temperature ("C), and wind (kt).
FIQURE 9.-Topographic map (ft) of the Malagasy Republic, in-
cluding 12-hr positions of the unnamed cyclone and sea water
temperature ("C) from ship data for February 1970.
definite cyclonic surface circulation. Over the ocean south-
east of the center, squalls and 25- to 35-kt winds were oc-
curring. It is possible that the course of the storm was
closer to the coast than figure 9 indicates, but the break-
down of thc organization of the central cloud mass (fig. 8)
indicated that the storm core was effectively removed
from its surface energy source.
4. UPPER AIR OBSERVATIONS NEAR THE CYCLONE
BEFORE, DURING, AND AFTER LANDFALL
The only consistently available upper air observations
over the Malagasy Republic near the cyclone track were
taken at Tananarive (18.8OS, 47.5OE, at an elevation of
4,147 ft). Figure 10 shows the vertical time section of the
observations available before, during, and after the passage
of the cyclone. Unfortunately, radiosonde observations
were unavailable a t the time the storm was nearest to the
station; but useful information was obtained from the
observed upper winds.
A general warming trend in the soundings occurred
between 0000 GMT on February 21 and 22. Thereafter,
the soundings remained relatively unchanged except
for cooling at 700 mb at 0000 GMT on the 23d, probably
due to precipitation associated with the storm. Tropo-
pause heights were difficult to ascertain from the data
available, but they appeared to be near 100 mb. Apparent
outflow aloft was evident as early as 0000 GMT on the 22d
and continued to appear at 1200 GMT on the same day.
The level of change aloft from a cyclonic circulation to
anticyclonic was near or above 200 mb.
The upper winds observed at 0000 GMT on February 23,
when the cyclone was approximately 120 ami. from the
station, revealed that the strongest cyclonic circulation
aloft was between 700 and 300 mb. At 1200 GMT on the
23d when the storm was at its nearest location to the
station (approx. 90 n.mi.), the strongest cyclonic circula-
tion aloft was between 700 and 500 mb.
5. SOME POSSIBLE CONCLUSIONS FROM UPPER
AIR DATA AND OTHER CONSIDERATIONS
From the winds aloft observed at 0000 and 1200 GMT
on February 23, it appears that the strongest height
gradients aloft at the 90-n.mi. radius in the lower tropo-
sphere were near 600 mb or 12,000-14,000 ft. This was
approximately 6 hr after the storm had moved inland
and was at its closest proximity to Tananarive.
Since height gradients are known to depend upon
thermal gradients and since cooling in the core of the
cyclone causes a decrease in height gradients (Miller
1963), it appears that the cyclone did not cool as much in
the 700- to 500-mb layer as below 700 mb in the inflow
layer.
I t is postulated that the height gradients of the central
core of the cyclone in the layer between 700 and 500 mb
were maintained dynamically during the land traverse.
This cannot be proven because of insufficient data. Sub-
sequent events would indicate that large upward vertical
motions were maintained near the storm center at some
upper level.
When assuming that this was the case, several factors
were apparent. One was the tropospheric environment
in which the cyclone existed. This was pure maritime
tropical air with 850- to 500-mb thicknesses of at least 438
dekameters (dam) or 1000- to 500-mb thicknesses of
579 dam. Another was the state of the summer hemisphere
oceanic at mospher e-one of maximum instability . The
482 MONTHLY WEATHER REVIEW vos. 99, No. 6
FIGURE 11.-ESSA 9 photograph of the unnamed cyclone as it
appeared near the Malagasy Republic at 1110 GMT on Feb. 25,
1970. Note the enlargement of the cloud (banding east and south-
east of the center) and a well-defined frontal band extending
nearly to 35'5.
large number of contemporaneous tropical storms of
varying degrees of intensity gave evidence of this. Finally,
a NMC computer-produced 250-mb analysis (not shown),
the only upper air chart available to the author, indicated
anticyclonic conditions over the cyclone in the upper
troposphere.
As the storm moved back over the warm oceanic
surface energy source (fig. 9), it is postulated that the
maximum upward vertical motions retained in the 700-
to 500-mb layer near the core interacted with the surface
heat source. A consequent redevelopment of the inflow
layer (Miller 1963) and of surface gradients was effected
(figs. 11-13), and a large cloud-free cyclonic center was
re-formed.
Some cooling of the upper core very likely took place
while the cyclone was over land. This cooling may have
led to a spreading out as well as a reduction in height
gradients. Such expansion may have been responsible for
the large size of the re-formed center.
A similar event was possible with the redevelopment of
the eye of hurricane Cleo in AugustSeptember 1964.
Cleo spent, almost 3 days over the interiors of North Caro-
lina and South Carolina (losing much of its surface organi-
zation) before moving back over an oceanic heat source
and redeveloping an eye (US. Weather Bureau 1964).
. RE-FORMATION OFTHE EYE
The remnants of the cyclone (fig. 9) moved over water
(in which the surface temperature was 28O-29OC) some-
time between 1200 GMT on February 24 and 0000 GMT on
February 25. By 1207 GMT on the 24th, the center of the
1 FIGURE 12.-ESSA 9 view of the unnamed cyclone near the Mala-
gasy Republic at '1209 GMT on Feb. 26, 1970.
\"
, *:
f
I I
FIGURE 13.-Surface chart for 1200 GMT on Feb. 26, 1970.
storm (fig. 8) was near 25%, 45'E, with weak but definite
cyclonic organization of the clouds. The storm (fig. 11)
was centered near 27.5'5, %'E, at 1110 GMT on the 25th.
The area of organized cloud banding east and southeast of
the center had greatly enlarged and intensified in 24 hr.
According to the NMC computer-produced 250-mb
analysis (not shown), conditions a t that level were anti-
cyclonic over the area of eye re-formation.
Satellite photographs taken at 1209 GMT on February 26
disclosed that the cyclone (fig. 12) had re-formed a large
cloud-free center over 60 n.mi. in diameter near 27OS,
~
June 1971 Cadesman Pope, Jr. 403
FIGURE 14.-ESSA 9 photograph of the unnamed cyclone near the
Malagasy Republic at 1112 GMT on Feb. 27, 1970.
49'E. The clouds surrounding the center presented a more
compact organization than those observed 24 hr earlier;
however, the bright, compact presentation of the storm
before moving over land was lacking.
Some available ship data (fig. 13) at 1200 GMT on
February 26 showed sustained winds of 50 kt in the west-
ern and northern quadrants of the storm. A surface and
inflow layer pressure gradient (sufEcient for tropical storm
to hurricane intensity) had been regenerated. A front
moving in from the south had joined the main cloud area
south to southeast of the center, but no cold air had
entered the tropical cyclone circulation.
7. FINAL STAGES OF THE CYCLONE
After 1200 GMT on February 26, no surface data were
available near the cyclone; satellite data were relied on
exclusively. A satellite photograph (fig. 14) taken at 1112
GMT on the 27th showed a circular cloud mass surrounding
the large cloudless center near 27'S, 53'E. Only slight
changes werenoted from the picture taken 24 hr previously,
The frontal band continued to be evident south and
southeast of the cyclone. Some possible presence of cooler,
drier air appeared from the cumuliform cloud banding
around the western, northern, and eastern periphery of
the storm. Also, the dry tongue east of 55'E may have
indicated this, with a possible front forming along 60'E
(east of the dry tongue).
Satellite data for 1016 GMT on February 28 showed con-
tinued movement of dry air around the cyclone (fig. 15)
with its cloud-free center near 3OoS, 54'E. The diameter of
the eye was considerably smaller, less than 30 n.mi. The
cyclone, appearing more intense than 24 to 48 hr earlier,
was beginning to approach the appearance of the storm
before it moved over land. The dry tongue east of the
center along 60'E had continued to sweep cyclonically
FIGURE 15.-ESSA 9 view of the unnamed cyclone over the Indian
Ocean at 1016 GMT on Feb. 28, 1970.
FIGURE 16.-ESSA 9 photograph of the unnamed cyclone over the
Indian Ocean at 1114 GMT on Mar. 1, 1970.
around the eastern edge of the cyclone while the cloud
band or possible front east of the dry tongue had intenssed.
Photographs taken by satellite at 1114 GMT on March 1
gave evidence that the cyclone (fig. IS), centered near
37'S, 60%, was becoming extratropical. The cloud-free
center was no longer visible. The dry tongue east of the
storm was still apparent, as was the front between 65'
and 70'E. Although the eye was no longer apparent, the
484 MONTHLY WEATHER REVIEW Voi. 99, No. 6
FIGURE 17.-ESSA 9 view of the unnamed cyclone over the Indian
Ocean at 1019 GMT on Mar. 2, 1970.
cloud circulation continued to be intense. The storm had
accelerated southeastward, with a movement of 7’ lati-
tude in 24 hr, indicating that the storm was strongly
imbedded in the westerlies. Figure 17 at 1019 GMT on
March 2 is a classical picture of an extratropical cyclone.
The storm, centered near 4OoS, 70’E, was moving south-
eastward at an accelerated rate.
8. SUMMARY AND CONCLUSIONS
From the foregoing data, we have seen a quasi-steady-
state tropical cyclone of hurricane intensity leave its
oceanic energy source, move over a rough landmass,
weaken, and lose its eye. After approximately 30 hr over
land, it moved back over the ocean. Within 30 hr after
moving over sea water in which the average temperature
was 28’-29’6, a large cloud-free vortex with a compact
cloud organization was re-formed. Upper air data suggests
the possibility of the cyclone having been maintained
dynamically aloft while over land. With the surface
energy source again becoming available, the lower tropo-
spheric upward vertical motions apparently interacted
with the surface heat source with a consequent rejuvena-
tion of surface gradients and an eye. Upper tropospheric
conditions over the area of eye re-formation apparently
were favorable (anticyclonic) for redevelopment.
FACTORS INHERENT IN MIS EVENT AND IN SIMILAR EVENT$
1. Occurrences of this nature are most apt to happen when the
summer hemisphere oceanic atmosphere is a t its maximum in-
stability [Le., when sensible and latent heat available are a t a
maximum, as is the case in late summer (February through March
in the Southern Hemisphere; August through September in the
Northern Hemisphere)]. As noted earlier, this cyclone formed and
existed while five other tropical cyclones of varying degrees of
intensity were in existence-an indication of the instability of the
oceanic atmosphere. This is not to exclude early summer or fall
events-merely to point out a time of apparent maximum proba-
bility of occurrence.
2. The tropical cyclone remains enveloped in a maritime tropical
air mass (with 850- to 500-mb thicknesses of a t least 438 dam or
1000- to 500-mb thicknesses of 579 dam) while traversing the
landmass and during eye re-formation.
3. The storm is a well formed steady-state cyclone of at least
moderate hurricane intensity before moving over land.
4. While the storm is traversing the landmass, maximum cy-
clonic vorticity and upward vertical motions are maintained aloft
(probably near 600 mb) by some dynamic mechanism.
5. The shorter the time period the cyclone is over land (removed
from its surface energy source), the greater the chance of eye re-
formation.
6. Finally, of necessity, the storm must move back over water
a t or above the threshold temperature for the formation of tropical
cyclones (27OC, Palmen 1948). The higher the sea temperature,
the greater the chance is of eye re-formation. Favorable conditions
in the upper troposphere (neutral or anticyclonic) also enhance
eye re-formation.
ACKNOWLEDGMENTS
Thanks are due Messrs. Edwin Fawcett of A&FD and Gilbert
Jager of the National Environmental Satellite Service (NESS) for
their helpful comments and Mrs. Beatrice Durkin for typing the
manuscript.
REFERENCES
Hubert, Lester F., and Timchalk, Andrew, “Estimating Hurricane
Wind Speeds From Satellite Pictures,” Monthly Weather Review,
Vol. 97, No. 5, May 1969, pp. 382-383.
Miller, Banner I., “On the Filling of Tropical Cyclones Over Land,”
National Hurricane Research Project Report No. 66, U.S. Weather
Bureau, Miami, Fla., Dec. 1963, 82 pp.
Palmen, Erik Herbert, “On the Formation and Structure of Tropical
Cyclones,” Geophysica, Vol. 3, Helsinki, Finland, 1948, pp. 26-38.
U.S. Weather Bureau, Hurricane Cleo, August BO-September 4, 1964,
Preliminary Report With Advisories and Bulletins Issued,
Washington, D.C., 1964, 31 pp.
[Received June SO, 18’70; rewised September 16, 19701