326 BULLETIN AMERICAN METEOROLOGICAL SOCIETY

The Development and Motion of Typhoon "Doris," 1950

GEORGE P. CRESSMAN

Hq.y Air Service, Washington, D. C.

ABSTRACT The development and motion of typhoon "Doris," which was observed during the first two weeks of May, 1950, are studied. The development of the is examined with respect to pre- viously published theories of storm formation. The original deepening occurred in the low latitude portion of an extended trough, after the fracture of the trough. This is in agreement with a model proposed by Riehl. The motion of the deepening storm relative to the high-level flow patterns differed from previously studied examples in that the deepening occurred as the low-level moved from under the west side of an upper toward a position un- der an upper cyclone. The storm developed as two cyclonic vortices, which gradually merged into one, in agreement with a principle of Fujiwhara. The motion of the storm northward, as it broke through the subtropical ridge line, is shown. After examination of several possibilities, this motion is attributed to the resultant of all the Coriolis forces acting on the storm, as discussed by Rossby. The suggestion is made that this resultant force becomes prominent in determining the motion of the storm due to changes in the radial velocity profile and the increasing geographical extent of the storm.

"^HE development and movement of ty- southern hemisphere and the phoon "Doris," which occurred in the west easterlies. This is suggested by a west wind at Pacific in the first two weeks of May 1950, 10,000 feet at Honiara (09°S, 160°E) on 1 May, offer an opportunity for the testing of several new and is confirmed by the streamline chart of 2 May. itheorie s relating to tropical . There were By 03GCT, 2 May, the northern part of the west an unusually large number of upper-air observa- tions available for this part of the world, and a reasonably complete life history of the cyclone can be derived.

THE DEVELOPMENT STAGE It is convenient to trace the development by starting with the 10,000-ft streamline chart (FIG. 1) and the 700-rnb chart (FIG. 2) for 0300 GCT, 1 May 1950. The pattern of middle latitude long waves consists of a trough over Manchuria, a ridge east of and Sakhalin, a trough at about 167°E, and a ridge at about 170°W. As a result of the short wavelength from trough to trough in this pattern, the west Pacific trough was moving rapidly eastward. It was shown in a previous in- vestigation [1] that if the higher-latitude portion of an extended trough is moving rapidly east- ward, it will be unable to retain its tropical ex- FIG. 1. 10,000-ft streamlines, 0300 GCT, 1 May 1950. In this and subsequent charts winds from aircraft are in- tension. In the present example, then, the portion cluded from a total period of twelve hours, centered on of the extended trough which lies in the tropics the map time. No correction in position is made in the should break off from the middle latitude trough. plotting for aircraft wind reports computed between fixes. According to Riehl and Burgner [10] this frac- Pilot balloon reports six hours off map time are indicated ture produces a situation which is favorable for by a dashed shaft. Apparent discrepancies occasionally low-latitude . An additional factor seen between winds and streamlines arise primarily be- of some importance is the existence of a line of cause the winds are off time or represent mean values shear between cross-equatorial from the between fixes.

Unauthenticated | Downloaded 10/04/21 08:35 AM UTC VOL. 32, No. 9, NOVEMBER, 1951 327

ception of the one ascent on 15 GCT, 4 May, strengthening of the winds through 5 May. This interpretation of the Truk winds is supported by the -distribution charts. The altostratus deck over the Marshalls on 1 May moves toward the WSW as a relatively small patch until 3 May. A large extension of the area of towering cumulus in the Carolines area can be observed on the charts for 2 and 3 May. Then from 3 to 4 May the deck of altostratus increases greatly over the Carolines and over the route from Kwajalein to . At the same time, the cumulus activity over the Carolines decreases along with the in- creasing horizontal circulation. It is at 21 GCT on 3 May that a 45 knot wind is reported from Truk at 3000 feet. By 4 May continuous is

FIG. 2. 700-mb chart, 0300 GCT, 1 May 1950. In this and subsequent charts, locations where the height of the 700-mb (or other) constant pressure surface was re- ported are indicated by heavy dots.

Pacific trough had moved 13° longitude eastward. As far south as 15°N the trough moved at least 5° longitude toward the east, as shown by FIGURE 3. By 5 May (see FIG. 7) the northern part of the trough is just west of Wake and Midway, which it passes before 6 May. However, the southern end of the trough broke off from the eastward mov- ing part, beginning 2 May. FIG. 4. Time section and 24-hr pressure changes from The gradual formation of a cyclonic center near 1-6 May 1950, for Truk. Truk (07°N, 152°E) is shown by the time-section for Truk (FIG. 4) and from the cloud-distribution falling at Truk and Ponape, and a closed cyclonic charts (FIG. 5). In the time section it appears circulation is drawn near Truk on FIGURE 6. that after the weak trough passage at Truk on the On 5 May (FIG. 7) a flight from Guam to first, there is a gradual veering and, with the ex- Truk passed through a new, apparently secondary, center near 11°N, 148°E. The existence of the center is indicated by a cyclonic wind circulation and by an area of cumulonimbus and bad weather which the aircraft encountered near the center of the secondary circulation. The tendency for "troughing" in the streamlines on the Kwajalein to Guam route is particularly evident on 5 May. This suggests the beginning of a split in the anti- cyclone north of this area. Such a split was defi- nitely confirmed by a flight from to Wake on the next day. A sudden backing of the low-level winds at Truk is observed at about 00 GCT 6 May (How- ever, the surface and 1,000-ft winds are unrepre- sentative due to terrain effects). This observation might be suspected to be an error. However, the graph of 24-hr pressure change indicates a maxi- mum of falls late in the fifth, and the first thunder- FIG. 3. 10,000-ft streamlines, 0300 GCT, 2 May 1950. storm activity at Truk occurs early in the sixth.

Unauthenticated | Downloaded 10/04/21 08:35 AM UTC 328 BULLETIN AMERICAN METEOROLOGICAL SOCIETY

The streamline chart for 6 May (FIG. 8) is there- fore analyzed to show the original center moving eastward past Truk as the two centers begin to rotate about each other, as described by Fuji- whara [3] [4] and Haurwitz [15]. The separate identity of these centers cannot be maintained in the analysis past the sixth, as a result of the rapidly increasing circulation and a lack of sufficiently de-

FIG. 5d. Cloud distribution chart 0300 GCT, 4 May 1950.

FIG. 5a. Cloud distribution chart 0300, 1 May 1950. Vertical hatching indicates areas of towering cumulus (tops above 8,000 feet). Horizontal hatching indicates areas of broken or overcast altostratus. Dots indicate observation locations.

FIG. 5e. Cloud distribution chart 0300 GCT, 5 May 1950.

FIG. 5b. Cloud distribution chart 0300 GCT, 2 May 1950.

FIG. 6. 10,000 ft streamlines, 0300 GCT, 4 May 1950.

tailed data. On the sixth the surface wind at Truk remains from S to SE, and the 24-hr pressure falls continue, although at a slower rate. From this it can be concluded that the second center is moving southward in the area west of Truk. After 6 May the only further indication of two centers comes FIG. 5C. Cloud distribution chart 0300 GCT, 3 May 1950. from the Air Weather Service reconnaissance

Unauthenticated | Downloaded 10/04/21 08:35 AM UTC VOL. 32, No. 9, NOVEMBER, 1951 329

double eye might be a final remnant of the origi- nal double system of circulations which, through low-level convergence operating over a period of time, gradually merged into a single system. Other solutions are also possible, in the absence of more detailed data, but they would be more complicated than the one presented above. The conclusion however, would be the same in that a system originally containing several vortices gradually intensified into a single typhoon, with the double eye as a final indication of the original complications. The development of "Doris" to typhoon strength concurrently with the appearance and amalgama- tion of the two centers is a remarkable verification of views presented by Fujiwhara in 1923 [3] and again in 1937 [5]. In these two papers Fujiwhara FIG. 7. 10,000 ft streamlines, 0300 GCT, 5 May 1950. presented a theory proposing that small adjacent vortices with the same sense of rotation will tend flight Vulture Three Doris on 8 May. The re- to amalgamate, leading to the appearance of a port of the weather observer, CWO K. R. single large vortex. Walters,1 includes the following paragraph: Recently Riehl [11] has pointed out the impor- "It is believed that there were two eyes in tance of the high-level flow pattern for the deepen- Doris at the time of this flight, a small one lying ing of tropical cyclones. The data available for approximately 30 miles southwest of a larger one. this situation permitted the analysis of the 300-mb While the surface indicated to the chart for the period from 15 GCT, 2 May to 15 navigator and weather observer that the typhoon GCT, 7 May, which fortunately embraces the pe- center was slightly to the left of the aircraft head- riod during which the typhoon developed. Since ing the radar observer found an eye 90 degrees to there were insufficient observations to indicate de- the right of the heading and this eye which we en- partures of the wind from the contours, the con- tered and took the fix was the smaller one." tours were drawn to fit the reported heights and The preceding analysis then suggests that the

1 Consolidated Report, "Doris," May 1950. Prepared by Typhoon Post Analysis Board, 15-2 Air Weather Service Detachment, Andersen Air Force Base, Guam, M. I.

FIG. 9. 300-mb chart, 1500 GCT, 5 May 1950. This chart was constructed with the use of (a) differential analysis from 500 mb, based on the available data and a series of correlations made by Major L. Garvin, Air Weather Service, and (b) time sections of height and FIG. 8. 10,000 ft streamlines, 0300 GCT, 6 May 1950. wind from all stations in the tropics and subtropics.

Unauthenticated | Downloaded 10/04/21 08:35 AM UTC 330 BULLETIN AMERICAN METEOROLOGICAL SOCIETY winds, and may be regarded as streamlines, even at low latitudes. From a comparison of the 10,000-ft streamline charts for 1-6 May with the 300-mb chart for 5 May (FIG. 9) it can be seen that the typhoon develop- ment began at a position midway between a 300-mb cyclone and a 300-mb anticyclone, under a flow which was from a southerly direction. Both cy- clone and anticyclone moved at first toward E and then toward ESE as the low-level cyclonic cir- culation increased throughout the period. Thus, with respect to the high-level pattern, the tropical cyclone slowly came from under the west side of a high-level anticyclone into an area under the in- fluence of a high-level cyclonic circulation. In this respect "Doris" seems unusual, its develop- ment being apparently of a different type than that discussed by Riehl. FIG. 11. 700-mb chart, 0300 GCT, 12 May 1950 On 2300 GCT, 6 May, an Air Weather Service reconnaissance plane left Guam for the storm area followed an oscillatory path, similar to the tro- and found surface winds of 70 knots north of the choid described by Yeh [10] [14]. Figure 2 in center, indicating that by this time the storm Horn's paper shows the track of the storm as de- could definitely be called a typhoon. termined from the observations. In pointing out It is interesting to note the low latitude at which the similarity to the theoretical tracks derived by the storm developed, most of the development hav- Yeh, Horn concludes, "It is now believed that in ing occurred at 5° or 6°N. the past a shadow has been unjustly cast on the validity of certain reconnaissance information, and MOTION OF THE STORM the accuracy of many good navigators has been The detailed motion of the storm is discussed in questioned through lack of knowledge of the true a short but excellent paper by Horn 2 in which he nature of typhoon movement." points out that the detailed information on the In addition to the oscillating motion he de- storm position, obtained by the 514th Weather scribes, the motion of the storm relative to the Reconnaissance Squadron, indicates that the storm larger scale circulation patterns is of interest. The 700-mb chart for 10 May (FIG. 10) shows a clear- 2 Horn, J. D., 1951: Movement of tropical cyclones in the Pacific. 2143d Air Weather Wing Technical Bulle- cut ridge extending from WSW to ENE north of tin, 1, 16-21; Bull. Amer. Met. Soc., v. 32, No. 9, Nov. the typhoon. The steering, if any, appears to be 1951, pp. 344-345. toward the west. Actually the motion of the storm from 10-11 May was slight, amounting to an average speed toward the NW of 7 knots. On the 11th (not shown) a weak opening in the ridge be- tween Formosa and Okinawa is evident. On the 12th the break in the ridge in this area is still evi- dent at 700 mb (FIG. 11). At 500 mb (map not reproduced) the ridge is weakest just north of the storm, which is moving due northward by this time. One might try to account for the north- ward motion of the storm by the concept of "high- level steering." However, from the 200-mb chart for 12 May (FIG. 12), it can be seen that the mo- tion of the typhoon is opposite to the flow at 200 mb. The steering hypothesis, low level or high level, must then be discarded in an explanation of the storm motion. The 700-mb charts for 13 (FIG. 13) and 14 May FIG. 10. 700-mb chart, 0300 GCT, 10 May 1950. (FIG. 14) show the typhoon as it breaks completely

Unauthenticated | Downloaded 10/04/21 08:35 AM UTC VOL. 32, No. 9, NOVEMBER, 1951 331

A number of recent papers report results show- ing that a cyclonic vortex such as a typhoon should be driven toward the north by various dynamical effects. Davies [2] neglects the effects of the altogether in the basic flow of the vortex. While for certain cases this is a good ap- proximation, the effect of variation of the Coriolis force, as indicated by Petterssen [9] and Rossby [12], is at its maximum in the low latitudes. Kuo [7] discusses the distribution in the sur- rounding fluid. This distribution is difficult to treat when the flow in the surroundings is as stag- nant as that shown in the charts from 6-14 May. Petterssen showed that a cyclonic center in which there is convergence will tend to have a northward component of motion, while divergence would re- sult in a southward component. In a typhoon, FIG. 12. 200-mb chart, 0300 GCT, 12 May 1950. This with low-level convergence balanced by high-level chart was constructed with the aid of time sections and with the aid of a vertical extrapolation system developed by Mr. H. S. Appleman, Air Weather Service. through the ridge. The chart for 14 May shows the ridge re-established again after the northward passage of the typhoon. This is not an isolated example of this kind of storm motion. The writer was informed in a verbal conference with some of the meteorologists at the Central Meteorological Observatory in Tokyo that the Japanese forecasters are very fa- miliar with this phenomenon, and do not hesitate to forecast some typhoons to break northward through the subtropical ridge line. An example familiar to American meteorologists is the notable of October 1944, which moved north- ward into an anticyclone. FIG. 14. 700-mb chart, 0300 GCT, 14 May 1950.

divergence, the resultant force from this effect should be small, and should be relatively invariant through most of the life history of the typhoon. This type of force should result in a tendency for the axis of the storm to tilt with height. The force on the cyclonic center according to Rossby is di- rected northward at all levels having cyclonic ro- tation. It can also vary considerably as the radial velocity profile of the storm changes with time. This type of force will be discussed in the following section, with particular reference to typhoons in the tropics. Rossby gives the northward force F acting on a unit slice of a vortex as /*R F = fipir I r*udr (1) FIG. 13. 700-mb chart, 0300 GCT, 13 May 1950. Jo

Unauthenticated | Downloaded 10/04/21 08:35 AM UTC 332 BULLETIN AMERICAN METEOROLOGICAL SOCIETY where /? is the northward rate of change of the due to the decrease in n, the value of C changes but Coriolis parameter, p is the density, r is the radius little during the growth of the storm. of curvature of particles moving in a vortex, R is Referring back to (5), it can be seen that as a the outer boundary of the vortex, and o> is the rela- typhoon matures and grows in intensity and ex- tive angular velocity of the air participating in the tent, the value of A will increase from near zero vortex. The force F is the resultant of the Coriolis to a significant northward acceleration. forces acting on the various parts of the vortex, and The velocity profile of a typhoon probably varies arises from the lack of balance between pressure from one quadrant to another as well as with the forces on the one hand and the sum of Coriolis and life history of the storm. Consequently, only the centrifugal forces on the other hand. most reserved type of conclusions can be drawn from As the first approximation for a typhoon we the above results. However, in order to illustrate can assume a velocity profile such that from the the results of variations of the various parameters center to a distance rlf where a maximum value of the following numerical applications are made. Ex- ro) is found, solid rotation exists, and = con- ample (1) is a hypothetical situation representing stant = C\. From r1 to R a vrn_1 vortex exists, a young immature storm. Examples (2) and (3) and wrn — Constant = C. At the distance rlf o)± = are computed from data gathered by weather re- to and C1 = C/rxn. Then connaissance flights of the Air Weather Service into "Doris" on 8 May and 12 May respectively. coi = C/rin, co = C/rn. Example l 2 3 Substitution of the above wind profile in (1) gives fmax (knots) 65 70 120 /•n nR 50 F = pPtt r'uidr + ppir I r*udr, (2) ri (nautical miles) 15 35 J0 Jri R (nautical miles) 150 300 600 which after integration yields Latitude 10° N 10° N 20° N

n 21 2 11

1.67 X103 2.45 X103 2.26 X103 As a rule, n will not be much different from two. C Since it is seldom observed that r1 exceeds R/10, A (knots per day) 1 4 17 an error of only about two percent will result from a simplication of (3) to From the above table one can conclude that the RA~n \ northward acceleration to which a typhoon is sub- jected will be small when the storm is intense, but (r=rn)- (4) covers only a small area. This is the stage of im- Since the mass M of the unit slice of the vortex is maturity as discussed by MacDonald [8] and pirR2, the northward acceleration A is James [6]. As the storm spreads out to cover a R2~n \ large area the northward acceleration becomes pro-

S nounced. The magnitudes of A in examples (2) (4^)' < > and (3) are only very approximate, but at least Various studies by Macdonald [8], U. S. indicate that significant accelerations can arise as Weather Bureau forecasters [13], and James [6] a storm matures. have suggested that n varies from values of slightly This type of storm, which attains dimensions greater than two for small immature systems to comparable to some of the larger high latitude cy- values near 3/2 for large mature typhoons. Thus clones, is the type which has a reputation for defy- the fraction R2_n/4 — n will grow from near-zero ing forecasters, particularly with its tendency to values to about 2R1/z/5 as the typhoon grows to move toward the north and destroy warm anticy- maturity. clones.3 The increase of the northward force is The constant C can be expressed in terms of the 8 In a lecture given at Headquarters, Air Weather maximum wind speed, found in the storm. Service in May 1950, Dr. H. Riehl pointed out that al- Since C — then C = r^'Hw. As a typhoon though small hurricanes can be expected to be steered well, large ones tend to move independently of the steer- grows from a small intense storm to a large ma- ing flow. This statement would be in agreement with the ture system, vmax and r1 both increase. However, above results.

Unauthenticated | Downloaded 10/04/21 08:35 AM UTC VOL. 32, No. 9, NOVEMBER, 1951 333

primarily due to the transformation of the radial some future investigation would be the compiling velocity profile and to the increase of the outer ex- of some statistics from the excellent weather re- tent of the vortex. connaissance data now available, in which the val- ues of n would be examined with respect to the CONCLUSIONS development stage, the size, and the sector of ty- The most significant features of the development phoons. In practice, the northward acceleration, of typhoon "Doris" appear to be : A, must be estimated qualitatively from the size and extent of the typhoon. Thus the forecaster (1) The existence of two centers during the will be required to forecast the future growth of development of the storm. This double system is the intensity and area of a typhoon in order to fore- followed by the double eye found by a later; weather cast the future track of the storm. reconnaissance flight, and is in agreement with a theory of storm development by Fujiwhara [3], REFERENCES (2) The development of the storm with respect [1] Cressman, G. P., 1948: Relations between high and to the high-level flow patterns. The storm de- low latitude circulations. Dep. Meteor. Univ. Chi- veloped as it moved from beneath a 300-mb anti- cago, Misc. Rep., No. 24: 65-100. [2] Davies, T. V., 1948: Rotary flow on the surface of cyclone to an area southeast of a 300-mb cyclone. the . Philosophical Mag., 39: 482-491. (3) The unusually low latitude (5° to 6°N) at [3] Fujiwhara, S., 1923: On the growth and decay of which the storm formed. vortical systems and on the mechanism of extra- tropical cyclones. Japanese Journal of Astronomy The most significant features of the motion of and Geophysics, I (5) : 125-182. the storm are: [4] Fujiwhara, S., 1931: Short note on the behaviour (1) The oscillations of the cyclone about its of two vortices. Proceedings of the Physical and Mathematical Society of Japan, Series 3, V. 13, mean path, discussed by Horn, which might be ex- No. 3. plained by Yeh's theory [14]. [5] Fujiwhara, S., 1937: On the quick development of (2) The motion of the storm to the north and cyclones by amalgamation. The Geophysical Mag- azine, XI (1) : 41-50. its breakthrough of the subtropical ridge. This [6] James, R. W., 1951: On the evolution of tropical northward motion is attributed to the resultant of cyclones. /. Meteor., 8(1): 17-24. the Coriolis forces acting on the storm, in accord [7] Kuo, H., 1950: The motion of atmospheric vortices and the general circulation. J. Meteor., 7 (4) : 247- with the principle of Rossby. The northward ac- 258. celeration of typhoons in general is shown to be- [8] Macdonald, W. F., 1942: On a hypothesis concern- come prominent as a result of the changes in the ing normal development and disintegration of tropi- cal hurricanes. Mon. Wea. Rev., 70 (1) : 1-7., also radial velocity profile and the increasing extent of Bull Amer. Meteor. Soc., 23 : 73-78 and 117-121. the storm. [9] Petterssen, S., 1950: Some aspects of the general The problem of forecasting the motion of ty- circulation of the atmosphere. Centenary Pro- ceedings of the Royal Meteorological Society, 1950: phoons is then a problem of evaluating the steer- 120-155. ing [10] and related effects such as the "Fujiwhara [10] Riehl, H., and N. M. Burgner, 1950: Further stud- effect" (the tendency for two cyclones to rotate ies of the movement and formation of hurricanes and their forecasting. Bull: Amer. Meteor. Soc., counterclockwise about each other [4]), and a 31 (7) : 244-253. problem of estimating the northward acceleration [11] Riehl, H., 1948: On the formation of typhoons. J. which must be added to the steering effect. Due Meteor., 5 (6) : 247-264. [12] Rossby, C-G., 1948: On displacements and intensity to the lack of symmetry of most typhoons, a quan- changes of atmospheric vortices. J. Marine Res., titative computation seems impractical. The quan- VII: 175-187. tities n and rt can occasionally be evaluated from [13] U. S. Weather Bureau, 1948: Hurricane notes. Weather Bureau Training Paper No. 1, 210 pp. velocity profiles obtained by individual weather [14] Yeh, T. C., 1950: The motion of tropical un- reconnaissance flights, but the application of values der the influence of a superimposed southerly cur- obtained from one cross section to the whole storm rent. /. Meteor., 7 (2) : 108-113. [15] Haurwitz, B., 1946: The motion of tropical cyclone is not likely to be of much value. A possibility for pairs. Trans. Amer. Geophys. Un., 27 (V) : 658.

Unauthenticated | Downloaded 10/04/21 08:35 AM UTC