Cyclogenesis Off the African Coast: the Case of Cindy in August 1999

SEPTEMBER 2005 NOTES AND CORRESPONDENCE 2803

NOTES AND CORRESPONDENCE

Cyclogenesis off the African Coast: The Case of Cindy in August 1999

SAÏDOU MOUSTAPHA SALL Ecole Supérieure Polytechnique Laboratoire de Physique de I’Atmosphère Siméon Fongang, Universitè Cheikh Anta Diop, Dakar, Senegal

HENRI SAUVAGEOT Observatoire Midi-Pyrénées Laboratoire d’Aérologie, Université Paul Sabatier, Toulouse, France

(Manuscript received 24 August 2004, in final form 14 March 2005)

ABSTRACT

Using radar data from Dakar (Sengal), National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalyses, outgoing longwave radiation provided by the Na- tional Oceanic and Atmospheric Administration (NOAA) Television Infrared Observation Satellite (TIROS) satellite series as well as data from the National Hurricane Center (NHC), a cyclogenesis process leading to the birth of a tropical cyclone from a Sahelian mesoscale convective system (MCS) off the African coast of Senegal is described. The cause of this evolution seems to be the coincidence of the MCS with an easterly wave over a warm sea, the presence of a wide area of precipitable water vapor, strong convergence in the low and midtropospheric layers, and an easterly vertical shear of the zonal wind. As a result, a dynamically well organized convective system built up and the system rapidly strengthened. Before moving away from the African coast of Senegal, this perturbation, which became the tropical cyclone “Cindy,” caused the wreck of more than a hundred fishing pirogues and the deaths of many fishermen because of the suddenness and speed of the phenomenon.

1. Introduction shown that hurricanes reaching the coast of the United States are markedly more numerous when wetter con- Mesoscale convective systems (MCSs) and notably ditions with numerous MCSs are observed over the Sa- squall lines are the most frequently observed rain- hel (Gray 1990; Gray and Lansea 1992; Gray et al. 1992; bearing systems in the Soudano–Sahelian strip (e.g., Landsea et al. 1998; Lawrence et al. 2001). The power- Laurent et al. 1998; Laing et al. 1999). Sahelian MCSs ful, frequent, and well-organized continental MCSs of rapidly move westward over hundreds or thousands of the very wet years are correlated with a high number of kilometers and frequently reach the Atlantic Ocean. At cyclonic perturbations over the Atlantic Ocean. The the end of their continental path, when they cross the transition from mesoscale convective systems toward Atlantic Coast, most Sahelian MCSs weaken and dis- cyclonic vortices has been discussed by Reed et al. appear over the sea near the coast. However some sys- (1977), Bartels and Maddox (1991), and Harr et al. tems strengthen. These strengthening systems play a (1996), among others. major part in the birth of the tropical cyclones observed Many previous studies have led to a better under- in the late summers over the western Atlantic area standing of the processes that govern the dynamics of (e.g., Gray and Landsea 1992; Gray et al. 1992; Landsea the cyclonic tropical disturbances. Thorncroft and et al. 1998; Thorncroft and Hodges 2001). Authors have Hodges (2001) showed that there is a correlation between the number of easterly waves over West Africa and the number of tropical cyclones observed over the Corresponding author address: Dr. Henri Sauvageot, Université Paul Sabatier, Observatoire Midi-Pyrénées, Laboratoire Atlantic. Lawrence et al. (2001) analyzed the charac- d’Aérologie, 65300 Lannemezan, France. teristics of the 1999 cyclonic season including Cindy, E-mail: [email protected] the case studied in the present paper. They found that

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC

MWR3003 2804 MONTHLY WEATHER REVIEW VOLUME 133

FIG. 1. Area observed by the radar of Dakar-Yoff. five tropical cyclones reached level 4 on the Saffir– width 3 ␮s, and the 3-dB beamwidth 1.3°. The data are Simpson code (Simpson 1974) making this season the in the form of plan position indicator (PPI) and range– most active one since 1985. Their study also showed height indicator (RHI) of reflectivity gathered with a that 11 of the 12 tropical cyclones recorded during that time interval of 10 to 20 min. The reflectivity is digi- year (i.e., 92%) originated from an easterly wave, which talized on 8 bits (Sauvageot and Despaux 1990) and is much higher than the mean of 62% for the period used over a circular area with a radius 250–300 km (Fig. between 1967 and 1999. However, the transition be- 1). It is calibrated and corrected for attenuation and tween continental MCSs and the first cyclonic organi- variability of the reflectivity vertical profile using a zation off the African coast is not well understood be- probability matching method (Calheiros and Zawadzki cause it has rarely been observed and described. 1987) as described in detail by Nzeukou and Sauvageot The goal of this paper is to present some observa- (2002). Using these data, which are available for the tions about the birth of Cindy, a particularly clear case 1993 to 1999 period, these authors have studied some of a filiation (or link) between a radar-observed distur- characteristics of the rain field in this coastal area and bance originating from Sahelian Africa and the begin- shown that it is ergodic; that is, the mean and variance ning of a cyclogenesis over the Atlantic Ocean. The of the rain rate are not dependent on the latitude and National Hurricane Center (NHC; http://www.nhc.noaa longitude, although the cumulative annual rain rate dis- .gov) classified the studied disturbance as a “tropical plays strong meridional and sea–land gradients. These depression” on 19 August 1999 when it was located results are confirmed by those of Kebe et al. (2005) who close to the African coast, southwest of Dakar, Senegal, show that the ratio of the cloud-top areas having an around the point of coordinates (13.5°N, 18.9°W). equivalent blackbody brightness temperature lower Later, this disturbance became Hurricane Cindy. Such than Ϫ35°C to the radar-observed rain areas is constant radar observations have never been presented before. throughout the entire observed area and equal to about 1.68. In the whole area considered in the present paper, 2. Data the land surface is flat with an altitude lower than 200 m. Reanalyses of National Centers for Environmental The precipitation field in the coastal area is observed Prediction–National Center for Atmospheric Research with a meteorological radar located at the airport of (NCEP–NCAR) are used to describe the air motion, Dakar-Yoff (14°34ЈN, 17°29ЈW, altitude 30 m). It is a instability, and water distribution associated with the non-Doppler C-band radar, with a peak power of 250 disturbance. These distributions, obtained from a com- kW. The pulse repetition frequency is 250 Hz, the pulse bination between observations and general circulation

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC SEPTEMBER 2005 NOTES AND CORRESPONDENCE 2805

FIG. 2. Time series of radar reflectivity factor PPI gathered with the radar of Dakar-Yoff from 17 to 19 Aug 1999 showing the transition from the MCS (a), (b) over land to (g), (h) a spiral-shaped structure over sea. Reflectivity is in dBZ. The beam elevation angle was 0.8°.

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC 2806 MONTHLY WEATHER REVIEW VOLUME 133

FIG. 3. RHI of the radar reflectivity factor obtained with the radar of Dakar-Yoff on 18 Aug 1999 at 0855 UTC along the 218.6° azimuthal direction. Range markers are 10 km from each other (total distance 50 km). models, provide good representations of the main char- Atlantic Ocean, outside the radar-observed area. Low acteristics of the atmospheric circulation. Details about values of the OLR tend to indicate the location of con- the models, assimilation techniques, quality control, vective systems with deep vertical development. Infor- and validity limits can be found in Kalnay et al. (1996). mation on the characteristics of the tropical cyclones A discussion on the validity of these reanalyses for provided by the NHC are used to follow and identify West Africa is given by Diedhiou et al. (1999). Grist the time evolution of the disturbances. and Nicholson (2001) have used the reanalyses to em- phasize the difference between four humid years (1958– 3. Results and discussion 61) and four dry years (1982–85) in the Sahelian area. Nicholson and Grist (2003) show the agreement be- On 17 August 1999, around 2300 UTC, a mesoscale tween the meteorological profiles obtained by radio- convective system moving westward entered in the area soundes and retrieved from the NCEP reanalyses over observed by the radar. Its western part appeared on the West Africa. The reanalysis provides air temperature, radar screen 250–300 km southeast of Dakar (Fig. 2a). geopotential altitude, as well as the three components After 0230 UTC 18 August, when reaching the coast, its of wind velocity for 17 pressure levels in a grid mesh of shape was that of a short squall line extending 250 km 2.5° ϫ 2.5°. It also provides the relative humidity be- along the meridional direction (Fig. 2b). While pro- tween 1000 and 300 hPa. gressing over the sea, its area and reflectivity increased The outgoing longwave radiation (OLR) provided by (Fig. 2c). Considering the main rain cell clusters, its the National Oceanic and Atmospheric Administration westward average velocity before 18 August at 0715 (NOAA) Television Infrared Observation Satellite UTC was 40 km hϪ1 (320 km in 8 h). After that time, (TIROS) satellite series is also used in the present study the westward motion slowed down (about 20 km hϪ1) to follow the evolution of the studied system over the and the heading turned toward the southwest. Figure 3

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC SEPTEMBER 2005 NOTES AND CORRESPONDENCE 2807

FIG. 4. Wind field at 0000 and 1200 UTC from 18 to 21 Aug 1999 at 1000 hPa from the reanalyses of NECP–NCAR.

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC 2808 MONTHLY WEATHER REVIEW VOLUME 133

FIG. 5. Precipitable water in the atmosphere at 15°N from 17 to 23 Aug 1999 from the NCEP–NCAR reanalyses. presents an RHI at 0855 UTC along the southwest di- every day from 18 to 21 August. It shows that the wind rection limited to 50 km of radial distance (along this velocities were low over the land south of about 18°N. direction the system had an extent of 200 km). It shows Over the ocean, the intertropical convergence zone the convective cell nearest the radar. The RHI of Fig. 3 (ITCZ) appeared clearly between 8° and 15°N, with an shows that the cloud top reached almost 19 km of alti- eastward motion of increasing intensity near the coast. tude with maximum reflectivities higher than 55 dBZ, When the convective system approached the coast on which is equivalent to a rain rate of 180 mm hϪ1 [using 18 August, it was associated with an area of conver- the relation Z ϭ 368 R1.30 found by Sauvageot and gence west of Dakar. When the system passed over the Lacaux (1995) and Nzeukou et al. (2004) for West Af- ocean the cyclogenesis began—it first appeared on 18 rica]. In Fig. 2c, the organization is no longer that of an August at 1200 UTC—and intensified while moving ordinary squall line but is a compact and roughly cir- away from the coast. The center of the cyclonic meso- cular convective system. In Figs. 2c–h, two radials of scale circulation remained around 15°N. It moved from low reflectivity are observed along the 190° and 210° about 15° to 29°E from 18 August at 1200 UTC to 21 azimuthal directions. These radials are linked to beam August at 1200 UTC, with a mean velocity of 466 km blocking by two small volcanic hills, about 80 m high, dayϪ1 or about 20 km hϪ1 westward. named the Mamelles, located near the airport. Figure 5 gives the distribution of precipitable water Beginning around 1900 UTC, the organization of the in the atmosphere at 15°N from 17 to 23 August 1999 system modified and took an annular shape (Figs. 2d–f) from the NCEP–NCAR reanalyses. It shows that the evolving into a kind of spiral shape clearly visible on 18 cyclogenesis process is associated with the presence August around 0000 UTC in Figs. 2g and 2h. However over the Atlantic Ocean of a wide area of wet air to in Figs. 2d–h, the radar acquisition has been restricted supply the approaching MCS (notably on 17 August). by the operator to 250 km (for unknown reason) giving During that period, the sea surface temperature in the a circular shape to the farthest limit of the echo distri- concerned area was higher than 26°C. bution, which is not the real limit. During the coastal crossing of the MCS on 18 Au- The wind field at 1000 hPa associated with this evo- gust, the surface pressure at Dakar fell by about 4 hPa. lution is presented in Fig. 4 at 0000 and 1200 UTC for Figure 6 presents the vertical profiles obtained from

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC SEPTEMBER 2005 NOTES AND CORRESPONDENCE 2809

FIG. 6. Profiles of radiosonde observations at Dakar on 19 Aug 1999 at 0000 and 1200 UTC (from Meteorological Services of Senegal); E and S are for east and south, respectively, and indicate the direction the wind is from. radiosonde observations at Dakar on 19 August at 0000 Senegal were swept by a northward wind. At Dakar, and 1200 UTC. It shows that the winds aloft had a the wind velocity was 43 km hϪ1 at 0730 UTC and the westward direction at all levels at 0000 UTC and had visibility was reduced to 6 km. turned toward the north below 700 hPa at 1200 UTC. Vertical wind shear is an important dynamic param- At that time, all the coastal meteorological sites of eter for convective development. Weak vertical shear is

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC 2810 MONTHLY WEATHER REVIEW VOLUME 133

Ϫ Ϫ FIG. 7. Field of horizontal divergence at 850 hPa on 18 and 19 Aug 1999, at 0000 and 1200 UTC, expressed in 10 6 s 1, from the reanalyses of NCEP–NCAR. however observed to be favorable to the organization 8 presents the wind at 700 hPa for the period from 18 to of the tridimensional circulation associated with the cy- 22 August with superimposed cloud cover deduced clogenesis (e.g., McBride and Zehr 1981; De Maria from the OLR signal. What can be seen in Fig. 8 is an 1996). Figure 6 shows that, in the present case, the ver- easterly wave around a latitude of 17.5°N on 18 August tical wind shear between the surface and 600 hPa was with the cyclonic depression under the ridge of the about 20 m sϪ1, which is not really weak since 15 m sϪ1 wave. Between 20 and 22 August, the depression inten- is often quoted as an upper bound that prevents storm sified just in front of the wave. The vertical wind shear formation. A possible explanation of why Cindy was moderate, a factor that is favorable to the cyclo- formed despite fairly large values of vertical wind shear genesis. can be found in a numerical modeling study by Tuleya Figure 9 gives the classification of the studied con- and Kurihara (1981). They found that “when the mean vective system between 18 and 22 August (from surface wind is easterly, easterly vertical shear (i.e. east- Lawrence et al. 2001) with respect to the criteria of the erlies increasing with height) is conductive for storm World Meteorological Organization. The “tropical de- genesis” because the coupling of the upper troposphere pression” character is attributed to the system on 18 with the moving disturbance is thus more complete. In August and maintained on 19 and 20 August. It became one of their experiments with environmental wind val- a tropical storm on 20 August at 1800 UTC and Hur- ues very similar to that of the present case, the initial ricane Cindy on 22 August at 0000 UTC. disturbance developed into a vigorous tropical storm Ϫ1 with surface wind exceeding 22 m s . 4. Conclusions The field of divergence (10Ϫ6 sϪ1) at 850 hPa at 0000 and 1200 UTC on both 18 and 19 August is presented in In this note the successive steps of the transformation Fig. 7. It clearly shows the location, intensification, and leading to the birth of a tropical depression on the near offshore motion of the area of convergence associated Atlantic from a Sahelian MCS is described. This evo- with the cyclogenesis (and also a diurnal effect). Figure lution occurred over 4 days. On 18 August the “pri-

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC SEPTEMBER 2005 NOTES AND CORRESPONDENCE 2811

FIG. 8. Wind at 700 hPa on 18 to 22 Aug 1999 from the NCEP– NCAR reanalyses showing the easterly wave leaving the continent (bold line). The cloud cover provided by the OLR is superimposed upon the flux.

mary” MCS approached the Atlantic Coast around northward at lower latitude, low winds on the conti- 13°N, and on 22 August it was first identified as a hur- nent, and the presence of a wide area of a wet air over ricane by the NHC and named Cindy. the Atlantic Ocean. Over the sea, the MCS rapidly The primary disturbance had a meridional size of spread out while a wide echo-free area appeared in its about 250 km and was organized as a short squall line central part. The spiral shape, which is characteristic of moving westward with a velocity of 40 km hϪ1. When a mesoscale vortex, appeared at a distance to the Afri- moving over the sea between 14°40Ј and 12°30ЈN—that can coast no larger than 150 km (Fig. 2h). The time is, between Cape Verde and Gambia—the system interval between the arrival of the primary MCS on the strengthened and its maxima of reflectivity were higher coast and this first radar-observed mesoscale vortex than 55 dBZ, or 180 mm hϪ1 of rain rate. The sea was only 24 h (from 0000 to 2400 UTC on 18 August). surface temperature was higher than 26°C. The horizontal size of this mesoscale vortex was about Associated with the easterly mean surface wind, the 200 km in the radar field (with the restriction that a part vertical wind shear was high, which is rather unusual. In of the system was beyond the maximum radar-observed fact, the observed environmental wind vertical profile distance). The footprint of the MCS and its evolution was found in agreement with the results of a numerical toward a hurricane was very clear on the low-level modeling study by Tuleya and Kurihara (1981) showing (1000 hPa) wind field provided by the NCEP–NCAR that the coupling of the upper-tropospheric circulation reanalyses. The wind flux at 700 hPa clearly showed with the moving disturbance is more complete with an that an easterly wave was associated with this cyclogen- easterly vertical wind shear. esis. This evolution was associated with a strong mass con- Most of the factors favorable to cyclogenesis were vergence at low levels over the sea (Fig. 4), with a simultaneously present, and it is hypothesized that this strong, wide southward motion off the Mauritanian is the main cause of the rapid cyclone genesis process. coast north of 15°N, a wind veering eastward then Tropical cyclogenesis has been observed to occur in a

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC 2812 MONTHLY WEATHER REVIEW VOLUME 133

FIG. 9. Time series of the classification of the studied disturbance between 18 and 23 Aug 1999: tropical depression, tropical storm, and hurricane (Cindy).

variety of ways, however, and it is still impossible to from Senegal. Thanks also to two reviewers for valu- prove that Cindy’s birth was an archetype of this pro- able comments and suggestions. cess. Numerous other similar observations would be necessary to document and comprehensively depict the REFERENCES tropical cyclogenesis off the African coast and to iden- Bartels, D. L., and R. A. Maddox, 1991: Midlevel cyclonic vortices tify with a correct weighting of the decisive factors. generated by mesoscale convective systems. Mon. Wea. Rev., The observations presented in the present paper sug- 119, 104–118. gest that a powerful S-band radar (which is not the Calheiros, R. V., and I. Zawadzki, 1987: Reflectivity–rain rate re- modest C-band Dakar-Yoff radar available, with its lationships for radar hydrology in Brazil. J. Climate Appl. Meteor., 26, 118–132. small antenna and peak power) located on the African De Maria, M., 1996: The effect of vertical shear on tropical cy- coast or on the Cape Verde Islands could be very useful clone intensity change. J. Atmos. Sci., 53, 2076–2087. to an early identification of tropical depressions and Diedhiou, A., S. Janicot, A. Viltard, P. de Félice, and H. Laurent, cyclones and to more accurate marine warning fore- 1999: Easterly wave regimes and associated convection over casts for African fishermen. In the annual summary of West Africa and tropical Atlantic: Results from the NCEP/ NCAR and ECMWF reanalyses. Climate Dyn., 15, 795–822. the Atlantic hurricane season of 1999 by Lawrence et Gray, W. M., 1990: Strong association between West African rainfall al. (2001), no direct deaths were attributed to Cindy, and US landfall of intense hurricanes. Science, 249, 1251–1256. but the lives of numerous African fishermen were lost ——, and C. W. Landsea, 1992: African rainfall as a precursor of in the process of its formation. hurricane-related destruction on the U.S. East Coast. Bull. Amer. Meteor. Soc., 73, 1352–1364. Acknowledgments. The authors thank the Agence ——, ——, P. W. Mielke Jr., and K. J. Berry, 1992: Predicting Atlantic seasonal hurricane activity 6–11 months in advance. Universitaire de la Francophonie (AUF) for funding Wea. Forecasting, 7, 440–455. the stay in France of one of them (SMS). They are also Grist, J. P., and S. A. Nicholson, 2001: A study of the dynamic grateful to the National Centers for Environmental factors influencing the variability of rainfall in the West Af- Prediction–National Center for Atmospheric Research rican Sahel. J. Climate, 14, 1337–1359. and to the National Hurricane Center for making their Harr, P. A., R. L. Elsberry, and C. L. Chan, 1996: Transformation of a large monsoon depression to a tropical storm during data available to them, to the Direction de la TCM-93. Mon. Wea. Rev., 124, 2625–2643. Météorologie du Sénégal and the Agence pour la Sécu- Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Re- rité de la Navigation Aérienne for providing the data analysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC SEPTEMBER 2005 NOTES AND CORRESPONDENCE 2813

Kebe, C. M. F., H. Sauvageot, and A. Nzeukou, 2005: The relation Nzeukou, A., and H. Sauvageot, 2002: Distribution of rainfall between rainfall and area–time integrals at the transition parameters near the coasts of France and Senegal. J. Appl. from an arid to an equatorial climate. J. Climate, in press. Meteor., 41, 69–82. Laing, A. G., J. M. Fritsch, and A. J. Negri, 1999: Contribution of ——, ——, A. D. Ochou, and C. M. F. Kebe, 2004: Raindrop size mesoscale convective complexes to rainfall in Sahelian Af- distribution and radar parameters at Cape Verde. J. Appl. rica: Estimates from geostationary infrared and passive mi- Meteor., 43, 90–105. crowave data. J. Appl. Meteor., 38, 957–964. Reed, R. J., D. C. Norquist, and E. E. Recker, 1977: The structure Landsea, C. W., G. D. Bell, W. M. Gray, and S. B. Goldenberg, and properties of African wave disturbances as observed dur- 1998: The extremely active 1995 Atlantic hurricane season: ing phase III of GATE. Mon. Wea. Rev., 105, 317–333. Environmental conditions and verification of seasonal fore- Sauvageot, H., and G. Despaux, 1990: SANAGA: Un systéme casts. Mon. Wea. Rev., 126, 1174–1193. d’acquisition numérique et de visualisation des données radar Laurent, H., N. D’Amato, and T. Lebel, 1998: How important is pour la validation des estimations satellitaires de précipita- the contribution of the mesoscale convective complexes to tions. Veille Climatique Satellitaire, 30, 51–55. the Sahelian rainfall? Phys. Chem. Earth, 23, 629–633. ——, and J. P. Lacaux, 1995: The shape of averaged drop size Lawrence, M. B., L. A. Avila, J. L. Bevan, J. L. Franklin, J. L. distributions. J. Atmos. Sci., 52, 1070–1083. Guiney, and R. J. Pasch, 2001: Atlantic hurricane season of Simpson, R. H., 1974: The hurricane disaster potential scale. 1999. Mon. Wea. Rev., 129, 3057–3084. Weatherwise, 27, 169–186. McBride, J. L., and R. Zehr, 1981: Observational analysis of tropi- Thorncroft, C. D., and K. I. Hodges, 2001: African easterly wave cal cyclone formation. Part II: Comparison of non-developing variability and its relationship to Atlantic tropical cyclone versus developing systems. J. Atmos. Sci., 38, 1132–1151. activity. J. Climate, 14, 1166–1179. Nicholson, S. E., and J. P. Grist, 2003: The seasonal evolution of Tuleya, R. E., and Y. Kurihara, 1981: A numerical study on the the atmospheric circulation over West Africa and equatorial effects of environmental flow on tropical storm genesis. Mon. Africa. J. Climate, 7, 1013–1030. Wea. Rev., 109, 2487–2506.

Unauthenticated | Downloaded 09/29/21 11:12 PM UTC