The so-called “perturbation D” was a nondeveloping West African disturbance observed near Dakar (Senegal) during special observing period (SOP) 3 of the African Monsoon Multidisciplinary Analysis (AMMA) in September 2006. Its mesoscale environment is described with the dropsonde data obtained during flights on three successive days with the Service des Avions Français Instrumentés pour la Recherche en Environnement Falcon-20 aircraft. Processes involved in this evolution are studied qualitatively with ECMWF reanalyses and Meteosat-9 images. The evolution of perturbation D was the result of an interaction between processes at different scales such as the African easterly jet (AEJ), a midtropospheric African easterly wave (AEW), a series of mesoscale convective systems, the monsoon flow, dry low- to midlevel anticyclonic Saharan air, and a midlatitude upper-level trough. The interaction between these processes is further investigated through a numerical simulation conducted with the French nonhydrostatic Méso-NH model with parameterized convection. The growth of the simulated disturbance is quantified with an energy budget including barotropic and baroclinic conversions of eddy kinetic energy, proposed previously by the authors for a limited domain. The development of the simulated system is found to result from barotropic–baroclinic growth over West Africa and baroclinic growth over the tropical eastern Atlantic. It is suggested that these energy conversions were the result of an adjustment of the wind in response to the pressure decrease, presumably caused by convective activity, and other synoptic processes. A comparison with the developing case of Helene (2006) reveals that both perturbations had similar evolutions over the continent but were associated with different synoptic conditions over the ocean. For perturbation D, the anticyclonic curvature of the AEJ, caused by the intensification of the eastern ridge by a strong flow of dry Saharan air, prohibited the formation of a closed and convergent circulation. Moreover, a midlatitude upper-level trough approaching from the northwest contributed to increase the northward stretching and then weakened the perturbation. It is therefore suggested that at least as important as the intensity of the AEW trough and associated convection leaving the West African continent are synoptic conditions associated with the Saharan heat low, the subtropical high pressure zone, and even the midlatitude circulation, all of which are instrumental in the (non)cyclogenetic evolution of AEWs in the Cape Verde Islands region.
Seventy years ago, Piersig [1936; see Piersig (1944) for a partial translation into English] was probably the first to suggest a relationship between Atlantic hurricanes and westward-moving low pressure systems accompanied by disturbed weather off West Africa during the monsoon season. Hubert (1939) associated the so-called tropical cyclone of New York (1938) with such a West African monsoon perturbation. More generally, Dunn (1940) proposed that Atlantic hurricanes develop from easterly wavelike disturbances, and that some of them arise near the Cape Verde Islands with a frequency of 3–4 days. During the summer months, Schove (1946) observed that the strongest West African squall lines are associated with a synoptic trough, confirming the “popular idea that Antilles hurricanes originate in West African line squalls” (p. 110). Erickson (1963) referred to these wavelike disturbances as African easterly waves (AEWs), characterized by a wavelength of 3000–5000 km, a period of 3–5 days, and a maximum amplitude near 3000-m altitude, with a succession of midtropospheric troughs and ridges. Erickson (1963) suggested that Hurricanes Debbie and Esther in 1961 originated from such AEW troughs. Carlson (1969a) also proposed a similar origin for Hurricanes Beulah and Chloe and Tropical Storm Ginger in 1967. Using satellite images and radiosonde data, Carlson (1969a) focused more particularly on the apparent interaction between AEWs and convection over West Africa, and Carlson (1969b) suggested that part of AEWs’ energy comes from convective processes. According to Carlson (1969a), however, there is no evident correlation between the intensity of AEW troughs leaving the West African coast and cyclogenetic evolution over the tropical Atlantic Ocean. Nevertheless, systematic operational tracking of AEW has been done since that time and it is now admitted that a large proportion of Atlantic hurricanes evolve from AEW troughs (e.g., Avila and Clark 1989).
Using statistics of AEWs in a 20-yr European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis dataset, Thorncroft and Hodges (2001) proposed that the cyclogenetic evolution of AEW troughs depends on their low-level amplitude when they leave the West African coast. Using composite statistics on a 60-yr National Centers for Environmental Prediction reanalysis dataset, Aviles (2004) applied the Lorenz (1955, hereafter L55) energetic analysis to differentiate developing AEW troughs from nondeveloping ones. Surprisingly, she found that “the sources of eddy kinetic energy of developing AEWs are not as clearly barotropic and baroclinic as for non-developing ones” (Aviles 2004, 136–137). Indeed, the cyclogenetic evolution of AEW troughs does not depend much on their energetic growth over West Africa but rather on their more or less favorable internal structure when they leave the West African coast (“anticyclonic flow in the upper levels and surrounding the positive vorticity center, deep rising motion, weak and small cold core, a moist environment”; Aviles 2004, p. 140) and, to lesser extent, on a favorable environment in the Atlantic basin (e.g., Gray 1968).
Lin et al. (2005) studied the AEW trough that spawned Hurricane Alberto (2000) and identified three successive convective genesis and lysis periods before the final cyclogenesis evolution occurred off the Guinean coast. For the same case study, Berry and Thorncroft (2005) suggested that the cyclonic vorticity structure of this AEW trough merged with cyclonic vortices of convective origin over the Guinean Highlands, thus resulting in its intensification. They finally hypothesized that this interaction process between the AEW and the so-called mesoscale convective systems (MCSs) led to the cyclogenesis of Alberto (2000), one day after the disturbance left the West African coast.
Arnault and Roux (2009, hereafter AR09) revisited the L55 analysis of the energy cycle when applied to a finite domain. In particular, they considered potential energy, or enthalpy, instead of available potential energy and they found that the baroclinic conversion between potential energy and eddy kinetic energy is proportional to the work of the horizontal pressure forces by the ageostrophic circulation. This budget was applied to a 4-day mesoscale simulation of the pre-Helene (2006) West African disturbance with the French nonhydrostatic model Méso-NH (Lafore et al. 1998). This developing disturbance was associated with an AEW trough and periods of enhanced convective activity over West Africa and the nearby Atlantic Ocean, before cyclogenesis occurred close to Cape Verde Islands on 12 September 2006. Over the continent on 9–10 September 2006, the simulated disturbance was associated with a positive baroclinic conversion, associated with the geostrophic adjustment of the wind field to the pressure decrease of convective origin. On the opposite, over the ocean on 11–12 September 2006, the simulated disturbance was associated with negative baroclinic conversion and strong positive barotropic conversion. The positive barotropic conversion was related to an intensification of the cyclonic curvature of the easterly flow in the low pressure system. The negative baroclinic conversion was associated with an up-gradient flow induced by strong and divergent northeasterly winds on the southeastern flank of a subtropical high pressure zone to the northwest of the AEW trough. Late on 12 September 2006, baroclinic conversion became positive again due to strong convective activity and an associated pressure decrease in an environment where the AEJ was weak and the northwestern anticyclonic circulation had moved farther west. At that time, the system became Tropical Depression Helene and was able to extract enough energy from the sea surface to become self-sustained (Emanuel 1986).
Here, we study the case of the so-called “perturbation D”, a nondeveloping West African disturbance observed during the Special Observing Period 3 (SOP-3) of the African Monsoon Multidisciplinary Analysis (AMMA; Redelsperger et al. 2006), conducted from Dakar (Senegal) on 15–29 September 2006. Three other disturbances of weaker intensity (perturbations A, B, and C; Jenkins et al. 2010) were observed during the first 10 days of AMMA SOP-3. Perturbation D has been sampled with dropsondes on three successive days (25–27 September) with the Service des Avions Français Instrumentés pour la Recherche en Environnement (SAFIRE) Falcon-20 aircraft. These data are used in section 2a to describe the evolution of its mesoscale structure. A synoptic analysis of the interaction between processes involved in this noncyclogenetic evolution is given in section 2b, based on meteorological fields from the ECMWF and Meteosat-9 satellite images. In section 3, we present a 5-day mesoscale simulation conducted with Méso-NH. The energetic growth of this simulated perturbation D is then quantified using the energy budget proposed by AR09. Conclusions and perspectives are finally presented in section 4.
2. Synoptic analysis
The evolution of perturbation D is investigated with three kinds of data: (i) dropsonde sampling from the SOP-3 of AMMA, (ii) ECMWF reanalysis, and (iii) Meteosat-9 images, which respectively resolve mesoscale, synoptic, and convective processes.
a. Mesoscale environment
Perturbation D was a westward-moving midtropospheric trough associated with enhanced convective activity that crossed the West African coast during the period 25–27 September 2006, as can be deduced from vorticity at 700 hPa and as was analyzed operationally by ECMWF and from Meteosat-9 images (Fig. 1). Three successive flights on 25–27 September 2006 were dedicated to its observation near the West African coast. More precisely, 12 successful dropsondes were launched during flight 74 between 1031 and 1317 UTC 25 September, 8 were launched during flight 76 between 1545 and 1804 UTC 26 September, and 12 were launched during flight 77 between 1332 and 1601 UTC 27 September (see Fig. 1 for the location of horizontal domains sampled with dropsondes). Considering the size of the observed domains, the sampling resolution, and assuming that large-scale atmospheric features remained approximately unchanged during the dropsonde sampling (2–3 h), these dropsonde data can be used to document some mesoscale characteristics of perturbation D.
These dropsonde data have been analyzed with the variational method developed by Moine (2001) to deduce three-dimensional fields of wind components, temperature, and humidity on a regular grid in domains of approximately 500 km × 500 km horizontally and 10 km vertically. As seen in Fig. 2 from 25 to 26 September, the troposphere progressively moistened (relative humidity > 90%) and cyclonic vorticity intensified (up to 5 × 10−5 s−1) between 2 and 9 km altitude. On 27 September, however, this evolution was hampered by a dry southeasterly Saharan flow extending from the surface to 9-km altitude, weakly cyclonic below 5 km and anticyclonic above.
These dropsonde data were not assimilated operationally, but they were included in a reanalysis program dedicated to AMMA by ECMWF (Agustí-Panareda et al. 2009). Another important change in the AMMA reanalysis compared to operational analyses is the use of the more realistic convection scheme proposed by Bechtold et al. (2008). The reanalyzed fields are more coherent with the observations, with a closer connection between cyclonic vorticity maxima and cold cloud tops associated with MCSs (Fig. 1).
b. Interaction between convective and synoptic processes
The synoptic environment of perturbation D is deduced from the ECMWF AMMA reanalyses. Air temperature, relative humidity, geopotential, wind components, relative vertical vorticity ζ, and horizontal divergence are available on a regular latitude–longitude grid of 0.5° resolution, at 21 pressure levels between 1000 and 1 hPa, at 0000, 0600, 1200, and 1800 UTC. The convective activity associated with perturbation D is deduced from the Meteosat-9 brightness temperature Tb images in the water vapor channel at 6.85–7.85 μm, available every 15 min. The Meteosat-9 raw images have been interpolated on a regular latitude–longitude grid of 0.027° resolution and a brightness temperature resolution of 0.5 K.
2) Hovmöller diagrams
Perturbation D was an AEW trough associated with several convective genesis and lysis periods, similar to the developing case study of Lin et al. (2005). Figure 3a shows a Hovmöller space–time diagram for the ECMWF-reanalyzed relative vertical vorticity at 700 hPa averaged between 8° and 16°N from 22 to 29 September. According to Fig. 3a, perturbation D was a well-defined westward-moving AEW trough that formed on the afternoon of 22 September over Ghana (approximately 0° longitude) and dissipated on the afternoon of 28 September near the Cape Verde Islands (approximately 25°W).
The Hovmöller diagram computed with Meteosat-9 brightness temperature (Fig. 3b) confirms that this AEW trough was associated with convection. More particularly, there were two continental convective developments embedded in the AEW trough during the afternoons of 22 and 23 September, although these convective systems moved faster and dissipated west of the trough. This was followed by convective redevelopments on 25–26 September near the Guinean highlands and the Atlantic coast, in parallel with an intensification of the AEW trough at 700 hPa. This enhanced convective activity near the Guinean coast is comparable with that of the pre-Alberto (2000) (Lin et al. 2005; Berry and Thorncroft 2005) and pre-Helene (2006) disturbances (AR09). Finally, convection developed over the ocean on the afternoon of 27 September, but it was short lived and this AEW trough did not lead to tropical cyclogenesis.
3) Composite horizontal cross sections
Synoptic and convective processes involved in the evolution of perturbation D have been further investigated with composite horizontal cross sections from ECMWF synoptic fields and Meteosat-9 brightness temperatures (Fig. 4). Brightness temperatures below −30°C indicate convective activity. The synoptic environment is characterized at 925 hPa by Saharan dry air (wind with northerly component) and monsoon flow (humidity flux with a southerly component), and at 700 hPa by the AEJ (winds with a zonal component less than −10 m s−1). The trough axes of the AEWs have been computed with the objective method proposed by Berry and Hewson (2007). Only the period from 0600 UTC 25 September 2006 to 1800 UTC 27 September 2006, when perturbation D crossed the West African coast, will be discussed here (Fig. 4).
On 25 September (Figs. 4a,b), perturbation D was relatively active over southern Senegal (10°–13°N, 13°–16°W), due to the southwesterly monsoon flow bringing moist air in the low levels. The northern anticyclonic conditions, however, intensified at low and mid levels with a low-level northeasterly flow of dry Saharan air coming from northern Mali. This midlevel anticyclonic circulation was related to an enhanced AEW ridge east of perturbation D (Figs. 4c,d). The dry low-level northerly flow extended westward to Mauritania and the midlevel AEW ridge east of perturbation D strengthened during the following days. As a consequence, midtropospheric southeasterly winds intensified east of perturbation D, moving it northwestward, away from the low-level monsoon flow on 26–27 September. This enhanced anticyclonic Saharan flow stretched the midlevel cyclonic vortex associated with perturbation D, thus partly “disorganizing” it (Figs. 5a–d). The cyclonic vortex was still present on the morning of 27 September but with weaker intensity (cf. Figs. 5a,d). During the afternoon of 27 September (Fig. 4f), one last MCS associated with perturbation D developed off the Senegal coast, but it dissipated during the next day. At the same time, midlevel cyclonic vorticity intensified and “reorganized” (Figs. 5e,f), but anticyclonic Saharan flow finally disorganized this cyclonic structure on 28 September (Figs. 5g,h,j,k).
While low- to midlevel anticyclonic circulation of Saharan origin was intensifying and moving westward on 26–27 September, an upper-level trough in the midlatitude westerlies was moving southward toward the Cape Verde Islands region (Fig. 6). Upper tropospheric divergence in the southern part of the trough (15°–25°N, 15°–25°W in Figs. 6b–d) was probably instrumental for the development of the oceanic MCS on the afternoon of 27 September. The influence of this midlatitude upper-level trough could also be seen at the midlevel (700 hPa) although with weaker intensity. This midlatitude trough was located west-northwest of the midlevel AEW ridge on 27 September and contributed to intensifying the southerly flow northeast of perturbation D. These observations suggest an interaction between the eastern AEW ridge, the eastern midlevel anticyclonic Saharan air, the northern midlatitude trough, and the failed evolution of perturbation D. Dust in the Saharan air could also have played some role in this evolution, although its impact on oceanic convective development during tropical cyclogenesis off West Africa is not clearly established (Jenkins et al. 2008).
4) Composite profile
Perturbation D was associated with a cyclonic maximum of relative vertical vorticity at 700 hPa (Fig. 7). Figure 8a displays the evolution of the vertical profile of ECMWF relative vertical vorticity averaged over the domain enclosing perturbation D, and the evolution of Meteosat-9 cloud area below different temperature thresholds within the same domain. Following Mathon and Laurent (2000), convection is quantified by the evolution of cloud surfaces with brightness temperatures colder than −30°, −50°, and −70°C. Evolution of areas colder than −30°C in Fig. 8a shows the disturbance’s life cycle. Three successive periods can be identified for the evolution of perturbation D: Stage I—development and propagation over the continent (22–24 September), Stage II—convective redevelopments when it passed over the Guinean highlands near the Atlantic coast (25–26 September), and stage III—decay over the ocean (27–29 September).
The area colder than −50°C in Fig. 8a focuses on convective activity and the evolution of the area colder than −70°C indicates periods of intense convection. The onset of the disturbance and stage I started with surges of intense convection on the evening of 22 September and during 23 September. This was followed by an enhancement of relative vertical vorticity in the middle troposphere (Fig. 8a), suggesting that the growth of the synoptic wave was influenced by convective processes. This first surge of deep convection dissipated on 24 September, but the vorticity maximum remained visible during the next five days (Fig. 8a). Stage II is related to the crossing of the Guinean coast. A weak convective redevelopment occurred over this mountainous area on the morning of 25 September (Figs. 4a and 8a), followed in the afternoon by a convective burst over the ocean, which quickly dissipated on 26 September (Figs. 4b–e and 8a). These relatively short-lived systems were associated with intensifying midlevel cyclonic vorticity at synoptic scale in the AEW trough (Fig. 8a). The resulting long-lived cyclonic perturbation may have helped to trigger the successive convective developments (e.g., Davis and Trier 2007). Perturbation D was dissipating in the beginning of stage III when a last MCS rapidly grew up off the Senegal coast on the afternoon of 27 September (Figs. 4f and 8a), which probably helped to temporarily amplify the midtropospheric vorticity maximum (Figs. 5e,f and 8a).
3. Energy analysis
a. Mesoscale modeling
To further investigate the evolution of perturbation D during its transition from the continent to the ocean, a 5-day numerical simulation, starting from 24 September at 0000 UTC, was conducted with the French nonhydrostatic Méso-NH model. The characteristics of this simulation are similar to those discussed in AR09 for the pre-Helene (2006) West African disturbance.
The simulated domain has 200 × 200 horizontal grid points and is located at 6°S–36°N, 35°W–9°E. The horizontal resolution is 24 km and parameterized convection was used (Kain and Fritsch 1993; Bechtold et al. 2001). The vertical grid has 66 levels from the surface to 28-km altitude, with a grid spacing of 60 m near the surface up to 600 m at the tropopause level, and it takes the orography into account. The simulation outputs are saved every hour. The boundary conditions are from the ECMWF AMMA reanalyses (Agustí-Panareda et al. 2009).
b. Qualitative validation
Composite horizontal cross sections similar to those discussed in section 2b(3) (Fig. 4) were derived from the simulation outputs (Fig. 9). The brightness temperature in the water vapor channel is calculated from Méso-NH simulation outputs using the Radiative Transfer for Television and Infrared Observation Satellite Operational Vertical Sounder code, version 8.7 (Chaboureau et al. 2000, 2002; Saunders et al. 2005) to facilitate comparisons with Meteosat-9 observations. The dynamic features associated with simulated perturbation D are very similar to those obtained from ECMWF reanalyses, but the simulated convective activity shows some differences from the observed ones. The simulated MCS over southern Senegal on morning of 25 September was as intense as the observed one, with comparable size and location. During the next day, however, the simulated convective activity near the Guinean coast was weaker than the observed one, and its location was slightly different. The final offshore development of 27 September started earlier in the simulation and was substantially weaker.
Simulated midlevel vorticity shows features globally comparable to ECMWF reanalyses (Fig. 7). Méso-NH displays small-scale cyclonic vorticity structures south of Senegal on the afternoon of 25 September (Fig. 7f), certainly in relation with the MCS simulated in that region at that time (Fig. 9b). These small-scale cyclonic vortices contributed to enhance the synoptic vorticity of the associated AEW trough, as in the developing case discussed by Berry and Thorncroft (2005). On the afternoon of 26 September, the remaining small-scale cyclonic vortices were stretched by the enhanced anticyclonic flow of the AEW ridge to the east (Fig. 7g). This “disorganizing process” occurred slightly faster in the simulation than in ECMWF reanalysis, probably due to the weaker convectively induced cyclonic vorticity in the simulation at that time.
A composite profile similar to that discussed in section 2b(4) (Fig. 8a) is derived from the simulation outputs (Fig. 8b). A noticeable result is the weaker proportion of coldest (Tb < −50°C, and Tb < −70°C) cloud tops in the Méso-NH simulation, compared to Meteosat-9, which resulted probably from the use of parameterized convection. The simulated vorticity at 0000 UTC 24 September shows a cyclonic anomaly between the altitudes of 2000 and 5000 m, a probable consequence of the convective surges that initiated perturbation D during the two days before the beginning of the simulation. The simulated MCS on 24–25 September was associated with an enhancement of cyclonic vorticity between 1000 and 4000 m, with a stronger intensity than in the ECMWF analysis, which probably resulted from the slightly finer horizontal resolution of Méso-NH and different convective parameterizations (Molinari and Dudek 1992). This enhanced cyclonic vorticity, however, was short lived and it decayed to values comparable to those before the convective event on the afternoon of 25 September. The enhancement of midlevel cyclonic vorticity on 27 September was less intense in the Méso-NH simulation due to the weaker convective activity simulated at that time (Figs. 8a,b). A more precise study of convective vortices and their interaction with the synoptic AEW requires numerical modeling with higher horizontal resolution and explicit convection, which is out of the scope of this paper and will be done in a future work.
c. Energy budget
AR09 derived an energy budget at local and global scale between moist enthalpy Hm, considered as the potential energy of the atmosphere, and the vertical, zonal, and eddy horizontal components of kinetic energy KV, KZ, and KE, respectively. This energy budget is displayed in Fig. 10 and is given by
In (1a)–(1d) BH, BKV, BKZ, and BKE are the boundary terms of the Hm, KV, KZ, and KE budgets. CPZ (CPE) is the zonal (eddy) baroclinic conversion between Hm and KZ (KE) and is equal to the work of the zonal (eddy) pressure forces by the zonal (eddy) circulation. When the zonal (eddy) circulation is in geostrophic balance, there is no zonal (eddy) baroclinic conversion. CPV is the vertical baroclinic conversion between Hm and KV and is proportional to the vertical pressure work. In the budget of KV, CPV is mainly compensated by the work of the gravitational force DVgravity. Note that Hm can be produced through (i) radiative heating Gdiab, (ii) convective heating Gphase, (iii) an increase of humidity associated with evaporation (sublimation) of liquid (solid) water GWV, (iv) variations of pressure Gp, and (v) cooling of hydrometeors Ghydro. The conversion (CK) between KZ and KE is the barotropic conversion. Also, DZ and DE are the frictional dissipation of KZ and KE. The conversion between horizontal kinetic energy (KZ + KE) and KV due to the Coriolis force is negligible compared to CPZ and CPE, so the budget of KV, except for CPV, will not be discussed here [the terms of the budget of KV that are not explicitly given in Eq. (1b) are precisely discussed in AR09]. This energy analysis is similar to that of Smith (1970), except that it has been derived in a nonhydrostatic framework so KV is not invariant and its contribution has to be considered in the budget. As the vertical baroclinic conversion term CPV is two orders of magnitude larger than the horizontal baroclinic conversion term (CP = CPZ + CPE) (AR09), horizontal (K = KZ + KE) and KV kinetic energy have been separated to clarify the budget.
d. Results for Perturbation D
The energy budget of AR09 is applied to the Méso-NH simulation of perturbation D within domain Δ, defined as 8°–18°N, 4°–27°W and 1000–18 000 m above mean sea level.
1) Eddy vorticity
As stated by AR09, it is first necessary to objectively quantify the contribution of the AEW trough and ridge to the eddy perturbation with respect to the zonal mean in Δ to correctly analyze the budget of KE. The eddy vorticity ξE is defined as
where the asterisk indicates a deviation with respect to the zonal mean. Figure 11a shows a time series of vertical profiles of horizontally averaged values of ξE over Δ. On average, the eddy circulation was cyclonic between 1000 and 5000 m on 24 September. On the afternoon of 25 September, the eddy vorticity became negative at these heights and positive above, due to the presence of anticyclonic Saharan air that began to interact with perturbation D. On 27 September, the eddy circulation became cyclonic again below 3000 m, due to the weak oceanic convective redevelopment and negative between 3000 and 16 000 m due to enhanced anticyclonic flow southeast of the midlatitude upper-level trough (Figs. 6c,d). On the morning of 28 September, the eddy circulation was anticyclonic in the whole troposphere as the dissipating perturbation D went out of Δ.
2) Eddy geostrophic vorticity
As proposed by AR09, the cyclogenetic evolution of an AEW trough could result from wind adjustment following a pressure decrease of convective origin. This process can be analyzed through a comparison between ξE and the eddy geostrophic vorticity ξG, defined as the opposite of the horizontal divergence of the eddy pressure force normalized by the Coriolis parameter f:
where ρa is the density of dry air. Note that ξE and ξG are equal when the atmospheric flow is geostrophic. Positive (negative) values of ξG are associated with a pressure minimum (maximum) and the eddy pressure forces are convergent (divergent). Comparison between the time series of horizontally averaged ξE and ξG (Fig. 11) shows that the eddy circulation was approximately geostrophic; however, the difference between ξE and ξG was slightly larger than for the pre-Helene disturbance (Fig. 14 in AR09). On 24 and 27 September, ξG was positive between 1000 and 6000 m and larger than ξE, which could indicate that the wind field was adjusting to a pressure perturbation of convective origin and the cyclonic eddy circulation was converging toward the pressure low. On 25–26 September, ξE was negative between 1000 and 5000 m with a larger amplitude than ξG, which could correspond to the fact that the strong anticyclonic winds arriving from the east did not have time to adjust to the weaker pressure field associated with the AEW trough and convective activity in the western part of Δ (Fig. 7g).
To further analyze this simulated evolution, horizontal cross sections of total pressure at 3000 m are displayed in Fig. 12. Perturbation D was associated with a pressure increase during the first day of its interaction with anticyclonic Saharan air, with minimum values of 712 hPa at 0000 UTC 25 September near 10°N, 14°W and 714 hPa at 0000 UTC 26 September near 14°N, 17°W. This was followed by a small pressure decrease with a minimum value of 713 hPa at 0000 UTC 28 September near 15°N, 24°W due to the convective activity offshore (Fig. 9f). For comparison, AR09 observed that the pre-Helene disturbance was associated with a minimum pressure of 711 hPa at 0000 UTC 11 September, the day before it crossed the West African coast. This minimum value then decreased to 709 hPa at 0000 UTC 12 September and to 702 hPa at 0000 UTC 13 September. It is interesting to note that the nondeveloping perturbation D was relatively similar to the developing pre-Helene disturbance during its continental stage, as the associated pressure minimum values were comparable. The difference between the cases results from the pressure increase observed on 26 September resulting from the enhanced midlevel AEW ridge east of the disturbance.
3) Eddy kinetic energy
The vertical profiles of KE (Fig. 13) indicate a maximum between 2000 and 5000 m, which corresponds to the energetic signature of the simulated AEW. On the afternoon of 24 September, KE increased between 1000 and 5000 m and decreased slightly at the same altitudes during the afternoon of the next day due to the increased convective activity during the continental stage of perturbation D and the beginning of its interaction with anticyclonic air from the east. On morning of 26 September, KE increased between 2000 and 5000 m and decreased with similar amplitude during the afternoon, which is the energetic signature of the anticyclonic Saharan air discussed above. On the morning of 27 September, KE increased again between 1000 and 5000 m, in association with the increase of eddy cyclonic vorticity and weak convective activity simulated at that time. Convection was short lived, however, and KE started to decrease after 1200 UTC between 1000 and 3000 m, while it remained relatively large above 3000 m until 28 September. Finally, on the afternoon of 27 September, KE increased between 5000 and 16 000 m due to the enhanced anticyclonic circulation southeast of the midlatitude upper-level trough.
4) Budget of moist enthalpy
The evolution of the moist enthalpy budget is displayed with vertically integrated terms and vertical profiles of the first-order terms [O(100 W m−2) in Fig. 14; O(10−2 W m−3) in Fig. 15]. The Hm tendency (Fig. 15a) is alternatively positive during the day and negative during the night, due to the diurnal cycle of radiative heating Gdiab (Fig. 15e) and the semidiurnal variation of Gp (Fig. 15f) related to the atmospheric tide (e.g., Dai and Wang 1999) and the coupling with the ECMWF analyses every 6 h. The vertical baroclinic conversion term CPV is predominantly positive (Fig. 15c shows negative values of −CPV). Since CPV is equal to the vertical pressure work −w∂p/∂z [see Eqs. (25) and (32) in AR09] and pressure decreases with height, this indicates that the mean vertical velocity was predominantly positive, with a maximum on 27 September during a period of convective activity offshore. Downward motion (positive values of −CPV) then occurred during the night of 27–28 September between 3500 and 9000 m when the oceanic MCS dissipated under the influence of anticyclonic Saharan air to the east and a midlatitude upper-level trough to the north. Since the mean vertical velocity was mostly positive and Hm decreased with heights, the vertical flux divergence −[∂(Hw)/∂z] dominated the boundary term BH (Fig. 15b). Hence, enthalpy produced by diabatic processes (Gphase in Fig. 15f; Gdiab in Fig. 15e) was mostly transformed into vertical kinetic energy through vertical baroclinic conversion (CPV > 0), accompanied with a vertical redistribution of Hm through the mean upward motions. The variations of enthalpy due to humidity changes (GWV in Fig. 15g) were mostly negative below 6000 m due to water vapor condensing throughout Δ, while positive values in the lowest levels revealed surface evaporation at the hottest time of the day (maximum positive values of Gdiab; Fig. 15e). These results are very similar to those concerning the developing pre-Helene disturbance (AR09).
Two second-order terms [O(1 W m−2) in Fig. 16a and O(10−4 W m−3) in Figs. 16b–f] balance the residual of the first-order terms. Here Ghydro (Fig. 16d) is the upward transport of heat by nonprecipitating hydrometeors. It was maximum during the night of 24–25 September and again on 27 September, due to simulated convective activity south of Senegal and over the ocean (Fig. 9), respectively. The most important secondary process is the horizontal baroclinic conversion term CP (i.e., the work of the horizontal pressure force, which can be decomposed into its zonal CPZ and eddy CPE contributions). There were two periods on 24 and 27 September during which the horizontal baroclinic conversion was positive (−CP < 0) below 5000 m and moist enthalpy was transformed into KE + KZ, which is observed for both zonal and eddy baroclinic conversions (respectively CPZ and CPE), in greater proportion for CPZ on 24 September and in comparable proportions on 27 September. Because the primary processes in the budget of moist enthalpy were partly caused by the vertical circulation of convective origin, and convection was associated with mostly hydrostatic pressure decrease induced by latent heat release, we suggest a close relationship between convective activity and horizontal baroclinic conversion through the ageostrophic flow converging toward the low pressure zone. The pressure decrease associated with the oceanic convective development of 27 September was more efficient in generating an ageostrophic circulation and in producing eddy kinetic energy through positive eddy baroclinic conversion, in comparison with the pressure decrease associated with the continental convective event on 24–25 September. Inverse baroclinic conversion occurred on 25 and 26 September below 5000 m, when the mean circulation changed from cyclonic to anticyclonic (Fig. 11) and the ageostrophic circulation was directed toward regions where pressure was increasing.
5) Budget of eddy kinetic energy
Evolution of the vertically integrated terms involved in the budget of KE is shown in Fig. 17, and their vertical profiles are in Fig. 18. The energy budget was computed following AR09, who proposed an alternative to L55 to alleviate conceptual difficulties when considering a limited domain. An important point in AR09’s analysis is that the CPE between enthalpy (used to quantify the thermodynamic reservoir) and horizontal kinetic energy can be expressed as the horizontal pressure work resulting from the ageostrophic circulation. It is therefore difficult to compare the present results with previous ones (e.g., Burpee 1972; Norquist et al. 1977; Thompson et al. 1979; Thorncroft 1995; Paradis et al. 1995; Hsieh and Cook 2007). Instead, we focus on a comparison with the developing case of Helene (2006) analyzed by AR09 with the same energy budget.
On the afternoon of 24 September KE grew through positive barotropic (CK > 0) and baroclinic (CPE > 0) conversions between 1000 and 4000 m, while pressure was increasing (positive ζG was decreasing in Fig. 11). This continental evolution resulted from the adjustment of the wind field to a negative pressure perturbation of convective origin. Here, the eddy ageostrophic circulation was mostly convergent, down the eddy pressure gradient in the low- to midtroposphere, leading to a positive work of the horizontal pressure force (CPE > 0). Consequently, this eddy circulation was filling the depression, which is coherent with the fact that in tropical regions the Coriolis force is not strong enough to rapidly bring such a pressure perturbation into geostrophic equilibrium. This result is similar to that derived by AR09 for the continental stage of the pre-Helene disturbance.
On the morning of 25 September KE was still growing, baroclinic CPE became negative and barotropic CK increased, while pressure decreased then increased quickly (see the opposite tendencies of ζG in Fig. 11) between 1000 and 3000 m. The increasing positive barotropic conversion CK resulted from an intensification of the cyclonic curvature of the easterly flow (see Figs. 7e,f) in the AEW trough. The negative baroclinic conversion CPE starting on the morning of 25 September is attributed to an interaction between the low- to midlevel anticyclonic Saharan air and perturbation D. Negative values of CPE reveal that the wind was mostly directed toward high pressure zones, which resulted from the simultaneous presence of the low pressure zone associated with perturbation D and a strong southeasterly flow west of the anticyclonic Saharan air (Fig. 7f). The positive barotropic conversion is similar to that obtained by AR09 for the oceanic stage of pre-Helene disturbance but not as intense. AR09 also found a negative baroclinic conversion at this stage and related it to a diverging eddy cyclonic circulation, in relation to a high pressure zone west of it. The situation is different during the continental stage of perturbation D, with a converging eddy anticyclonic circulation in the AEW ridge east of it.
On 27 September, the cyclonic eddy circulation of perturbation D in the low and midlevels reinforced. At that time, the production of KE was mainly due to positive baroclinic conversion, while pressure decreased then increased quickly (see the opposite tendencies of ζG in Fig. 11), which suggests that the pressure perturbation of convective origin could not reach geostrophic equilibrium and quickly dissipated, as is generally the case for small and short-lived disturbances in tropical regions (e.g., as happened during the continental evolution on the afternoon of 24 September). In comparison with the oceanic stage of developing pre-Helene disturbance, this failed cyclogenetic evolution resulted from a lack of barotropic conversion in the low to midlevels, although baroclinic growth was stronger during a limited period. For the developing case of Helene, AR09 showed that barotropic conversion at midlevels was associated with a substantial increase of the cyclonic curvature of the AEJ in the AEW trough, partially strengthened by a subtropical high to the northeast. For perturbation D, however, the AEJ had an anticyclonic curvature caused by the enhanced ridge east of the trough, so the zonal flow could not increase the cyclonic eddy circulation of perturbation D.
The large positive baroclinic conversion in the upper levels from the morning of 27 September was associated with the ridge southeast of midlatitude upper-level trough (Fig. 6). The associated anticyclonic eddy circulation was divergent and down the eddy pressure gradient (CPE > 0), which confirms the well-known result that a midlatitude ridge (trough) in nearly geostrophic balance is divergent (convergent). On the other hand, this nondeveloping case study and the developing case study of AR09 show that an AEW trough can be associated with a positive (negative) eddy pressure work, so it can be convergent or divergent, depending on the surrounding synoptic circulation, and accordingly lead or not to tropical cyclogenesis.
4. Summary and perspectives
The so-called perturbation D was a nondeveloping AEW trough observed over the West African coast and the nearby Atlantic Ocean during the SOP-3 of AMMA in September 2006. This system has been sampled with dropsondes on three successive days (25–27 September) with the SAFIRE Falcon-20 aircraft. Composite thermodynamic and dynamic fields retrieved from these data show that perturbation D evolved favorably on 26 September with substantial moistening and increase of cyclonic vorticity in the lower and middle troposphere. This evolution, however, was thwarted on 27 September by the adverse influence of dry, anticyclonic, southeasterly Saharan flow at low and mid levels.
A qualitative analysis based on ECMWF reanalyses and Meteosat-9 images confirmed that perturbation D was an AEW trough associated with a series of convective developments. More precisely, this disturbance was associated with a MCS initiated by a monsoon surge over Ghana on the afternoon of 22 September. This MCS propagated westward with the AEJ, but it dissipated during the night of 23–24 September. Several convective redevelopments occurred, especially near the Guinean highlands on 25–26 September, in association with an increase of the cyclonic circulation in the AEW trough. From 26 September, however, a low- to midlevel flow of Saharan origin increased the anticyclonic circulation of the AEW ridge east of perturbation D. The enhancement of the AEW ridge was also related to a midlatitude upper-level trough approaching the Cape Verde Islands region from the northwest. The increased southeasterly flow east of perturbation D stretched and disorganized its cyclonic vorticity structure, so it finally failed to develop into a tropical depression off shore.
This failed cyclogenetic evolution has been simulated with the nonhydrostatic Méso-NH model using parameterized convection for the period 24–28 September. A MCS in south Senegal on 25 September was well reproduced by the simulation. The associated production of midlevel cyclonic vorticity was slightly stronger than in ECMWF reanalyses. However, the coastal convective activity on the following day was weaker in the simulation. Then, the final convective development on 27 September over the nearby ocean started earlier and was weaker in the simulation. Nevertheless, the synoptic and convective processes were relatively well reproduced in the simulation.
This failed cyclogenetic evolution has then been analyzed with the energy budget of AR09 adapted from L55 for a limited domain. During the enhanced convective activity of the morning of 25 September, the eddy kinetic energy KE increased in the low to midlevels as a result of positive barotropic conversion (CK > 0) and negative baroclinic conversion (CPE < 0), while pressure first decreased and then increased quickly. The increasing positive CK is attributed to an intensification of the cyclonic curvature of the easterly flow in the low pressure system. The negative baroclinic conversion CPE on 25 September is attributed to the low- to midlevel anticyclonic Saharan air that interacted with perturbation D from that time.
During the convective redevelopment over the ocean on 27 September, the production of KE was mainly due to positive baroclinic conversion, while pressure decreased then increased quickly, which suggests that the pressure perturbation of convective origin could not reach geostrophic equilibrium and quickly dissipated. In comparison with the oceanic stage of developing pre-Helene disturbance, this failed cyclogenetic evolution was mostly due to a lack of barotropic conversion in the low to mid levels. The AEJ associated with perturbation D had an anticyclonic curvature due to the enhanced ridge to the east, so the zonal flow could not increase the cyclonic eddy circulation of perturbation D, which is the main difference with the developing pre-Helene disturbance analyzed by AR09. We suggest that strong low- to midlevel positive barotropic conversion is a necessary ingredient for an AEW trough to reach equilibrium following a pressure decrease of convective origin at such low latitudes. More case studies of developing and nondeveloping AEW troughs close to the West African coast should be analyzed with this energy budget to test this hypothesis.
This work was part of the first author’s thesis in Université Paul Sabatier Toulouse III, France. Based on a French initiative, AMMA was built by an international scientific group and is currently funded by a large number of agencies, especially from France, the United Kingdom, the United States, and Africa. It has been the beneficiary of a major financial contribution from the European Community’s Sixth Framework Research Programme. Detailed information on scientific coordination and funding is available on the AMMA International Web site, http://www.amma-international.org. The AMMA SOP-3 field campaign was funded by Centre National d’Etudes Spatiales (CNES, French space agency). Pilots and technical crews (SAFIRE: UMS 2859 CNRS/Météo France/CNES) of the Falcon must be thanked for their cooperation and dedication. Numerical simulations were conducted on CNRS/IDRIS computers under Grants 070591 and 080591. The ECMWF reanalyses used in the Méso-NH simulation were provided by Dr. Anna Agusti-Panareda. We thank Didier Gazen and Juan Escobar for the technical support in Méso-NH and Dr. Jean-Pierre Chaboureau for the scientific support in Méso-NH.
Corresponding author address: Frank Roux, Laboratoire d’Aérologie, Observatoire Midi-Pyrénées, 14 avenue Belin, 31400 Toulouse, France. Email: firstname.lastname@example.org