Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
390 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 62 Extratropical Dry-Air Intrusions into the West African Monsoon Midtroposphere: An Important Factor for the Convective Activity over the Sahel RÉMY ROCA Laboratoire de Météorologie Dynamique, École Normale Supérieure, Paris, France JEAN-PHILIPPE LAFORE, CATHERINE PIRIOU, AND JEAN-LUC REDELSPERGER Centre National de la Recherche Météorologique, Toulouse, France (Manuscript received 3 October 2003, in final form 28 June 2004) ABSTRACT This paper investigates the relationship between large-scale dynamics, water vapor, and organized convection over West Africa. Making use of a simplified condensation hypothesis, a back-trajectory model fed by NCEP-analyzed winds is used to reconstruct the midtropospheric humidity field over Africa during July to August 1992. The approach documents both the moisture content and the origin of the air masses. Meteosat satellite infrared imagery is used to characterize the convective systems. A case study analysis reveals that very dry air patches (RH ⬍ 5%) are located in the immediate midtropospheric environment of a typical squall line. Such dry-air structures are shown to originate in the upper levels (200–250 hPa) on the anticyclonic side of the polar jet stream at 50°N. Focusing on the Sahel region, dry events are isolated using the time series of the 500-hPa relative humidity distribution during the monsoon period. These dry events are shown to be composed of extratropical air. Composite analysis of the convective activity indicator exhibits a strong positive association between dry intrusions and convection on the eastern side of the Sahelian region. Organized convective systems that are fast moving and long lasting are more likely over this region when a dry intrusion is present. This coincides with the well-established theory that midtropospheric dry air, when combined with sufficient wind shear, can maintain and intensify previously triggered deep convection through rain evaporation that feeds the cold pools, especially within squall lines. This paper suggests that the extratropical dry-air intrusions modulate the occurrence and duration of convective systems and, therefore, the mode of variability of rainfall over West Africa during the monsoon. 1. Introduction Strong variability of free tropospheric humidity has been observed over the Pacific Ocean during the Tropical Ocean Global Atmosphere Coupled Ocean– Atmosphere Response Experiment (TOGA COARE) campaign with extremely dry layers found within regions that are usually convectively active (e.g., Mapes and Zuidema 1996). These dry-air tongues were shown to originate from the extratropical upper levels yielding to the notion of extratropical dry-air intrusions in the deep convective regions (Yoneyama and Parsons 1999; Parsons et al. 2000). Locally, dry air inhibits deep convection through two processes: first, the radiatively built thermal inversion due to the dry layer that pre- Corresponding author address: Dr. Rémy Roca, Laboratoire de Météorologie Dynamique, École Normale Supérieure, 24 rue Lhomond, 75005 Paris, France. E-mail: [email protected] © 2005 American Meteorological Society JAS3366 vents the development of convective clouds and second, the entrainment of dry air as the parcel rises that decreases the parcel’s buoyancy (Mapes and Zuidema 1996; Sherwood 1999; Parsons et al. 2000; Redelsperger et al. 2002b). Both processes control the cloud-top altitude. It has been suggested that these dry-air intrusions play an important role in the synoptic and intraseasonal variability of deep convection in the warm pool region (Brown and Zhang 1997; Johnson et al. 2001). While such phenomena received a great deal of attention in the TOGA COARE community, their distribution over the Tropics is still unclear. For instance, the Indian Ocean Experiment (INDOEX) sounding analysis revealed the existence of dry-air layers that originate from extratropical upper-level flows over the Indian Ocean during the boreal winter (Zachariasse et al. 2001); these dry structures have also been seen over the Atlantic Ocean in the summer (Pierrehumbert and Roca 1998). In the present paper, we investigate the humidity distribution of the environment of deep con- FEBRUARY 2005 ROCA ET AL. vection in the midtroposphere over the West African monsoon region and show the existence of extratropical dry-air intrusions. Over West Africa a number of processes govern continental convection, ranging from the role of topography in the triggering phase to the role of wind shear for moving convective lines, including the diurnal cycle, etc. Much of the Sahelian seasonal rainfall is produced by organized mesoscale convective systems (Lebel and Lebarbé 1997; Mathon and Laurent 2001). The environmental midtropospheric water vapor content of the West African squall lines has been investigated during the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE). Composites of the environment of squall lines were formed using soundings and by separating slow- and fast-propagating lines. Barnes and Sieckman (1984) showed that fast lines propagated in an environment where the midtropospheric equivalent potential temperature (e) was smaller than that for the slow lines, and where the wind shear was significant and normal to the fast lines while it was parallel to the slow-moving cloud lines. This midtropospheric dry environmental air, together with the wind shear, was further shown to be an important element of the mesoscale circulation associated with convection. The evaporation of hydrometeors in the dry-air layer acts as a fuel for the density current that drives the new convective cells’ development (Redelsperger and Lafore 1988; Lafore and Moncrieff 1989) and favors intense and long-lived systems, if the lowlevel shear is strong enough, according to the Rotunno et al. (1988) theory. More recently, Diongue et al. (2002) conducted numerical simulations of the 21 August 1992 squall line observed during HydrologyAtmosphere Pilot Experiment in the Sahel (HAPEXSahel). Their analysis indicates a strong sensitivity to moisture patterns in the free troposphere of the initial state of the simulations, further emphasizing the importance of the structure of water vapor in the environment of the squall line for its development. This suggests a specific relationship between dry-air intrusions and convection over West Africa that is also investigated in this paper. The monsoon season of 1992 is selected for the study because it is a well-documented period corresponding to the HAPEX-Sahel campaign (Goutorbe et al. 1994) and was shown to be an typical monsoon season with respect to rainfall variability (Redelsperger et al. 2002a, hereafter R02). The paper is structured as follows: section 2 presents the data and methodology used to reconstruct the water vapor content and the origin of midtropospheric air parcels. Section 3 presents a case-study analysis of a strong convective event that occurred on 21 August 1992 and how dry-air intrusions from the extratropics occur in the midtroposphere of its environment. Section 4 is dedicated to the entire 1992 African monsoon period and an analysis of the link between dry intru- 391 sions and convection. Finally, a summary and discussion are given. 2. Data and method a. Satellite observations 1) INDICATORS OF CONVECTION ACTIVITY AND ORGANIZATION Deep convection over West Africa spans a wide range of degrees of organization: local convection, slow- and fast-moving squall lines, stationary systems, and families of mesoscale convective systems (MCSs) that sometimes become as large as mesoscale convective complexes (MCCs; Laing and Fritsch 1997; Mathon and Laurent 2001). This hierarchy of systems is presently analyzed using a cloud-tracking technique developed at the Centre National de la Recherche Météorologique (CNRM) applied to Meteosat infrared imagery. The Instrument de Suivi de l’Imagerie Satellitale (ISIS) algorithm is an automated method for MCS identification and tracking developed by Morel and Senesi (2002a) to derive a climatology of MCSs over Europe using IR imagery (Morel and Senesi 2002b). It allows organized convective systems (e.g., fast-moving, long-lasting systems) to be separated both in space and time from other convective events (e.g., short-lived and stationary systems). Here the tracking takes place from July to August 1992, using half-hourly full-resolution Meteosat IR data. This description of convection is further completed by a simple convective activity index built from the full-resolution imagery (5 km at nadir) consisting of the number of pixels colder than a given threshold in a 0.625° ⫻ 0.625° grid. This cold-pixel surface is computed for an IR threshold of 253 K, which represents the upper-level cloudiness in the Tropics (e.g., Duvel 1988; Roca and Ramanathan 2000). The National Oceanic and Atmospheric Administration (NOAA) outgoing longwave radiation (OLR) 2.5° ⫻ 2.5° daily interpolated fields (Liebmann and Smith 1996) are further used to document the seasonal variability of convection over West Africa. 2) WATER VAPOR IN THE MID- TO UPPER TROPOSPHERE Mid- to upper-tropospheric water vapor is documented using the Meteosat observations in the water vapor channel (5.3–7.1 m). This channel is centered over a strong water vapor absorption band and, in clear skies, is sensitive mainly to the mid- to uppertropospheric moisture. Following the approach of Soden and Bretherton (1993) and Schmetz et al. (1995), the brightness temperatures are inverted in terms of the mean relative humidity of the clear column weighted by the captor weighting function. This free tropospheric relative humidity (FTH) corresponds to the whole free troposphere, although it is the most sensitive to the 392 JOURNAL OF THE ATMOSPHERIC SCIENCES mid- to upper troposphere (600–200 hPa). Roca et al. (2003) presents the technique in detail. Figure 1 shows the July–August mean-FTH distribution. The very dry areas (FTH ⬍ 10%) span a large portion of the region and correspond with the large-scale subsidence regions (Picon and Desbois 1990). A strong gradient separates the moist ITCZ (FTH ⱖ 40%) from the dry areas. These observations offer an estimate of the average tropospheric humidity but not its vertical structure. Processing the satellite data are hence completed by an estimation of the 500-hPa water vapor field as discussed below. b. Reconstruction of the midtroposphere water vapor field Investigating the water vapor distribution over Africa using atmospheric analysis and reanalysis remains a challenge (e.g., Schmetz and van de Berg 1994; Salathé and Chesters 1995; Vesperini 2002). For instance, the European Centre for Medium-Range Weather Forecasts (ECMWF) 15-year Re-Analysis (ERA-15) fields were compared to Meteosat water vapor data for the summer 1992 and shown to suffer from a longitudinal shift in the tropospheric dry structures that impact initial conditions for mesoscale simulations (Diongue et al. 2002). In this data-void region, only a few soundings are available and assimilated by the ECMWF and the National Centers for Environmental Prediction (NCEP) making it difficult for a model to reproduce details of the moisture field. Note that recent advances in the assimilation of satellite moisture data have yielded much better results in operational analysis systems (Hólm et al. 2002). In order to investigate both the water vapor distribution and the origins of air masses over West Africa, we use an idealized advection– FIG. 1. Mean FTH (%) derived from 3-hourly Meteosat water vapor data for Jul–Aug 1992. VOLUME 62 subsidence model (Pierrehumbert 1998; Pierrehumbert and Roca 1998, hereafter PR98) that allows for the reconstruction of water vapor fields in the clear-air region of subsidence using back trajectories. The model is fully described in PR98 and references therein; so, here, we only give a brief overview of its functions and underlying assumptions, as well as the slight changes made to the original approach. This model belongs to the so-called last temperature of saturation models to estimate the water vapor distribution (Held and Soden 2000). This fully Lagrangian transport scheme relies on analyzed winds and simplified moisture source/sink parameterizations. No diffusion is considered, neither are the complex processes associated with the cloud microphysics of the moistening/drying of the air mass. It is assumed that the entire moisture source is located in a saturated boundary layer and that deep convection transports saturated air upward where it is detrained and made available for the large-scale flow to transport away from this source region. The transport scheme is run as a reverse domain-filling technique (e.g., Sutton et al. 1994) using back-trajectory computations and allows the reconstruction of the humidity field at any given time and level in the atmosphere. The NCEP-analyzed winds (Kalnay et al. 1996) are used to compute the 3D back trajectories. Instead of diagnosing the vertical velocities from the horizontal winds using the continuity equation, as in PR98, the vertical velocities from NCEP are retained. While the reconstructed field is slightly noisier in the latter case, the major features of the recomputed moisture field, as well as the ensemble trajectories, are not greatly changed. Comparisons with another trajectory model run with ECMWF winds (P. Peyrillé and J. P. Cammas 2002, personal communication) also did not modify the large-scale patterns we consider in this study. The trajectories are run backward in time until they encounter deep convection. Instead of using ancillary information, like a global satellite archive, to determine convection it is simply diagnosed using the NCEP precipitation field. Not only is this an easy procedure for isolating active convection, it also insures consistency between the wind field and the convection indicator, although it does suffer from the limitations of the analyzed precipitation field. A threshold of 1 cm day⫺1 is used to determine when convection occurred. At this point, the air mass is considered saturated and its mixing ratio is set to the saturation mixing ratio. Furthermore, the mixing ratio along the trajectory is reset to the minimum between the saturation mixing ratio and the current mixing ratio, the excess moisture implicitly being rained out. As a result, the mixing ratio at a given point corresponds to the saturation mixing ratio at the coldest point encountered along the back trajectories. This accounts for the cold-air processing of the air mass in the extratropical upper levels. Once the mixing ratio of the individual air mass is obtained, relative humidity with respect to water is computed using the NCEP- FEBRUARY 2005 ROCA ET AL. analyzed temperature profiles. The model is integrated at the 500-hPa pressure level using a resolution of 0.5° ⫻ 0.5° over the studied region. As discussed in PR98, the individual level reconstructed field exhibits small-scale filamentary structures that are not discerned in the Meteosat imagery. These small-scale structures result from the advection of a passive scalar by the large-scale flow. While the existence of these very finescale filaments (100 km wide) is difficult to establish, the large dry patterns that we focus on appear as robust features of the water vapor field. Indeed a sensitivity analysis of the time of back advection indicated that the features to be discussed next appear and remain in the recomputed field from 5 to 25 days. The smaller-scale features are more sensitive to the integration time and becomes finer as the advection time increases. A trade-off between the filaments’ resolution and the central processing unit (CPU) time to perform the computations has been made. As a result, the back trajectories are integrated over 12.5 days. c. Evaluation of the reconstructed water vapor field On average, during July–August 1992, the reconstructed 500-hPa relative humidity field exhibits largescale structures similar to both the ECMWF ERA-40 reanalysis and the NCEP reanalysis (Fig. 2). The three estimates qualitatively resemble the Meteosat-derived FTH field (Fig. 1) characterized by strong latitudinal gradients and a dry region located over the eastern Mediterranean Sea as well as a moist ITCZ. Overall, ERA-40 exhibits a moister distribution than NCEP and the reconstructed field over the Sahara. The meridional gradients between the moist ITCZ and the dry central Sahara are weaker in ERA-40 than in the two other analyses. The Lagrangian-reconstructed moisture field shows very moist conditions (RH ⬎ 80%) in the ITCZ that are not seen in any other analysis. This is due to the simple saturation assumption, which strongly moistens the convective regions. Quantitative evaluation of the product is conducted 393 using soundings performed at 1200 UTC during July– August 1992 over a dry region at Tamanrasset, Algeria (22.7°N, 5.5°E) as well as over a typical Sahelian region at Niamey, Niger (13.48°N, 2.17°E). Figure 3 shows the scatter diagram of the difference between the NCEP analysis and the radiosonde RH and between the reconstructed relative humidity field and the radiosonde RH at 500 hPa. Overall the NCEP RH is biased moist by 5% over Tamanrasset and dry by 10% over Niamey, while the reconstructed RH is biased dry by around 10% in both cases. In the dry regimes (RH ⬍ 20%), NCEP bias is positive at around 10% in both regions while the reconstructed field bias changes sign and reaches 3.1% and 6% for Tamanrasset and Niamey, respectively. In summary, the reconstructed water vapor field agrees reasonably well with the radiosonde observations in the dry region over the Sahel. Unlike the analysis, such reconstruction gives access to smaller scales, which are of interest to our study, with consistent quality over the entire African subsidence regions. It is independent of the assimilation of moisture information from the soundings. These comparisons complete and confirm the results of Pierrehumbert (1998) and PR98 concerning the ability of this approach to yield realistic and consistent water vapor content in the free troposphere subsidence areas of the Tropics. 3. A case study of the 21 August 1992 squall line The strongest convective event of the 1992 season over the Sahel started on 21 August 1992 (R2002). The squall-line life cycle is described as follows: individual cells form over the Aïr Mountains late in the morning of the 21 August and they slide down the slope of the topography in a southwesterly direction. The individual cells organize into a well-shaped squall line that propagates to the west at a speed of around 17 m s⫺1. The mature stage of the system is reached at around 1600 UTC and the squall lasts through the next morning (Diongue et al. 2002). FIG. 2. Maps of Jul–Aug 1992 mean RH (%) at 500 hPa. (a) ECMWF ERA-40, (b) NCEP, and (c) the reconstructed humidity field. 394 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 62 FIG. 3. Scatter diagram of the difference between analyzed and radiosonde 500-hPa RH as a function of the radiosonde RH (%) over (top) Tamanrasset and (bottom) Niamey. (a), (c) NCEP and (b), (d) the reconstructed field. The vertical dashed line delimits the dry regime (RH ⬍ 20%). Large-scale and regional context of the case study The West African monsoon (WAM) region exhibits a number of specific local and large-scale circulations features. These features are important to the understanding of the water vapor transport during the days preceding the squall line and are summarized in Figs. 4 and 5. The tropical easterly jet (TEJ) is located at latitudes around 5°N with a zonal extension across the whole region of study. Other features of interest include the subtropical jet (STJ) located over the northern part of the Himalayas at latitudes around 40°N. The polar jet stream is located over the North Atlantic Ocean at a latitude of 50°N. The jet core is centered at 50°W and the jet extends eastward up to 10°E. These upper-level westerlies spread across a wide region in latitudes down to 20°N over the Atlantic Ocean and West Africa. Another interesting structure is the local FIG. 4. Averaged 200-hPa circulation over 11–21 Aug 1992 from NCEP. Contours show the wind module (m s⫺1). Topography above 750 m is shaded. FEBRUARY 2005 ROCA ET AL. 395 FIG. 5. Zonal circulation for 11–21 Aug from NCEP averaged over the 10°W–5°E longitude band. Thin contours show potential temperature (K) and thick contours show the zonal wind module (m s⫺1). anticyclonic circulation centered over 25°N, 30°E. The zonal mean vertical slab over the 10°W–5°E band of the zonal wind, together with the potential temperature, shown in Fig. 5 denotes these structures in the vertical. The polar jet stream core is located at 250 hPa around 55°N with a maximum velocity of 26 m s⫺1. The entry of the STJ is centered at 200 hPa around 30°N with a smaller velocity (13 m s⫺1). The TEJ is limited to the 150–250-hPa altitude with a velocity around ⫺18 m s⫺1. The African easterly jet (AEJ) is located at 15°N with its core centered on 600 hPa with a velocity of ⫺9 m s⫺1. Another important feature of the West African climate is the deep dry convective layer over the Sahara, which extends equatorward to Sahelian latitudes. This deep mixed layer reaches up to a 6-km height over some regions (Gamo 1996) and is believed to complement the role of the ITCZ in maintaining the AEJ (Thorncroft and Blackburn 1999). This mixed layer prevents us from investigating the water vapor transport below 500 hPa with the present methods because of the difficulties in establishing meaningful trajectories in the strong mixing regimes. Before the squall-line occurrence, the deep mixed layer extends up to 600 hPa over the 20°–30°N region (Fig. 5). This large-scale context determines the water vapor distribution in the regional environment of the squall line. Over the 12.5°–17.5°N band where convection occurs, the whole 10°W–5°E region is very dry during the day preceding the squall line, while in the eastern part of the region, moist conditions associated with active convection prevail to the south of the region (Fig. 6a). Extratropical air (at a latitude greater than 40°N) intrudes into the region on the night of 20 August, then it spreads into the region during the 21st, reaching a maximum during the squall-line mature phase at 1800–2400 UTC (Fig. 6b). To the north of the region (17.5°– 22.5°N), extratropical dry-air conditions prevail up to 10°E from the 20th (0000 UTC) to the 22d (0000 UTC). It is then interrupted by a moist plume (Fig. 6c) that originates from the squall line as evidenced by the low latitudes of origin (Fig. 6d). Accordingly, we investigate the water vapor distribution and origins at 1800 UTC on 21 August when the squall is fully developed. Figure 7 shows the Meteosat-derived FTH map. The squall line is clearly seen at 15°N, 5°E as well as the large-scale dry zones surrounding it. The large-scale dry features (FTH ⬍ 20%) are separated by a moist plume extending from the convection to the northeast of the 396 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 62 FIG. 6. Hovmöller diagram around 21 Aug 1992: (a) reconstructed RH, and (b) maximum latitude of air mass at 500 hPa averaged over the 12.5°–17.5°N latitude band. (c), (d) Same as (a), (b) at 500 hPa but averaged over the 17.5°–22.5°N latitude band. Isolines correspond to the fraction of Meteosat pixels colder than 253 K. FEBRUARY 2005 ROCA ET AL. 397 FIG. 7. Meteosat-derived FTH (%) for 1800 UTC 21 Aug 1992. Note that cloudy regions are white and that the estimate could be contaminated by thin cirrus on the edges of the convective systems. region previously mentioned. The satellite product indicates relatively dry regions, both ahead and behind the squall line, that are better characterized using the reconstructed moisture field. The 500-hPa relative humidity map indeed exhibits a number of dry features (Fig. 8a). The first feature (referred to here and later as A) is located over the 12°–20°N, 8°W–0° region on the western flank of the squall line and corresponds to a large dry region almost 800 km wide. Relative humidity is as low as 5% and does not exceed 15% to 20%. The second dry spot (referred to as B) is located on the eastern side of the squall line around 15°N. The relative humidity is greater than in the previous dry tongue with values ranging from 5% to 30%. The horizontal distribution of relative humidity in this second tongue is less homogenous than in the first one. A moist plume separates the dry region, in qualitative agreement with the satellite information (Fig. 6). The large-scale mixing between the Tropics and the midlatitudes is demonstrated in Figs. 8b and 8c, showing the spatial distribution of the maximum latitude encountered by the particles during their flights. The moist plume is associated with low-latitude air exported from the Tropics toward the northeast. The dry region ahead of the squall line originates north of 60°N. The longitude coordinates of the air particles when they reach their maximum latitude during their flight is shown in Fig. 8c. Overall, the western part of the region appears to be fed by western air masses while the opposite holds true for the eastern region. However these two-dimensional views do not reveal the whole path of air particles which is now assessed for the two key dry features (A and B) west and east of the squall line, respectively. Figure 9 shows an ensemble of back trajectories for these two regions. The air masses feeding the region (A; see Fig. 9a) all originate from the polar jet stream between 300 and 200 hPa, which explains their relative dryness. A close examination of the trajectories’ position relative to the polar jet (Fig. 4) indicates that they always stay on the southern anticyclonic flank of the jet centered at 50°N. This is confirmed by the fact that these air masses exhibit low potential vorticity (PV), characteristic of the upper troposphere, all along their trajectories (not shown). The air masses are transported eastward within the polar jet up to around 0° (longitude) where they make a strong southward turn and start to subside. They take around 5 days to subside to 500 hPa at a rate of 40–60 hPa day⫺1, which is in agreement with magnitudes usually quoted for tropical subsidence. Note that the ensemble of 25 trajectories spanning a 2.5° ⫻ 2.5° region remains very compact during the 5 days of subsidence indicating the robust- 398 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 62 ness of this feature. On the eastern side, the situation is less clear. The plume from the 25 trajectories reveals three regions of origin for the dry feature (B; Fig. 9b). One corresponds to the local anticyclonic circulation seen in Fig. 5, where particles take up to 12 days to recirculate from the upper level of the region north of the ITCZ to east of the squall line. The air masses originate from the 300-hPa level and subside slowly. The second region of origin is from the eastern part of the analyzed area and is characterized by a compact flow possibly associated with the upper part of the AEJ. Owing to their lower altitudes of origin and their history, these particles have a greater water vapor content than the previous ones. The third region only concerns a few back trajectories originating for the extratropical upper level (200 hPa) with subsiding characteristics similar to the one from the region (A) ahead the squall line. In summary, this case study reveals a large signal of intrusion of extratropical dry air originating from the anticyclonic side of the polar jet stream and extending down to the midtroposphere in the environment just ahead of a typical WAM fully developed squall line. The occurrence of these intrusions and their links with convection for the whole monsoon season is investigated next. 4. Dry-air intrusions and convective activity during the 1992 monsoon The 1992 season corresponds to the HAPEX-Sahel campaign (Goutorbe et al. 1994) and was shown to be a typical monsoon with respect to the rainfall variability over Sahel including synoptic (3–5 days) and longer (10–25 days) variability (R02). First we present the mean structure of convective activity and airmass origins. Then, we analyze the relative humidity variability and link it to the large-scale dynamics as well as to the convective activity. a. Mean structure of convective activity and of airmass origins FIG. 8. Map of (a) relative humidity, (b) maximum latitude, and (c) longitude at maximum lat for 1800 UTC 21 Aug 1992, at 500 hPa. Areas where Meteosat IR temperature is colder than 273 K are indicated with black horizontal lines. Relative humidity for initially ascending trajectories are not reported and are shown in white. During the monsoon, the convectively active regions form a zonal structure corresponding to the ITCZ (shaded area in Fig. 10). Two centers of high-convective activity are located over the eastern part of the domain (15°E) at a latitude of around 7.5°N and over the Atlantic Ocean west of Guinea, respectively. The OLR variability (isolines in Fig. 10) is at its highest on the northern flank of the ITCZ centered on the large gradient of convective activity. The maximum variability (standard deviation of 35 W m⫺2) appears over the Sahelian region (12.5°–17.5°N, 10°W–5°E). Figure 11 shows the characteristics of the convective systems during the season. All the systems contribute up to 50 h month⫺1 of occurrence over the centers of convection previously highlighted (Fig. 11a). The distribution of FEBRUARY 2005 ROCA ET AL. 399 FIG. 9. Three-dimensional view of an ensemble of back trajectories initially located at 500 hPa centered at (a) 15°N, 6°W and (b) 16°N, 18°E. The color scale indicates the duration of the air mass along the trajectory in days since 1800 UTC 21 Aug 1992. the systems with a life cycle longer than 12 h (Fig. 11b) corresponds to the region where the convective variability is at its greatest (Fig. 10). These long-lasting systems make up more than 60% of the convective activity over the Sahel. Figure 11c shows that the mean westward speed of these systems reaches up to 16 m s⫺1 on average. Long-lasting systems are less common on the southern flank of the ITCZ and move more slowly. 400 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 62 acterized by two bimodal behaviors: dry air can originate from tropical/extratropical regions (Pierrehumbert 1998) and dry and moist air coexists at small and large scales (Zhang et al. 2003). In the present case over the Sahel, for example, dry-air structures that interact with convection can originate from the upper levels either from the Tropics or from the PJ. Similarly, saturated moist air from local convection can be found simultaneously all over the region with dry air (e.g., Fig. 8a). This strong bimodality in water vapor and origins precludes the use of a simple regional average to characterize the relative humidity distribution and its variability. Defining quantitatively what could be a dry intrusion remains a difficult task that is attempted in the next section. FIG. 10. Map of the mean Jul–Aug 1992 OLR (W m⫺2). Darkest shading is 180 W m⫺2, and shading bands correspond to a 10 W m⫺2 increment. Isolines show standard deviation of the daily OLR over the period. The black rectangle highlights the key Sahelian region discussed in the text. These findings are in agreement with previous climatological studies (Lebel and Lebarbé 1997; Mathon and Laurent 2001) and confirm the typical characteristics of the 1992 season (R02) for the MCS distribution. Thus, over the Sahelian region convective activity is dominated by organized fast-moving convective systems, lasting more than 12 h, presumably squall lines. Note that this region corresponds to the mean position of the AEJ (Fig. 5) resulting in strong low-level wind shear underneath, which, with dry air is a key element for the fast-moving squall-line occurrence. Figure 12 shows the spatial distribution at 500 hPa of the airmass fraction with an extratropical origin (the maximum latitude encountered north of 40°N) for July–August. The distribution of the extratropical contribution is mainly meridional with a steady increase poleward from 10% at 12.5°N up to 80% at 35°N. Nevertheless, a strong nonmeridional structure is centered at 25°N, 10°E. This commalike feature indicates that more than 70% of the air mass over Tamanrasset at 500 hPa originates from the midlatitudes. The 20% isoline extends south down to 14°N over Sahel, where the variability in convection is at a maximum, as indicated by the OLR variance overlaid in Fig. 11. Over this region, these extratropical air masses originate from the southern anticyclonic flank of the PJ at 50°N, 250 hPa (Fig. 13). On average, these trajectories lie almost on the isentropic surfaces starting at the 330–340-K level at the jet altitude and reach the Sahel region over the 325– 330-K level where strong low-level wind shear prevails below the AEJ, about 5 days later. The other regions where the Sahelian midtropospheric air masses originate include the tropical latitudes with both midtropospheric and upper-tropospheric sources. This large-scale mixing influences the distribution and variability of the Sahel midtroposphere’s relative humidity. Indeed, in the Tropics, water vapor is char- b. Variability of the midtropospheric relative humidity and airmass origins Considering the 12.5°–17.5°N, 10°W–5°E Sahelian region, a dryness index is built as the percent of the grid points with RH below 5%. The mean value of the index is 15% and its standard deviation is 11%. The index varies from 0% to 55% during the season with no clear seasonality (Fig. 14a). Figure 14b also shows the time series of the fraction of extratropical air over the same Sahelian region. The fraction ranges from 0% to 100% over the season and exhibits a smooth seasonal cycle with its greatest point toward the end of July. By defining the dry (moist) event as conditions where the dryness index is greater (lower) than the seasonal mean plus (minus) one standard deviation, one can further inspect the link between dryness and origin. All of the dry events are associated with a positive seasonal anomaly of the extratropical fraction and all the moist events are associated with a negative anomaly. On the other hand, not all the positive (negative) anomalies in the extratropical fraction are associated with a dry (moist) event. In particular, the 18 July event (Fig. 14b) does not produce a strong positive anomaly in the dryness index. Although this extratropical intrusion is associated with dry air, it is not dry enough to show up in the dryness index. The case study of 21 August is associated with a 2-day dry intrusion where the dryness index reaches 40% even though the extratropicalfraction anomaly is modest and involves only a third of the region. The 23 July case, on the other hand, yields a similar value of the index, with twice as much of an extratropical fraction as that for the case study. These examples remind us of the complex variability associated with the water vapor and the airmass origin and confirm that the dry events are associated with intrusions of extratropical air in the region. c. Dry-intrusion events and convective activity In order to link relative humidity and convective activity, we need to account for the strong diurnal cycle of convection observed over West Africa. Our case-study FEBRUARY 2005 ROCA ET AL. 401 FIG. 11. Convective system distribution and characteristics for the Jul–Aug 1992: (a) occurrence of all systems (h month⫺1), (b) occurrence of systems lasting more than 12 h (h month⫺1), and (c) mean westward speed (m s⫺1) of systems in (b). analysis reveals that dry air from the extratropics exists over the Sahel for a whole day before the squall line occurs (Fig. 6). So we propose to define a dry-intrusion (moist exhaust) period as the 24 h period starting from 0800 UTC and continuing to the next morning (0800 UTC) when a dry (moist) event occurs; 0800 UTC corresponds to the time when convective activity is minimum in this region. Using this definition, the shadings 402 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 62 FIG. 12. Spatial distribution of the 500-hPa airmass fraction (%) with an extratropical origin (max lat ⬎ 40°N) over Jul–Aug 1992. The OLR variance from Fig. 10 is overlaid. in Fig. 14 highlight the eight dry-intrusion periods (dark) and seven moist-exhaust periods (light) that are found during the 1992 season. The dry intrusions last from 1 to 4 days and the moist exhausts from 1 to 3 days. The moist exhausts exhibit more stable structures since they are always associated with almost no extratropical fraction and less than 7% of the region being dryer than 5%. These moist exhausts could be related to tropical surges over western North Africa as discussed by Knippertz et al. (2003), but a detailed analysis of their fate is beyond the scope of this paper. Using these two types of events, the composite of Meteosat cold-cloud pixels is built for days with dry intrusions and days with moist exhausts. In the dry-intrusionevents composite (Fig. 15a), the cold-pixel fraction indicates strong convective activity northeast of the region while the moist composite (Fig. 15b) shows an opposite pattern with maximum convection on the west side of the region. The difference between the dry and the moist events (Fig. 15c) shows that the eastern reinforcement of convection is about 10%–15% in coldpixel fraction. The western decrease reaches also 10%– 15%. This is a significant modulation of the convective activity, which in terms of cold-pixel fraction, reaches about 30%. The composite analysis is further down- scaled to the individual convective systems scale. When dry intrusions are present over the Sahel, the systems occurrence is enhanced on the eastern part of the region (Fig. 15d), and decreased in the south and west of the region (Fig. 15e) in a similar way to the cold cloudpixel composite. South of 13°N all types of convective systems occur less frequently during dry-intrusion events. The next dry phase composite indicates a strong occurrence of the long-lasting systems (with life cycles greater than 12 h) on the eastern Sahel from 5°W to 5°E (Fig. 15g) over the region where, on average, these systems are more frequent (Fig. 10). The moist phase is associated with an increase of the long-lasting systems occurrence on the western Sahel and south of the Sahel (Fig. 15h). As a result, the composite of the different dry-intrusion minus moist-exhaust periods (Fig. 15i) shows a strong reinforcement of the long-lasting, fastmoving systems on the eastern Sahel during dryintrusion events, which explains a large part of the mean distribution seen in Fig. 10. There between 5°W and 5°E, the deep convection overall positive anomaly during dry-intrusion events appears to be mainly due to the long-lasting systems’ positive anomaly. Changing the threshold of the dry-intrusion selection from 5% to 10% and keeping the same definition does not alter FEBRUARY 2005 ROCA ET AL. 403 FIG. 13. Zonal circulation and thermal structure of the atmosphere from NCEP averaged over the 10°W–5°E longitude band and the Jul–Aug period. Thin lines correspond to potential temperature (K) and thick lines to the zonal wind module (m s⫺1). Shading represents the trajectories’ density in arbitrary units for the air mass with trajectories ending at 500 hPa in the Sahelian target region (12.5°–17.5°N, 10°W–5°E). The density is integrated over the 45°W–45°E longitude domain. The lighter shading corresponds to the higher density. the patterns discussed above or the magnitude of the signal. It appears that the relationship between dry intrusions and convective activity over West Africa is more complex than previously found over the west tropical Pacific where the inhibiting effect of dry air is dominant (Sherwood 1999; Parsons et al. 2000), and where dryintrusion events are associated with inactive phases of convective activity (Brown and Zhang 1997). Over the eastern Sahel, dry-intrusion events are associated with more long-lasting and fast-moving MCS activity than during moist exhausts; on the other hand, over the western Sahel convective activity is diminished overall by dry-intrusion events. Thus, the intense variability of the convective activity over the Sahel region seems to be related to the dry-intrusion occurrence that is associated with the long-lasting and fast-moving convective systems. 5. Summary and discussion Using a Lagrangian water vapor transport model, the moisture distribution of the midtroposphere over West Africa has been investigated together with the occurrence of the convective systems. The new findings of this study are the following: • Dry air structures (RH ⬍ 5%) exist in the Sahelian midtroposphere and originate from the anticyclonic flank of the polar jet stream. • These dry-intrusion events are associated with longlasting convective systems, likely squall lines, over the eastern Sahel. This highlighted relationship between convection and dry intrusions is not easy to explain and relies on the following physical reasons. Midtropospheric humidity and low-level wind shear have been shown to be key 404 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 62 FIG. 14. Time evolution over the 12.5°–17.5°N, 10°W–5°E region: (a) the surface fraction of the grid points with RH (%) below 5%. The horizontal thick line is the mean and the thin lines the mean plus or minus one standard deviation. (b) The extratropical (max lat greater than 40°N) fraction over the region (%). The thick line is the seasonal cycle corresponding to the first two harmonics of the filtered signal. The dark shading indicates dry-intrusion periods and the light shading the moist-exhaust periods. See text for details. ingredients for intense and long-lived convective systems. This is well supported both by observations (Zipser 1977; Barnes and Sieckman 1984) and by modeling studies (Redelsperger and Lafore 1988; Lafore and Moncrieff 1989; Rotunno et al. 1988) resulting in conceptual models and a theory. One important characteristic of the Sahel region during the monsoon period is that both the convective available potential energy (CAPE) and the convective inhibition (CIN) are large (Diongue et al. 2002), so that specific processes are needed to trigger and maintain the convection. As extensively developed here, this region is affected by midlevel dry-air intrusion events and by strong lowlevel shear induced by the AEJ fulfilling the necessary conditions for strong and long-lived convective systems. Based on these considerations we propose the following two-step scenario. First, a combination of forcing mechanisms such as topography, strong diurnal cycles over land, easterly waves, etc., allows the onset of local convection within an environment that is dry in the midtroposphere. The forcing is supposed to be strong enough to break the CIN and it is further assumed that enough CAPE is available for these cells to develop (see Diongue et al. 2002 for a discussion of the earlier stage of the 21 August 1992 case). In the second step, these cells organize themselves into a convective line that turns into a mature squall line because of favorable conditions (CAPE, shear and dry air) encountered over the region. At this time hydrometeor evaporation in the dry midtroposphere fuels a density current and favors the new cells’ development. Due to a good balance between the low-level shear and the cold pool, the latter can become large, widespread, and well-organized so that it can supply long-lived convection even in regions of strong CIN as well as during the night when convective instability (or CAPE) is reduced. This scenario differs from the interaction of dry intrusions and convection over the warm pool where the inhibiting effect of dry air dominates (Sherwood 1999; Parsons et al. 2000). We suggest that it is due simply to the fact that the shear is generally weak over the warm pool so it is not possible to build up efficient cold pools, except on particular days when shear is present as observed during TOGA COARE (LeMone et al. 1998). The present analysis investigates the water vapor distribution and origins at a large scale but lacks the very finescale details in the vicinity and within mesoscale convective systems. The use of cloud resolving models should make it possible to investigate the fate of dry air in the environment very close to convection and establish how active this dry air is in these convective systems so as to confirm the above scenario. The analysis indicates that the long-lived, fastmoving systems are related to the dry intrusions. Most of the Sahelian seasonal rainfall is produced by organized mesoscale convective systems, mainly squall lines. FEBRUARY 2005 ROCA ET AL. 405 FIG. 15. Composite maps of convection indices for the dry-intrusion and moist-exhaust phases. Cold-cloud fraction at 253 K (%): (a) dry phase, (b) moist phase, and (c) dry minus moist. Occurrence of all the convective systems (h month⫺1): (d) dry phase, (e) moist phase, and (f) dry minus moist. Occurrence of the convective systems lasting more than 12 h (h month): (g) dry phase, (h) moist phase, and (i) dry minus moist. As a result, the dry intrusions are expected to influence rainfall and its variability over the Sahel through their influence on the squall lines. Over the Sahel, rainfall is characterized by two important modes of variability at synoptic and suprasynoptic time scales (R02). The large-scale circulation associated with the WAM is indeed perturbed at the synoptic scale (3–5 days) by African easterly waves (AEW) associated with the barotropic and baroclinic instabilities of the AEJ (Burpee 1972). These synoptic scales are also seen in the rainfall variability (Reed et al. 1977; Diedhiou et al. 1999), although the link between these AEWs, the squall lines, and rainfall is complex (Duvel 1990; Fink and Reiner 2003). Sultan et al. (2003) recently documented the second mode of variability between 10 and 20 days of rainfall over West Africa. The link between this suprasynoptic mode and squall lines is yet to be investigated. The period of analysis is currently too short to fully extract the spectral characteristics of the dry intrusions. Nevertheless, from the present 2-month analysis, it appears that dry events occur in series that are wellseparated from each other. Figure 14 indicates three dry-event series (early July, late July through the be- ginning of August, and mid-August). This intraseasonal variability of the dry intrusions could explain the 10– 20-day mode of variability of the Sahelian rainfall. Furthermore it could also affect the unclear relationship of the AEW and squall lines and hence modulate the AEW-rainfall efficiency (Diedhiou et al. 2003; Fink and Reiner 2003). A long-term climatology of the Sahelian dry intrusions should be established to provide the needed database to investigate the relationship between their temporal characteristics and rainfall variability. The other finding of this study concerns the extratropical origin and the trajectory of the dry midtropospheric air over the Sahel that raises many questions about the dynamical processes at play in the midlatitudes. The analysis of such processes is clearly beyond the scope of the present paper, and a detailed analysis of the dynamics of this midlatitudes air outbreak will be provided in another paper (D. Raymond 2004, personal communication). Nevertheless, it is useful to stress that Sahelian dry intrusions exhibit different characteristics than intrusions from previously documented regions. Previous analysis of the warm pool dry-intrusion origins 406 JOURNAL OF THE ATMOSPHERIC SCIENCES suggests the instabilities of the boreal winter, Northern Hemisphere subtropical jet, and the corresponding Rossby wave breaking as one dynamical mechanism associated with the dry-air intrusions (Yoneyama and Parsons 1999; Yoneyama 2003). Here dry intrusions originate from the upper troposphere and from the anticyclonic side of the polar jet. They are associated with weak PV anomalies and interact in a specific way with the subtropical jet, which differs from the warm pool intrusion mechanism. Systematic comparison of the PV anomaly and midtropospheric dry layer (e.g., Yoneyama 2003 for the warm pool) over Africa and Europe could further our understanding of the dynamical processes at play in these extratropical dry-air outbreaks. Such effort could complete previous analyses that linked the upper-tropospheric high-PV intrusions with convection over the tropical oceans (e.g., Kiladis and Weickmann 1992; Kiladis 1998; Waugh and Polvani 2000; Waugh and Funatsu 2003) and contribute to form a more complete picture for dry intrusions on a global scale. Acknowledgments. Discussions with D. Ramond, J. P. Cammas, and P. Peyrillé on the midlatitude circulation are appreciated. The authors thank H. Brogniez for her help with the Meteosat free tropospheric humidity data. The authors are grateful to C. Morel, S. Senesi, and L. Labatut whose support and expertise have been helpful in applying the ISIS algorithm to Africa. The authors are deeply indebted to R. T. Pierrehumbert for providing us his model. This study was partially funded by the Programme National Atmosphere et Océan à Méso-échelle of the French Institut des Sciences de l’Univers et de l’Environnement. The anonymous reviewers are acknowledged for their comments that helped us to clarify the paper. REFERENCES Barnes, G. M., and K. Sieckman, 1984: The environment of fastand slow-moving tropical mesoscale convective cloud lines. Mon. Wea. Rev., 112, 1782–1794. Brown, R. G., and C. Zhang, 1997: Variability of midtropospheric moisture and its effect on cloud-top height distribution during TOGA COARE. J. Atmos. Sci., 54, 2760–2774. Burpee, R. W., 1972: The origin and structure of easterly waves in the lower troposphere of North Africa. J. Atmos. Sci., 29, 77–90. Diedhiou, A., S. Janicot, A. Viltard, P. de Felice, and H. Laurent, 1999: Easterly wave regimes and associated convection over West Africa and tropical Atlantic: Results from the NCEP/ NCAR and ECMWF reanalyses. Climate Dyn., 15, 795–822. ——, H. Laurent, T. Lebel, and A. Amani, 2003: Multiscale view of the Sahelian rainfall regimes. CLIVAR Exchanges, Vol. 8, No. 2/3, International CLIVAR Project Office, Southampton, United Kingdom, 24–26. Diongue, A., J. P. Lafore, J. L. Redelsperger, and R. Roca, 2002: Numerical study of a Sahelian synoptic weather system: Initiation and mature stages of convection and its interactions with the large-scale dynamics. Quart. J. Roy. Meteor. Soc., 128, 1899–1928. VOLUME 62 Duvel, J. P., 1988: Analysis of diurnal, interdiurnal and interannual variations during Northern Hemisphere summers using METEOSAT infrared channels. J. Climate, 1, 471–484. ——, 1990: Convection over tropical Africa and the Atlantic Ocean during northern summer. Part II: Modulation by easterly waves. Mon. Wea. Rev., 118, 1855–1868. Fink, A. H., and A. Reiner, 2003: Spatiotemporal variability of the relation between African easterly waves and West African squall lines in 1998 and 1999. J. Geophys. Res., 108, 4332, doi:10.1029/2002JD002816. Gamo, M., 1996: Thickness of the dry convection and large-scale subsidence above deserts. Bound.-Layer Meteor., 79, 265– 278. Goutorbe, J. P., and Coauthors, 1994: HAPEX-Sahel: A largescale study of land–atmosphere interactions in the semi-arid tropics. Ann. Geophys., 12, 53–64. Held, I. M., and B. J. Soden, 2000: Water vapor feedback and global warming. Annu. Rev. Energy Environ., 25, 441–475. Hólm, E., E. Andersson, A. Beljaars, P. Lopez, J.-F. Mahfouf, A. Simmons, and J.-N. Thépaut, 2002: Assimilation and modelling of the hydrological cycle: ECMWF’s status and plans. ECMWF Tech. Memo. 383, 55 pp. [Available from ECMWF, Shinfield Park, Reading RG2 9AX, United Kingdom.] Johnson, R. H., P. E. Ciesielski, and J. A. Cotturone, 2001: Multiscale variability of the atmospheric mixed layer over the western Pacific warm pool. J. Atmos. Sci., 58, 2729–2750. Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471. Kiladis, G. N., 1998: Observations of Rossby waves linked to convection over the eastern tropical Pacific. J. Atmos. Sci., 55, 321–339. ——, and K. N. Weickmann, 1992: Extratropical forcing of tropical Pacific convection during northern winter. Mon. Wea. Rev., 120, 1924–1939. Knippertz, P., A. H. Fink, A. Reiner, and P. Speth, 2003: Three late summer/early autumn cases of tropical–extratropical interactions causing precipitation in northwest Africa. Mon. Wea. Rev., 131, 116–135. Lafore, J.-P., and M. W. Moncrieff, 1989: A numerical investigation of the organization and interaction of the convective and stratiform regions of tropical squall lines. J. Atmos. Sci., 46, 521–544. Laing, A. G., and J. M. Fritsch, 1997: The global population of mesoscale convective complexes. Quart. J. Roy. Meteor. Soc., 123, 389–406. Lebel, T., and L. Lebarbé, 1997: Rainfall monitoring during HAPEX-Sahel: 2. Point and areal estimation at the event and seasonal scales. J. Hydrol., 188–189, 97–122. LeMone, M. A., E. J. Zipser, and S. B. Trier, 1998: The role of environmental shear and thermodynamic conditions in determining the structure and evolution of mesoscale convective systems during TOGA COARE. J. Atmos. Sci., 55, 3493– 3518. Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77, 1275–1277. Mapes, B. E., and P. Zuidema, 1996: Radiative-dynamical consequences of dry tongues in the tropical troposphere. J. Atmos. Sci., 53, 620–638. Mathon, V., and H. Laurent, 2001: Life cycle of the Sahelian mesoscale convective cloud systems. Quart. J. Roy. Meteor. Soc., 127, 377–408. Morel, C., and S. Senesi, 2002a: A climatology of mesoscale convective systems over Europe using satellite infrared imagery. I: Methodology. Quart. J. Roy. Meteor. Soc., 128, 1953–1992. ——, and ——, 2002b: A climatology of mesoscale convective systems over Europe using satellite infrared imagery. II: Characteristics of European mesoscale convective systems. Quart. J. Roy. Meteor. Soc., 128, 1973–1996. Parsons, D. B., K. Yoneyama, and J. L. Redelsperger, 2000: The FEBRUARY 2005 ROCA ET AL. evolution of the tropical western Pacific atmosphere–ocean system following the arrival of a dry intrusion. Quart. J. Roy. Meteor. Soc., 126, 517–548. Picon, L., and M. Desbois, 1990: Relation between METEOSAT water vapor radiance fields and large scale tropical circulation features. J. Climate, 3, 865–876. Pierrehumbert, R. T., 1998: Lateral mixing as a source of subtropical water vapor. Geophys. Res. Lett., 25, 151–154. ——, and R. Roca, 1998: Evidence for control of Atlantic subtropical humidity by large-scale advection. Geophys. Res. Lett., 25, 4537–4540. Redelsperger, J. L., and J. P. Lafore, 1988: A three-dimensional simulation of a tropical squall line: Convective organization and thermodynamic vertical transport. J. Atmos. Sci., 45, 1334–1356. ——, A. Diongue, A. Diedhiou, J. P. Ceron, M. Diop, J. F. Gueremy, and J. P. Lafore, 2002a: Multi-scale description of a Sahelian synoptic weather system representative of the West African monsoon. Quart. J. Roy. Meteor. Soc., 128, 1229– 1258. ——, D. B. Parsons, and F. Guichard, 2002b: Recovery processes and factors limiting cloud-top height following the arrival of a dry intrusion observed during TOGA COARE. J. Atmos. Sci., 59, 2438–2457. Reed, R. J., D. C. Norquist, and E. E. Recker, 1977: The structure and properties of African wave disturbances as observed during Phase III of GATE. Mon. Wea. Rev., 105, 317–333. Roca, R., and V. Ramanathan, 2000: Scale dependence of monsoonal convective systems over the Indian Ocean. J. Climate, 13, 1286–1298. ——, H. Brogniez, L. Picon, and M. Desbois, 2003: Free tropospheric humidity observations from Meteosat water vapor channel data. Preprints, 17th Conf. on Hydrology, Long Beach, CA, Amer. Meteor. Soc., CD-ROM, J3.7. Rotunno, R., J. B. Klemp, and M. L. Weisman, 1988: A theory for strong, long-lived squall lines. J. Atmos. Sci., 45, 463–485. Salathé, E. P., Jr., and D. Chesters, 1995: Variability of moisture in the upper troposphere as inferred from TOVS satellite observations and the ECMWF model analyses in 1989. J. Climate, 8, 120–132. Schmetz, J., and L. van de Berg, 1994: Upper tropospheric humidity observations from Meteosat compared with shortterm forecast fields. Geophys. Res. Lett., 21, 573–576. 407 ——, C. Geijo, W. Menzel, K. Strabala, L. van de Berg, K. Holmlund, and S. Tjemkes, 1995: Satellite observations of upper tropospheric relative humidity, clouds and wind field divergence. Contrib. Atmos. Phys., 68, 345–357. Sherwood, S. C., 1999: Convective precursors and predictability in the tropical western Pacific. Mon. Wea. Rev., 127, 2977–2991. Soden, B. J., and F. P. Bretherton, 1993: Upper tropospheric relative humidity from GOES 6.7 m channel: Method and climatology for July 1987. J. Geophys. Res., 98 (D9), 16 669– 16 688. Sultan, B., S. Janicot, and A. Diedhiou, 2003: The West African monsoon dynamics. Part I: Documentation of intraseasonal variability. J. Climate, 16, 3389–3406. Sutton, R., H. Maclean, R. Swinbank, A. O’Neill, and F. Taylor, 1994: High resolution tracer fields estimated from satellite observations using Lagrangian trajectory calculations. J. Atmos. Sci., 51, 2995–3005. Thorncroft, C. D., and M. Blackburn, 1999: Maintenance of the African easterly jet. Quart. J. Roy. Meteor. Soc., 125, 763–786. Vesperini, M., 2002: ECMWF analyses of humidity: Comparisons to POLDER estimates over land. Remote Sens. Environ., 82, 469–480. Waugh, D. W., and L. M. Polvani, 2000: Climatology of intrusions into the tropical upper troposphere. Geophys. Res. Lett., 27, 3857–3860. ——, and B. M. Funatsu, 2003: Intrusions into the tropical upper troposphere: Three-dimensional structure and accompanying ozone and OLR distributions. J. Atmos. Sci., 60, 637–653. Yoneyama, K., 2003: Moisture variability over the tropical western Pacific Ocean. J. Meteor. Soc. Japan, 81, 317–337. ——, and D. B. Parsons, 1999: A proposed mechanism for the intrusion of dry air into the tropical western Pacific region. J. Atmos. Sci., 56, 1524–1546. Zachariasse, M., H. G. J. Smit, P. F. J. van Velthoven, and H. Kelder, 2001: Cross-troposphere and interhemispheric transports into the tropical free troposphere over the Indian Ocean. J. Geophys. Res., 106 (D22), 28 441–28 452. Zhang, C., B. E. Mapes, and B. J. Soden, 2003: Bimodality in tropical water vapour. Quart. J. Roy. Meteor. Soc., 129, 2847– 2866. Zipser, E. J., 1977: Mesoscale and convective scale downdrafts as distinct components of squall-line structure. Mon. Wea. Rev., 105, 1568–1589.