Observations from the NASA 10 cm polarimetric Doppler weather radar (NPOL) were used to examine structure, development, and oceanic transition of West African Mesoscale Convective Systems (MCSs) during the NASA African Monsoon Multidisciplinary Analysis (NAMMA) to determine possible indicators leading to downstream tropical cyclogenesis. Characteristics examined from the NPOL data include echo-top heights, maximum radar reflectivity, height of maximum radar reflectivity, and convective and stratiform coverage areas. Atmospheric radiosondes launched during NAMMA were used to investigate environmental stability characteristics that the MCSs encountered while over land and ocean, respectively. Strengths of African Easterly Waves (AEWs) were examined along with the MCSs in order to improve the analysis of MCS characteristics. Mean structural and environmental characteristics were calculated for systems that produced TCs and for those that did not in order to determine differences between the two types. Echo-top heights were similar between the two types, but maximum reflectivity and height and coverage of intense convection (>50 dBZ) are all larger than for the TC producing cases. Striking differences in environmental conditions related to future TC formation include stronger African Easterly Jet, increased moisture especially at middle and upper levels, and increased stability as the MCSs coastally transition.

1. Introduction

Tropical cyclone (TC) lifecycles vary in duration, location, and human impact, but in general, TCs follow a similar progression from genesis, intensification, maturity to eventual decay. Mature systems often receive much attention due to their potential human impact, but one aspect of the TC lifecycle that remains difficult to predict is genesis. For many cases over the North Atlantic basin, tropical storms and hurricanes form as a direct result of African Easterly Waves (AEWs) and associated Mesoscale Convective Systems (MCSs) moving off the West African coast during the African monsoon season [1, 2]. On average, AEWs that move off the coast are found to produce TCs on the order of 15–20% of the time across the Atlantic basin [1, 3]. While Hopsch et al. [4] found that these numbers are true in the Main Development Region [58] during the early and later parts of the tropical cyclone season, they also found that the relationship between TC genesis and AEWs during the peak of the season (late August and September) is as high as 40%. In contrast, Avila [9] mentions that there is no correlation between the number of AEWs and the number of TCs that develop over the North Atlantic basin, but Landsea [10] notes that 58% of tropical storms and weak hurricanes (categories 1 and 2) and over 80% of intense hurricanes (categories 3 and higher) originate from AEWs.

These AEWs tend to originate and propagate across the Sahel region of Africa, which is bounded by the Sahara Desert to the north and tropical rainforests to the south. The dynamics behind AEW formation are dependent on several factors. Two of these factors are the African Easterly Jet (AEJ) and the low-level monsoon flow, both of which are instrumental to the formation and strengthening of AEWs in the Sahel [11, 12]. The Intertropical Convergence Zone (ITCZ) has been found to have a large impact on the formation of the AEJ, thus affecting AEW formation and progression [13]. For more information on the flow patterns over the Sahel and their relation to precipitation and AEW formation, see papers by Kanamitsu et al. [14], Chen [15], Lamb [16], Chen and van Loon [17], Fontaine et al. [18], Cook [11], and Diedhiou et al. [12].

As the synoptic flow of the AEJ and low-level monsoon flow interact to produce AEWs, the AEWs also act to produce precipitation and allow MCS formation in the Western Sahel. The maximum amount of precipitation across the Sahel occurs during the months of August and September, which corresponds with the peak in the number of AEWs [19, 20]. A majority of this rainfall is produced by squall-line-type MCSs [2123]. The dynamics and orientation of AEW are determinants of MCS generation on a synoptic level, but Convective Available Potential Energy (CAPE) and low-level wind shear are important for ensuring squall-line-type formation [24, 25]. Diurnal variation can have a profound impact on the strengths of the MCSs as well. The diurnal variation of precipitation over land and ocean environments in the tropics is well documented (e.g., [26, 27]). It was found that a majority of the squall line production over land occurred between the hours of 12 and 18 UTC. This result has been verified by Schumacher and Houze [28] and Futyan and Del Genio [29] through use of the Tropical Rainfall Measuring Mission (TRMM) precipitation radar whereby the more intense systems were able to maintain structure through the diurnal periods.

Being able to observe these aspects of AEWs and MCSs on a regular basis poses a potential problem to forecasting TC genesis because of the very sparse meteorological data networks across most of Sahelian Africa. During the summer of 2006, the African Monsoon Multidisciplinary Analysis (AMMA) international field campaign and NASA-AMMA (NAMMA) collaborated to provide an observation network of various platforms stretching from Nigeria to the Cape Verde Islands [30, 31]. Several studies have used the NAMMA data to examine the AEWs and MCSs moving off the African coast (e.g., [3235]). In addition, one of the three main goals of AMMA-NAMMA was to determine a relationship between AEWs and MCSs that could provide insight into TC genesis. Several numerical modeling studies have looked at NAMMA cases to associate AEW events with TC genesis (e.g., [3638]), but no studies have associated ground-based radar depictions of MCS structure and intensity with TC genesis. This is an area of uncertainty that requires further investigation into how the AEWs and squall-line-type MCSs develop and interact with each other. In many cases, there have been strong similarities to the timing between the two, especially with rainfall amount [12, 23, 39]. However, some cases have shown the opposite to be true [40, 41]. Studying the structure of precipitation from MCSs that are produced by AEWs is an important step to better understand the relationship between these interactions and with downstream TC formation. This study attempts to find a relationship between the structure and environment of these MCSs as they transition off the coast in the hopes of ultimately providing some insights to the genesis portion of the TC life-cycle.

The paper is organized as follows. Section 2 describes the data used in the study. In Section 3, the methodology used in the analysis is presented. Section 4 provides the results of a composite analysis of several AEW-MCS cases, and Section 5 presents the summary and conclusions to the study.

2. Data Description

The NAMMA field program utilized a variety of different observation platforms. For a description of all platforms used during NAMMA, consult Zipser et al. [31]. The key instrument used in this study is the NASA polarimetric Doppler weather radar (NPOL). Structures of observed MCSs were diagnosed with the data gathered by NPOL. The other observations used in the analysis are from the radiosonde measurements collected from the site near NPOL east of Dakar, Senegal, and at the Dakar launch site. The European Centre for Medium-Range Weather Forecasts (ECMWF) provided synoptic scale data in the form of the ERA-Interim Reanalysis as a means of validating some of the conclusions inferred through the analysis of the NAMMA dataset [42, 43].

2.1. NASA Polarimetric Doppler Weather Radar Data

NPOL is an S-Band, dual-polarized Doppler radar that was used to observe precipitating systems in the region near Dakar, Senegal. The radar was located near the village of Kawsara, Senegal (14.657° N, 17.098° W), which is approximately 40 km southeast of Dakar. NPOL was operational from 21 August 2006 through 30 September 2006 and was configured to scan at a maximum range of 270 km when scanning in a low-level Plan Position Indicator (PPI) surveillance mode. The surveillance scans were used to observe precipitating systems located at relatively long distances from the radar location. NPOL also sampled in a radar volume scan mode with 19 elevation tilts out to a maximum range of 150 km. These volume scans were repeated in 15 minute intervals. A map of western Senegal showing the location of NPOL is provided in Figure 1. Precipitation signatures of 19 MCS events were recorded with NPOL, but due to the scope of this work, only eight cases, which are mentioned in Table 1, were chosen for further study. As in Table 1 and hereafter, cases that produced tropical cyclones will be referred to as TCP and those that did not as NTCP.

2.2. Radiosonde and Sounding Data

Radiosondes launched from Dakar and the NAMMA site in Kawsara, Senegal, were used to assess environmental conditions such as instability, wind shear, and tropospheric moisture content. Referring to Figure 1, the location of the Kawsara site is near the position denoted by “NPOL.” The Kawsara radiosondes were launched by representatives of the University of Virginia and Howard University at a variety of times. A total of 72 radiosondes were successfully released between 19 August and 16 September 2006. As many as six radiosondes could be launched in a 24-hour period, and a decision to launch was usually related to weather in the proximity of NPOL in order to capture conditions prior to, during, and after a convective event. Additional radiosondes were launched in conjunction with TRMM satellite overpasses. Dakar radiosondes were used in place of NAMMA radiosondes when the latter were unavailable, but these data have a lower vertical resolution. Only 17 of the 72 atmospheric soundings from NAMMA data set were used for this study, and seven were used from the Dakar location. The temporal aspects of these data are listed in Table 1. The Kawsara radiosonde data were provided by the Global Hydrology Resource Center (GHRC) at the Global Hydrology and Climate Center in Huntsville, Alabama.

2.3. ECMWF ERA-Interim Reanalysis Data

The ECMWF ERA-Interim Reanalysis data were useful for validating the hypothesis that MCSs form due to the dynamics of the AEWs propagating westward across the tropical latitudes of Africa. Also, the ECMWF ERA-Interim Reanalysis data were used to determine how AEWs may impact TC formation. By analyzing meridional winds at 700 hPa, AEW troughs can be identified [1, 12, 44, 45]. The ECMWF ERA-Interim Reanalysis data were examined for periods that were centered on the times of each MCS event, and most of these data are examined prior to, during, and after a convective event similar to that of the radiosondes.

The AEJ was an important synoptic feature to consider for this study due to its effect on AEW formation. Because the location and strength are the most important information to be gained from this data and because the AEJ is mostly a zonally oriented jet, the zonal wind component is a very useful field to help locate the position and strength of this jet. It should also be noted that during the AMMA campaign, radiosondes were assimilated into the ECMWF analyses and forecasts, providing a more accurate structure of the AEJ in the reanalysis [46]. To obtain information on the location and strength of the AEJ, the ECMWF ERA-Interim Reanalysis 600 hPa zonal component wind field was used. Due to the variations in the height of the jet, it may be better to get a layer average of the zonal winds. However, the vertical resolution of the ECMWF ERA-Interim Reanalysis dataset prevents the use of this technique. Based on previous work by Cook [11] and Chen and van Loon [17], 600 hPa is likely the best level at which to attain information about the AEJ.

The derived equivalent potential temperature anomalies ( ) through the vertical column and at specific heights were necessary for determining certain characteristics of the synoptic environment. The field can be used to determine areas with moisture convergence, which provides a better understanding of the moisture in the synoptic environment. The was derived using air temperature data and the relative humidity between 1000–300 hPa. Total precipitable water (TPW) was also derived using the ECMWF ERA-Interim Reanalysis data set in order to validate the column moisture prior to, during, and after a convective event. The dates and times of selection were the same as the other ECMWF ERA-Interim Reanalysis data sets.

3. Methodology

3.1. NASA Polarimetric Doppler Radar Methods

AEW-MCS TCP events were identified using information provided by the National Oceanic and Atmospheric Administration (NOAA) National Hurricane Center (NHC) website. Three named systems (Florence, Gordon, and Helene) were associated with four of these cases. More specific information about these storms can be obtained at the NHC archive located at the website: http://www.nhc.noaa.gov/2006atlan.shtml.

The selection of the NTCP systems was limited by the fact that only four easterly waves developed into TCs over the Atlantic Ocean during the time period from late August until the end of September. In order to develop a fair comparison, cases that were of similar form and strength relative to the TCP cases but did not produce TCs were chosen. The remainder of the 19 cases could have been used in this comparison process, but it was decided that comparing only the cases that did not produce TCs yet had similar structure and organization to the TCP cases would be most beneficial to the goals of the study. The definition for squall-line-type organization is based on the discussion provided by Houze [47] and Glickman [48]. Houze [47] mentions that as squall lines progress, there is a period of intensification lasting for several hours, during which time the squall line begins to expand in length. At the squall line’s maturest stage, there should be a significant leading convective region along with a trailing stratiform region that on average should equate to 100 km across. These constraints were applied to the cases examined during NAMMA to eliminate events that would not compare well to the TCP cases. In addition, a mature, linearly organized squall line would need to propagate through the domain for a minimum period of two hours to be considered as a relevant NTCP case as these systems are more likely to be the most similar to the TCP cases. Of all the NTCP cases, only four meet these criteria. For this study, the best overall TCP case was the 01-02 September event due to its structural development. The best NTCP event occurred on 13-14 September due to similar structural qualities to the selected TCP case. While these were the best squall-line representations for each event type, the eight cases listed in Table 1 will be discussed in more detail.

For these MCS events, NPOL volume scans were mapped to a 2 km horizontal resolution Cartesian grid and contain data up through 18 km with vertical cross-sections perpendicular to the main convective line. Constant Altitude Plan Position Indicator (CAPPI) and cross-section products were created every 15 minutes through the entirety of each event as well. Echo-top heights, maximum radar reflectivity, and height of the maximum reflectivity were retrieved during this process. The maximum reflectivity and height were defined by the largest reflectivity value within the volume and the height at which the reflectivity maximum occurred. If there was more than one location of the same maximum reflectivity, the location at the highest height was used in the analysis. The echo-top height was determined to be the highest altitude of any recorded radar reflectivity >20 dBZ within the profile of maximum reflectivities. Intensities of the MCSs were then primarily determined by periods when the maximum reflectivity and height had their largest values. For these quantities, mean values were created for the most intense period for all TCP cases and for all NTCP cases.

With the limited number of cases being analyzed, traditional statistical methods for discriminating between events using NPOL observations are limited as well. However, a statistical application that is used in this study is the Contoured Frequency by Altitude Diagram (CFAD), which was first outlined by Yuter and Houze [49]. When using only cross-sections or maximum reflectivity profiles, it is sometimes hard to distinguish the importance and effect of convective and stratiform precipitation. CFADs eliminate this uncertainty and provide a more sophisticated method of viewing the evolution of precipitating systems. To compare the eight MCS events, average convective and stratiform CFADs were created for periods one hour prior to, during, and one hour after the most intense period of convection. Convective and stratiform regions are defined according to Steiner et al. [50], which is based on indentifying convective centers through a three-step method. The following discussion is a summary of the convective and stratiform algorithm. At an altitude of three kilometers, any reflectivity value greater than 40 dBZ is considered a convective center. If there are any values within an 11 km radius of each convective center with a reflectivity value greater than the background of that area, these are also listed as convective. Any reflectivity within a background intensity-dependent radius is also considered convective. All other reflectivity values are considered stratiform. Because land/ocean transitions are of interest as well, mean convective and stratiform CFADs were created for the eight cases for land and ocean environments. Within the CFADs, the mean reflectivity (in dBZ) was computed at every kilometer in the vertical up to 10 kilometers. The basic principles of CFAD development are expressed mathematically in Appendix A of Yuter and Houze [49].

3.2. Radiosonde and Sounding Methods

For analysis of the soundings, several parameters were computed and statistically analyzed to enable relevant comparisons between the TCP and NTCP systems for times prior to, during, and after MCS movement through the domain. As with the NPOL methods, composited soundings were created for these three time regimes for both case types. General stability parameters, such as CAPE and CIN, were calculated using RAOB sounding software. These stability parameters were then used to calculate Bulk Richardson number following the equation given in Bluestein [51]. According to Lenouo et al. [52], CIN is not a reliable stability measure in the Sahel due to the low LCL heights. Less emphasis was placed on CIN results due to this observation. Calculations of the dewpoint depression with increasing height were found for the composite soundings used in this study. Statistical comparisons were made between various atmospheric levels, including 1000–850 hPa, 850–600 hPa, and 600–400 hPa. Similarly to dewpoint depression, speed and directional vertical wind shear were calculated through the atmospheric column. In the numerical calculations, this parameter was computed from the surface up to 150 hPa. Similarly to the dewpoint depression, statistical comparisons were made at several levels within the column.

In addition, radiosonde relative humidity data were used to verify dewpoint temperature depression results and to compare relative amounts of moisture in the atmosphere. According to LeMone et al. [53], long-lived and well-developed tropical MCSs are formed in environments that have relatively high relative humidity values from the boundary layer to about 500 hPa. With this in mind, relative humidity was assessed and compared in a similar manner to the dewpoint depression calculations. To assess the height of the boundary layer, a more reliable parameter than relative humidity to use is the virtual potential temperature ( ). According to Stull [54], the virtual potential temperature is a useful parameter for finding the top of the boundary layer because the value changes abruptly at the boundary between two layers, such as the boundary layer and the free atmosphere. Definitions of virtual potential temperature and virtual temperature, which are needed to calculate virtual potential temperature, can be found in Rogers and Yau [55].

3.3. ECMWF ERA-Interim Reanalysis Methods

Several of the parameters computed in this study are derived from the ECMWF ERA-Interim Reanalysis dataset and are described in this section. They include tracking strengths and locations of AEW troughs using 700 hPa meridional wind anomalies, AEJ characteristics using the 600 hPa zonal winds and anomalies, TPW, and equivalent potential temperature ( ) anomalies. Detailed information about the ECMWF ERA-Interim Reanalysis dataset is discussed by Dee et al. [42].

For the ECMWF ERA-Interim Reanalysis dataset, which is provided every six hours, times were chosen to fully represent the environment prior to, during, and after a convective event. Time averages were used to account for this examination scheme so as to fully represent each relative time instead of using a single snapshot. For the times prior to the examined events, 18 hours were used up to six hours prior to the center time of the event. During and after the convective event, 12-hour time spans were used. Anomalies were also calculated over these three time periods for each case type as a way to identify differences. These anomalies are determined by calculating a monthly mean and then computing differences relative to this mean for each designated time. In addition, the events are also presented as Hovmoller diagrams of 600 hPa zonal winds over a span of three days for the two most representative cases for each system type. Detecting the areas where the jet was strongest was completed by looking for maxima of easterly winds (most negative zonal winds). This parameter was mainly used to point out differences in the general flow over West Africa at the times prior to, during, and after MCS progression through the domain. 700 hPa meridional wind anomalies were also calculated to determine the difference in strengths of the AEW troughs. These were calculated similarly to the 600 hPa zonal wind anomalies.

The other parameters evaluated in the study are equivalent potential temperature anomalies and TPW. Equivalent potential temperature anomalies were calculated between 1000–300 hPa and are averaged as mentioned previously. These anomalies were examined to provide additional information regarding moisture and instability of the atmosphere [55]. Higher values imply more moisture presence, and thus, these values are usually highest at the lowest heights where large amounts of moisture are located. For this study, is calculated using equation (38) from Bolton [56]. Anomalies of this value are determined similarly to the methods described in association with the 600 hPa zonal wind anomalies. TPW is calculated by following its definition given in Glickman [48]. It is also averaged over the time periods mentioned previously.

4. Results and Discussion

Due to the number of cases used, descriptions of individual cases will not be presented. However, a general description of each MCS type will be provided to explore reasons as to why the TCP cases later developed into TCs and the NTCP cases did not. Following the event descriptions, a discussion of the synoptic and mesoscale environmental conditions and influences for each system type is given, providing insight into the formation of the two MCS types. The MCS types are then discussed by describing the results of MCS structure and organization as determined from the NPOL data.

4.1. Brief Event Descriptions

For the TCP cases, the MCSs tended to move into the detectable range of NPOL as intensifying squall lines. These systems would quickly progress across the area, often within one to one and half hours, before transitioning off the coast. Once off the coast, the TCP systems would begin to slowly lose their squall line development and weaken in intensity. As the convective regions moved out of the region, the stratiform portions remained for several hours in the NPOL domain.

The NTCP cases produced different results in structure and development as compared to the NTCP cases. These MCSs began as scattered cells of organizing convection moving into the region. Their development was much slower compared to the TCP cases as they progressed slowly towards the coast. Somewhat weaker and less organized squall lines would form as they neared the coast, transitioning to the oceanic environment usually at their most organized and most intense stage of development. Like with the TCP cases, moving over the ocean allowed for the convection to decrease in intensity and produce longer-lasting stratiform precipitation.

4.2. Synoptic Influences: AEJ and AEW

The environmental aspects that led to and sustained development for the event types described previously are needed for further understanding of each MCS type. Examining the effect that the AEJ has on the formation and development of MCSs is one important aspect for this study. As stated by Cook [11] and Rowell and Milford [21], the location and strength of the AEJ has a very distinct impact on AEW formation, location, and strength. The purpose of this section is to locate the AEJ in relation to the area where the MCS formed and to determine if there were differences between the TCP and NTCP systems. For an example that shows the AEJ, the 600 hPa zonal winds are presented as a Hovmoller diagram in Figure 2 for the strongest and most organized cases for each system type. This figure presents the evolution of the AEJ through the lifetime of the event with the winds averaged over the latitude range from 10° to 20° N. In addition, Figures 3 and 4 depict the 600 hPa zonal winds and anomalies, respectively, averaged for all TCP and NTCP events for times prior to, during, and after MCS progression.

Figure 2 provides an insightful view of the strength of the AEJ for two AEW-MCS events observed during NAMMA. For the specific TCP case shown, zonal wind speeds are in excess of −14 m s−1 at a period of 6–12 hours prior to the convective event. During the event, wind speeds weaken slightly by 1-2 m s−1 with continued weakening 12 hours after the event has passed the NPOL location. In relation to the AEW, the AEJ remains strong as it moves off the coast, increasing to values in excess of 16 m s−1 as it begins to move over the eastern Atlantic. The NTCP case shows zonal wind speeds 7-8 m s−1 weaker than the TCP case during the 6–12 hours prior to the convective event. As the AEW passes, the wind speeds remain fairly constant near the NPOL location. The AEJ begins to weaken as it moves off the coast as well, signifying weaker easterly winds than seen with the TCP case.

The AEJ for each system type is shown clearly in Figure 3. Similarly to Figure 2, the 600 hPa zonal winds for the mean TCP case are in excess of −14 m s−1 for times prior to and during the MCS events with slight weakening after the event. Much weaker zonal winds are found for the NTCP cases with 4–6 m s−1 weaker winds at times prior to and during each event and nearly 8 m s−1 less than the TCP cases after the MCS passage. Figure 4 shows the 600 hPa zonal wind anomalies for the same times as Figure 3. Clearly, the TCP events experience higher negative anomalies (i.e., stronger easterlies) with easterly winds greater than the monthly mean by 4–6 m s−1. The NTCP cases show nearly no departure from the monthly mean for each time period of MCS progression. Figures 2, 3, and 4 depict that the AEJ is quite a bit stronger for the mean TCP case compared to the mean NTCP case.

Now that the strength of the AEJ is known for the two system types, it is important to determine the locations of the AEW troughs associated with these events. Stronger AEJ velocities have an increased influence on the strength of the AEW, and based on previous discussion, the TCP systems are likely to be associated with better developed AEWs. To investigate this idea, mean 700 hPa meridional wind anomalies are shown in Figure 5. These winds are plotted similarly to the zonal winds in Figure 4. From the 700 hPa meridional wind anomalies for the mean TCP case, it is clear to see there is a wave trough that passes through the domain, but meridional wind speed anomalies are weak for the period 6–12 hours prior to the convective event. As the trough passes over the NPOL site, the meridional wind anomalies increase, but by 12 hours after its passage, the northerly wind component begins to affect the area, allowing for a weak ridge to build over the area. However, the mean TCP panels clearly display a strong AEW passage through the area. In contrast, the mean NTCP case experiences weak meridional wind anomalies in association with the respective AEWs. Based on Figure 5, it is implied that AEWs associated with the NTCP events are much weaker than their TCP counterparts. Because TCP events have higher trough strengths, it reveals that this characteristic and the stronger AEJ could have a larger impact on the wave later forming a TC downstream.

4.3. Environmental Influences from Soundings

In this section, several components derived from the observed soundings are presented in tabular form to identify possible differences between the two AEW-MCS types. Environmental indices such as CAPE, 0–5 km CAPE, CIN, LFC height and Bulk Richardson number (R) are examined to identify any differences in stability and vertical profiles of the environment. In some cases, the dry, dusty Saharan air layer (SAL) can extend southwards and impact the environment in which the AEWs and MCSs propagate. However, discussion of SAL impacts is beyond the scope of this study. A summary of the information gathered from the radiosonde analysis is shown in Table 2.

For the mean TCP conditions, the CAPE decreased from a substantial 2290 J kg−1 prior to the MCS event to 555 J kg−1 after the MCS moved out of the region. The mean NTCP event showed less decrease in CAPE where prior to MCS propagation through the area, the CAPE was 1555 J kg−1 decreasing to 660 J kg−1 after the event passage. These trends are not reflected in the CIN values as the mean TCP case has lower CIN (−22 J kg−1) prior to the event and becomes increased by −102 J kg−1 by the time the MCS passes through the region. The mean NTCP case has higher CIN (−73 J kg−1) before the event and then becomes more positive by nearly 50 J kg−1 before increasing by −150 J kg−1 after the MCS passage. As mentioned previously, these CIN results have less weight on MCS formation due to the fact that LCL heights are very low in the Sahel region [52]. However, this confirms the idea that higher CAPE with low CIN will more than likely produce more intense storm conditions due to the fact that the TCP cases show more intense convection and more linear organization.

Examining the mean Richardson number for each period for both case types, the mean TCP case has more favorable values that allow for the production of sustained strong storms. Bluestein [51] mentions that Richardson numbers between 30 and 40 are capable of producing strong systems but wind shear is often too weak to produce intense storms. It is also mentioned that values between 15 and 30 are better indicators of long-lived strong systems, while values above 40 are less likely to produce intense storms. The Richardson numbers found for the mean TCP case show that prior to the MCS event, there is a good possibility of strong systems moving through the area, and during the event there is an even greater likelihood of strong, long-lived systems in the area. The mean NTCP Richardson numbers reveal that while CAPE is increasing, the likelihood of producing very strong, long-lived systems is low. Even though the NTCP MCSs gained CAPE before and during the propagation of the MCS, the maximum amount of CAPE determined from the data was too weak to produce conditions similar to the mean TCP case. These results show that much stronger, long-lived MCSs are predicted by the soundings for the mean TCP case, hinting that these types of systems may be more suitable for future TC development.

4.4. Moisture Analysis

As is shown subsequently, the sounding plots produced from the radiosonde data at Kawsara and Dakar showcased some key differences between the two case types. One associated analysis examined was the dewpoint depressions over several pressure ranges. Mid-level dewpoint depressions were calculated between 600 and 400 hPa. Low-level and surface level ranges were calculated between 850 and 600 hPa and surface to 850 hPa, respectively. Averages of the dewpoint depression along with statistical significance based on a Student's t-test at 95% were calculated for each range for soundings prior to, during, and after MCS propagation through the domain. Table 3 shows the results of these calculations.

For the mean TCP case, the average dewpoint depression prior to MCS progression through the domain within the 600 hPa to 400 hPa range is −9.2°C, whereas the mean NTCP case is −11.3°C over this same pressure range. This difference shows that the atmospheric layer from 600 hPa to 400 hPa was drier for the mean NTCP case, and this difference is statistically significant at 95% confidence level. At low levels, the mean TCP case is slightly drier than the mean NTCP case, and this difference is also statistically significant. The surface-layer shows a difference of just over 3°C between the two cases, with the mean NTCP case being the drier of the two. This difference at the lowest layer is also statistically significant. The values of dewpoint depression are very similar between the two case types during the MCS events; however, there is significant difference with this value after the MCS passage. As shown in Table 3, the mid-levels and low-levels are considerably drier for the mean NTCP than for the mean TCP case, both of which are statistically significant.

As a way to further confirm the moisture results, relative humidity (RH) and profiles are analyzed. The key to this section is the correlation between strength of system and ultimately TC formation to the average RH and and how these relate to the tropical boundary layer. According to Fitzjarrald and Garstang [57], tropical boundary layer depth is usually found to be on the order of 600 to 900 m. Although these results were obtained over oceanic surfaces, they do provide a baseline for expected values over land. Values calculated herein are provided in Table 4.

The boundary layer heights used here are a mechanism by which average RH over physically relevant layers can be calculated. LeMone et al. [53] found that for strong, organized MCSs, the average RH from the top of the boundary layer to 500 hPa was 80%. Figure 6 displays the RH with increasing height for the three times for both the mean TCP and NTCP case. In Figure 6(a), it is clear to see that the TCP case tends to stay slightly more humid throughout the column, especially within the boundary layer and the middle levels. Using the ideas from LeMone et al. [53], the mean TCP case average RH from the top of the boundary layer to 500 hPa prior to the MCS event was 66%. This value increases to 73% and 80% during and after the MCS event, respectively. This increased RH is expected due to the presence of convection and widespread cloudiness across the area. Under the same vertical constraints, the mean NTCP case has an average RH of 62% prior to MCS propagation through the region. During the MCS passage through the domain, the average RH increases to 69% before decreasing to 61% after the squall line has moved off the coast and out of the domain. In addition, the differences between the mean TCP and NTCP soundings prior to and after MCS passage are both statistically significant at the 95% confidence level. This result is important in showing that during the MCS progression through the study region, the mean NTCP case encounters drier air than the mean TCP case above the boundary layer. This idea is also seen with the mean RH within the boundary layer. The mean TCP case has an RH of 88% prior to the MCS event while the mean NTCP case has 77% RH for the same time. At middle levels (600–400 hPa), RH values are also consistently larger for the mean TCP case with humidities increasing from 55% to 57% to 79% from times prior to, during, and after the MCS event, respectively. The corresponding values for the mean NTCP case are 47%, 54%, and 53%. These large differences in RH at the middle levels of the atmosphere are statistically significant at 95% as well.

Moisture content in the atmosphere is also examined using equivalent potential temperature profiles and anomalies. Lucas and Zipser [58] during their study of West Pacific tropical precipitation found that strong, well-developed systems have higher values of from about 300 hPa down through the top of the boundary layer. The difference they found through this layer was on the order of 2–5 K. Figure 7 shows profiles for the three times for both cases. Similarly to Figure 6, Figure 7 shows substantially larger values within the boundary layer prior to the MCS event. This mean difference within the boundary layer is statistically significant as well. Above the boundary layer, values are closer between the two mean cases. Using the Lucas and Zipser [58] ideas, average values were calculated from the top of the boundary layer up to 300 hPa and these values are provided in Table 5 along with values from the surface to the top of the boundary layer. values above the boundary layer indicate mean differences of 2.4 K, 3.4 K, and 6.9 K for times prior to, during, and after MCS passage, respectively. While these differences are not large, Student's t-tests for each time reveal that the differences between the two columns above the boundary are statistically significant at the 95% confidence level. These values also provide proof that there is more moisture present within the mean TCP case.

Mean and TPW could also provide additional information regarding moisture fields. It has been found that strong MCS production increases in likelihood when there is an influx of higher values [59]. If there are strong positive anomalies of and TPW for a given time, the increased moisture could signify a better chance of observing more intense MCSs. Since TCP systems eventually produce TCs and conceivably are more intense, one might expect to see higher positive anomalies associated with TCP systems. While mean anomalies (not shown) do not reflect this idea, TPW anomalies shown in Figure 8 imply that larger amounts of moisture are present in association with the mean TCP case. In Figure 8(a), the average anomalies prior to MCS progression through the domain are in the order of 4.0 mm larger than the NTCP case. At times during and after MCS progression through the domain, the TCP case shows increased columnar moisture across the domain, while the NTCP case experiences less columnar moisture as compared with the monthly mean. Throughout the life-cycle, the anomalies remain between 4–8 mm for the mean TCP case, while the mean NTCP case experiences mostly negative anomalies. This analysis confirms that the mean TCP case has higher columnar moisture content than the mean NTCP case.

4.5. Vertical Wind Shear Analysis

Vertical wind shear is said to be important for the development of squall lines [6062]. Since all of the MCS events mentioned in this work produced linear organization during their lifetimes, it is likely that most if not all of the soundings will have significant vertical wind shear values. Figure 9 shows the mean wind speed profiles for both case types for times prior to, during, and after the MCS event. Associated hodographs of these profiles are also provided to give an idea of the directional wind shear. From Figures 9(a) and 9(b), the AEJ is clearly depicted near 600 hPa in the mean TCP case. These strong middle-level easterly winds, however, are not seen with the mean NTCP case. The wind speed maximum below 500 hPa for the NTCP case is near 10 m s−1, while the maximum for the mean TCP case below 500 hPa is near 18 m s−1. The mean TCP hodograph prior to the MCS event (Figure 9(d)) reveals west-northwesterly winds near the surface, followed by the winds backing with height and becoming easterly at 800 hPa, a change of over 180°. The mean NTCP hodograph (Figure 9(g)) shows southwesterly winds at the surface with the winds backing with height and becoming easterly at 800 hPa. However, the change in wind direction at this lowest level is on the order of 100°. Figures 9(b), 9(e), and 9(h) show the wind speed profiles and hodographs for times during the MCS event. The wind speeds for the mean TCP case remain stronger than that of the mean NTCP case through most of the atmospheric column, with the presence of a weakened AEJ still noticeable within the mean NTCP case. The hodographs reveal that similarly to the sounding prior to the MCS event, the mean TCP and NTCP cases both experience backing up to 800 hPa, but the amount of backing is much higher for the mean TCP case. While the importance of backing and veering is more noticeable at middle latitudes in relation to cold and warm air advection, storm dynamics reveal that a backing wind profile is more favorable for long-lived intense convection in the tropics and subtropics [51]. In this case, the mean TCP soundings show more favorable wind shear conditions to produce strong storms.

Calculations of the speed and directional wind shear reveal some interesting facts about the two MCS types. Tables 6 and 7 show the 1000–800 hPa, 800–600 hPa, 600–400 hPa, and 400–150 hPa mean wind shear values and their statistical significance. As was previously discussed, the directional change in the winds dynamically can help or hinder long-lived severe storms. Table 7 confirms that strong counterclockwise rotation in the winds at the low levels is important as well, proven by the intense squall lines that were produced in the TCP events. The strong wind shear at the middle levels in the TCP cases also helps sustain these long-lived systems. While the mean NTCP case had rotation at the lowest levels, the weaker turning with height acted to hinder stronger convective development. The times prior to and during the MCS show the most differences in wind shear and wind profile (Figure 9) with the most important differences being the directional shear at the low levels and speed shear at the middle levels. These wind shear calculations provide evidence that the wind profile in the mean TCP case is more favorable for strong MCS development and show that these are the type of conditions that could complement future TC genesis.

4.6. MCS Cases Observed by NPOL

The previous sections describe likely conditions that aided in the production of the MCSs. The aim of this section is to describe the MCS types in terms of organization of precipitation and vertical development of precipitation based on the NPOL data to identify any structural differences between the two types of systems. As an example, Figure 10 shows NPOL CAPPI and vertical cross-sections of radar reflectivity for the times when the two strongest of all MCS cases were most organized in development, meaning the time period throughout the MCS life-cycle where the squall line was most linear in shape. For the TCP case (Figure 10(a)), the period of the most organized convection was also the period of the most intense convection, meaning that the maximum reflectivity was found at the highest altitude within the MCS. This characteristic is not true for the NTCP case (Figure 10(b)). With this idea in mind, the TCP system tends to be more organized, more intense, and move more quickly through its progression across the region. While not shown, all TCP cases produced their most intense convection during their most organized development, and the NTCP cases maintained weak organization at different times than their most intense convection.

The echo coverage area for each case is described in terms of convective versus stratiform portions of the system. As was mentioned in Section 3, the Steiner et al. [50] method for determining convective areas was used to separate the two precipitation regimes, and Figure 11 provides an example of these two regimes at the same time as the CAPPI images in Figure 10. As can be seen in Figure 11(a), the locations of convective centers are much more concentrated and encompass a larger area than that of the NTCP case in Figure 11(b), both of which signify the times of greatest linear organization. As with radar reflectivity, the convective and stratiform separation for the TCP case tended to be higher than those seen with the NTCP case. Figure 11 provides an example of a better organized system that produces more intense convection, and with more intense convection and increased organization, there is a larger possibility of maintaining these characteristics once the system moves over water.

While specific examples are seen in Figures 10 and 11, Figure 12 provides a representation of the mean area coverage for each MCS type in terms of convective and stratiform precipitation for one hour prior to and one hour after the time of most intense convection. Figure 12 is divided into six panels showing the total area with height for the convective region, stratiform region, and convective region within four reflectivity thresholds. Error bars are included and are representative of the 95% confidence level of each area as determined by a Student's t-test. The main convective regions for each case type (Figure 12(a)) show very strong similarities in area coverage up to 11 km. Above this height, the mean NTCP case has slightly larger areal coverage of convection with no statistical significance at each height through 18 km. The stratiform region (Figure 12(b)) shows some drastic differences between the two case types with the mean NTCP case having a much larger area than the mean TCP case through 10 kilometers followed by similar coverage between the two cases above this height. Looking specifically at the convective regions, both cases have similar area coverage of reflectivities at lower heights with the mean TCP case tending to have larger areas above eight kilometers. While these areas are larger, they are not statistically significant. At reflectivities >50 dBZ (Figure 12(f)), the mean TCP case has greater coverage at heights between two and six kilometers, but these areas are not statistically significant either. While these results do not show statistical significance, they do show that throughout the MCS life-cycle, the TCP case type has more intense convection reaching higher altitudes within the system.

In addition to the area coverage, three characteristics were briefly analyzed to compare vertical structure of the MCSs, and these include echo-top height, maximum radar reflectivity, and the height of the maximum reflectivity. Based on the NPOL volumes that were produced, mean profiles of the maximum reflectivity at each height level up to 18 km were created for each time period and were averaged for time of most intense convection, for over land, and for oceanic conditions. Echo-top heights for the mean TCP case for the most intense period averaged to 15.0 km, while the average for the mean NTCP case was 15.1 km. Mean maximum reflectivity for the mean TCP case was 53 dBZ at an average height of 1.6 km while the same parameters for the mean NTCP case were 51 dBZ at a mean height of 1.5 km. While over land, the echo-top heights for the mean TCP and NTCP cases are 15.4 km and 15.9 km, respectively, and over the ocean, these are 15.2 km and 13.8 km, respectively. Maximum reflectivity and height over land for the mean TCP case is 52.3 km at 1.7 km, and while over ocean, these values decrease to 49.8 dBZ and 1.5 km. The mean NTCP case has a mean maximum of 51.3 dBZ at 1.7 km over land and a value of 48.9 dBZ at 1.6 km over the ocean. These results reveal that the mean TCP case has intense precipitation stretching higher into the atmosphere. Also, the mean NTCP case was slightly more vertically developed than the mean TCP case because of the higher echo-top heights. It can be said that the two cases have similar vertical development, but the mean TCP case developed deeper intense convection compared to the NTCP MCSs. Also, as the systems transition off the coast, it can be seen that the mean TCP case maintains better vertical development as well as deep convection. It is interesting to see that while many of the cases used occurred in the early morning hours, they still produce quite a bit of vertical development and maintain their convective qualities for an extended period of time [27, 63]. Despite the similarities between the mean cases here, it does seem that the TCP cases display better oceanic transitional structure, which could aid in future TC genesis downstream.

4.7. CFAD Analysis

To look at the structure and strength of these systems statistically, we utilize the use of CFADs. Mean CFADs were created for both case types for the most intense period and for land versus ocean environments. The first attribute examined is the difference between the convective and stratiform regions of each event. Figure 13 provides the mean CFADs for each precipitation regime for both MCS types for the most intense period of convection. The contours are plotted every five percent beginning at 2.5% up to 42.5% with reflectivity bins every 5 dBZ from 0 to 60 dBZ. Also, if the number of instances within a reflectivity bin is less than two percent of the total number at that height, the values are not included in the CFAD. Comparing the convective portions (Figures 13(a) and 13(b)) from each case type reveals that both system types tend to have their highest relevant percentages below five kilometers between 40 and 45 dBZ. The mean TCP case has a broader range of reflectivity values throughout the height of the column, indicating that the percentages will not be as high as seen in the mean NTCP case. The mean TCP case also has a greater mean reflectivity of between 3–5 dBZ at each level, especially between 1–5 km, indicating the presence of more intense convection at higher altitudes within the systems. These convective CFADs reveal that the mean TCP case has slightly higher frequency of intense convection in the lowest few kilometers of the atmosphere compared to the mean NTCP case, and the mean TCP case has better coverage of these stronger reflectivities at higher heights. This is indicative of a strengthening system while the trends seen in the mean NTCP case reveal more influence from downdrafts aloft, which could indicate weakening of the convection.

The stratiform portion of the system (Figures 13(c) and 13(d)) shows similar circumstances between the two cases. The largest differences are below five kilometers where the mean NTCP case shows that percentages are greater within the 25–30 dBZ range. However, the mean TCP case maintains higher percentages between 30–35 dBZ and indicates an increased presence of the bright band signature between four and five kilometers. The range of reflectivity coverage is larger for the mean NTCP case below five kilometers, which is due to the amount of coverage of the stratiform precipitation during the MCS event (Figure 12(b)). These stratiform CFADs also support the idea that the TCP systems are stronger due to the higher percentages of stronger reflectivities, especially near the bright band altitude.

During the oceanic transition, it was discussed previously that both cases experience weakening and disorganization of the squall line. The CFADs for convective and stratiform land and oceanic conditions for both cases are presented in Figures 14 and 15 in a similar fashion to Figure 13. Examining the mean TCP case (Figures 14(a), 14(c), and 15(a), 15(c)), the convective and stratiform CFADs are similar between surface types. However, within the convective CFAD, it is possible to see the increased frequency at the upper levels and weaker reflectivities. Also, the height of the CFAD begins to diminish over the oceanic surface, showing that the system is losing its vertical development. These results are both indicative of weakening. The stratiform portions are also very similar to each other, but the land condition shows more influence from the bright band due to the larger frequencies at heights between four and six kilometers. The mean NTCP case (Figures 14(b), 14(d), and 15(b), 15(d)) comparison has some similar results to the mean TCP case; however, the frequencies within the convective CFADs are much larger than was seen with the mean NTCP case. These higher frequencies are likely due to the lower total number of convective centers over each surface type, especially with increasing height. The convective CFAD over the ocean is noticeably thinner than that over land, encompassing a range of about 5 dB less than that of the CFAD over land. While the land CFAD here shows some weakening of the MCS, the narrowing of the ocean CFAD as well as the lower slope of the profile indicates a quickly weakening system. The key difference between the two systems revealed through the CFADs is that the TCP cases undergo less weakening as they transition off the coast. The NTCP cases spend much of their time over land developing into linear MCSs, and it is because of this lack of organization that they quickly decreases in strength.

5. Summary and Conclusions

African Easterly Waves and associated West African squall lines are vital to the prediction and determination of tropical cyclone formation over the Atlantic Ocean. Studying these seasonal occurrences using observations from NAMMA has provided insight into the structure and development of several squall lines that moved off the West African coast. Because four of these developed into TCs over the Atlantic, the characteristics of these cases were examined and compared with four similar cases that did not produce TCs to characterize differences in structure and organization. Through use of the NPOL radar products, coverage area was determined within various convective and stratiform conditions, and the mean TCP case was found to have more intense precipitation, especially at heights between two and eight kilometers. Stratiform precipitation had greater coverage on average for the mean NTCP case mainly due to the lack of convective locations within the MCS. Echo-top heights also revealed that the mean TCP case was similar in vertical development as compared to the mean NTCP case, but larger maximum reflectivity values reached higher altitudes during the mean TCP case. The transition from land to ocean also revealed that the TCP cases, because they tended to be stronger and more organized, were able to maintain their convection and in some cases intense convection for a longer period after moving over the water. NTCP cases, due to their lack of organization and strength, failed to maintain any kind of intensity once over the water.

When discussing the CFADs for each case, the mean TCP system convectively showed less weakening throughout the lifetime of the system, while the mean NTCP case showed some areas of strong convection that likely represented localized areas of intense development. The NTCP case displayed larger relative frequencies within the higher reflectivity bins, but these frequencies did not translate well to the upper levels where signs of weakening were more prominent. Stratiform precipitation also conveyed a stronger system as a whole due to the presence of the well-represented bright band observed in the CFAD. This bright band signature was not seen as clearly in the mean NTCP CFAD despite the larger area coverage of stratiform precipitation associated with this case. The oceanic versus land CFADs had similar results where weakening within the convection was found more prominently with the mean NTCP case, especially as the MCS transitioned over the ocean. Analyzing higher frequencies over larger reflectivity ranges while the system was over land was expected and the mean TCP case showed that it was able to maintain this idea for a longer period of time after oceanic transition.

Aside from the NPOL data, the NAMMA radiosonde data that were collected at Kawsara and Dakar showed some striking differences between the two cases. Stability indices such as CAPE and Bulk Richardson number revealed that a better environment was in place for a squall line to develop and move through the region. Mean profiles of relative humidity, equivalent potential temperature, and dewpoint depressions show a relatively dry mid-level associated with the mean NTCP case, whereas the mean TCP case had larger amounts of moisture throughout the column. The drier atmospheric conditions increase the entrainment effects on the convective updrafts and are a possible reason why the mean NTCP system did not organize or have distinct development. ECMWF ERA-Interim Reanalysis analyses of TPW and equivalent potential temperature anomalies confirm these conclusions that the mean NTCP case was significantly drier than the mean TCP case.

Wind profiles and vertical wind shear also played vital roles in the development of the two system types. Low-level wind shear was found within both cases, but the mean TCP case experienced high amounts of backing through the lowest levels. This condition is more favorable for squall line development in the tropics and could be a contributor to the strength and organization of the MCS. The mean NTCP case experienced weak backing conditions, which is less favorable for development. The wind profiles for each case also revealed that the presence of the AEJ was an important influence on the development of the TCP squall lines, whereas the lack of a strong AEJ in the NTCP cases contributes to their lack of development.

Overall, there were some obvious differences with the structure of the mean TCP case and mean NTCP case in relation to the systems and the environmental influences. Based on the results discussed, certain conditions may be more favorable for development and could be early signs of TC development over the Atlantic. Strong environmental influences such as the presence of an intense AEJ and a strong backing wind within the monsoon flow at low levels could be critical components to strengthening and organizing precipitation that forms in association with an AEW. Weak wind shear at middle and upper levels is also a positive influence on the development of a strong MCS. More intense and more organized convection determined from radar reflectivity could signify a greater possibility of development once the system coastally transitions from land to ocean. While it has been shown in this study that there were certain characteristics that are distinctly different between the two MCS case types, it is apparent that downstream development is reliant on future environmental and mesoscale interactions. Therefore, the conclusions from this work do not encompass all scenarios but provide a base for furthering the study of West African MCS as they move off the coast and propagate downstream.


The authors would like to thank the many representatives from the institutions and organizations that were involved in the data collection of the data sets used in this study, including Howard University, the University of Virginia, the University of North Dakota, and NASA. This paper was completed through a grant from the National Aeronautics and Space Administration (Grant no. NNX06AC73G).