Abstract

The geographic and diurnal variability of moist convection over tropical Africa and the east Atlantic is examined using the Tropical Rainfall Measuring Mission (TRMM) satellite and related to the variability of the convective environment. The stratiform rain fraction is highest within oceanic and continental regions just north of the equator. Both regions have high column relative humidity (CRH). In both monsoon and semiarid continental regions, stratiform rain fractions are significantly higher on days when the CRH is high, which suggests a relationship between these quantities. Large convective systems with high echo tops dominate the rainfall over the Sahel. The importance of CAPE and shear to the development of these types of systems is suggested by the fact these systems are especially common on days when the CAPE and shear are unusually high.

Both deep convective and stratiform conditional rain rates increase with the size and echo-top height of convective systems. According to the TRMM Precipitation Radar (PR) near-surface rain rate, the highest deep convective and stratiform conditional rain rates occur off the coast of West Africa. However, comparisons between the PR near-surface rain rate and rain rates computed from Z–R relationships from the literature suggest that deep convective conditional rain rates over the Sahel are underestimated by the TRMM precipitation algorithm. Over the Sahel, small (large) convective systems produce most of the rainfall in the afternoon (early morning). This is associated with enhanced convective rainfall in the afternoon and stratiform in the early morning. The transition from small to large convective systems as convection propagates away from topographic features is also observed.

1. Introduction

The rain that falls over tropical Africa and the east Atlantic during the West African monsoon (WAM) comes from convective systems with a wide range of properties. Over the Sahel, convection frequently takes the form of intense mesoscale convective systems (MCSs) characterized by bowing convective lines known as squall lines (e.g., Hamilton et al. 1945; Fortune 1980; Hodges and Thorncroft 1997; Mathon and Laurent 2001; Rickenbach et al. 2009). In contrast, rain over the east Atlantic comes from both MCSs and isolated shallow convection (e.g., Short and Nakamura 2000; Schumacher and Houze 2003a, 2006).

In this study, regional differences in the properties of convection and their diurnal cycle are examined and related to the large-scale convective environment. Particular attention is paid to the cloud type population and metrics indicating the potential for heavy rainfall (intense rain rates, large sizes, and high echo tops). Understanding the influence of the large-scale environment on the cloud type population is important because each cloud type is characterized by a different latent heating (LH) profile (e.g., Houze 1989, 1997, 2004; Schumacher et al. 2007). Both the large-scale flow of the WAM (Hagos and Zhang 2010) and African easterly waves (AEWs) (Janiga and Thorncroft 2013), the dominant weather systems in the region, are influenced by regional differences in LH profiles. The environmental conditions that give rise to convective systems with heavy rainfall are important because of their immediate effects and because of their importance to the seasonal rainfall in the Sahel (Mathon et al. 2002).

Studies using the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) have shown that stratiform rainfall tends to be more frequent over the east Atlantic than the African continent (Schumacher and Houze 2006). This land–ocean contrast is observed over most of the tropics (Schumacher and Houze 2003b) and has been linked to the higher “convective sustainability” of the oceanic boundary layer (Yuter and Houze 1998; Houze 2004; Schumacher and Houze 2006). In these studies, it is proposed that the warm moist boundary layer of the ocean—which in contrast to the land does not become stable at night—is able to maintain the deep convection within MCSs for longer periods, producing extensive stratiform cloud.

Other studies have suggested that the large-scale environment might modulate stratiform clouds directly. Examinations of TRMM PR reflectivity profiles by Geerts and Dejene (2005) and Fuentes et al. (2008) indicate that stratiform rain has more intense low-level evaporation over the African continent than over the east Atlantic. They proposed that this is related to the drier environment over the continent. In addition, Del Genio et al. (2012) found that during active periods of the Australian monsoon, when there is increased free-tropospheric humidity, sublimation of ice aloft within stratiform clouds is reduced and deposition increases. This led to additional latent heat release and mesoscale ascent above the freezing level within the stratiform clouds.

Examining 85-GHz ice scattering signatures, Mohr and Thorncroft (2006) found that MCSs with intense updrafts are strongly constrained to the Sahel. This region has high conditional instability due to the intense surface heating and high low-level shear due to the presence of the African easterly jet (AEJ). Schumacher and Houze (2006) found that conditional convective rain rates are greatest over the Sahel and related this to the high conditional instability and frequency of convective systems with deep intense updrafts in the region.

Nicholls and Mohr (2010) examined the environment of convective systems with the most intense ice scattering signatures over West Africa and found that they had slightly higher values of low-level equivalent potential temperature (θe) and shear. In addition, Guy and Rutledge (2012) found that high echo-top convection is most frequent in the northerly phase of African easterly waves where there is a tongue of high low-level θe air (Berry and Thorncroft 2005, 2012). Lastly, both idealized modeling studies (Rotunno et al. 1988; Weisman and Rotunno 2004) and studies using observations from the region (Barnes and Sieckman 1984) have found that shear plays a crucial role in organizing convection into MCSs.

Over tropical Africa, both convection and the convective environment—particularly shear and instability—are strongly modulated by the diurnal cycle (e.g., Parker et al. 2005a). Using geostationary infrared (IR) observations, Laing et al. (2008) showed that convection is typically triggered over topographic features and then propagates downstream as it grows into large MCSs. Because the diurnal cycle of rainfall in this region is so strongly influenced by the life cycle of organized convective systems, it is poorly simulated by models with parameterized convection (Pearson et al. 2013).

Convective systems detected by TRMM can be defined in a variety of ways (e.g., Liu et al. 2008). The method used in this study is to identify contiguous areas of rainfall detected by the TRMM PR, which are referred to as precipitation features (PFs). One of the simplest but most informative PF properties is their size, which can yield insights into the life cycle and organization of the convective systems (e.g., Romatschke and Houze 2011a,b). Intense updrafts within these PFs can be inferred through the detection of high-altitude echoes by the PR and low polarization corrected temperatures (PCTs), indicative of large ice hydrometeors, by the TRMM Microwave Imager (TMI) (e.g., Nesbitt et al. 2000; Zipser et al. 2006; Liu et al. 2007; Liu 2011).

The goal of this study is to use the TRMM platform to produce an integrated assessment of the properties of convection over this region and examine how these properties are influenced by the large-scale convective environment. Several studies have used TRMM to characterize the regional and diurnal variability of convection over tropical Africa and the east Atlantic. However, these studies have often focused on different metrics, such as the vertical structure of convection (Hirose and Nakamura 2004; Geerts and Dejene 2005; Fuentes et al. 2008), convective intensity (Mohr and Thorncroft 2006; Nicholls and Mohr 2010), and rain contributions from different cloud types (Schumacher and Houze 2006). In addition, most studies approach this problem by analyzing seasonal means. Only a handful of studies (e.g., Nicholls and Mohr 2010) have directly examined how the properties of convection vary during different large-scale environments in this part of the world.

Section 2 discusses the data sources and methodology. Section 3 examines regional differences in the properties of convection using simple metrics such as the conditional rain rate and the frequency of different rain types detected by the TRMM PR. Section 4 builds on this by analyzing regional differences in the properties of convective systems using TRMM PFs. Section 5 examines how the properties of convection vary as a function of moisture, conditional instability, and shear in several regions using data from TRMM and reanalyses. Section 6 examines how the preceding three sections relate to the diurnal cycle. Last, section 7 discusses linkages between the above sections and offers some conclusions.

2. Data and methodology

a. TRMM data sources

In this study, the properties of convection are primarily assessed using the TRMM PR (Kummerow et al. 1998). The PR is a KU band (2.17 cm) radar that can reliably detect echoes of 18 dBZ—equivalent to a rain rate of 0.5 mm h−1. The PR underestimates the rain from shallow cumulus clouds, which are frequently weaker than this (Berg et al. 2006). In August 2001 the satellite was boosted to a higher orbit to conserve fuel; this slightly reduced the sensitivity of the PR (Shimizu et al. 2009). The PR has a swath width of 245 km (215 km preboost) and a resolution of 5 km (4.3 km preboost). The PR samples the same location approximately once every 3 days. Reflectivity profiles have a vertical resolution of 250 m and are available from just above the surface to 19.75 km. In addition, the TRMM satellite has a precessing orbit that allows it to evenly sample the diurnal cycle.

All TRMM data used in this paper are from the version 7 release (TRMM Precipitation Radar Team 2011). The near-surface rain rate and three-dimensional reflectivity profiles are taken from product 2A25. The rain type classifications, taken from product 2A23, were used to define four categories of precipitation: convective, shallow convective, deep convective, and stratiform (Table 1). In short, horizontal reflectivity gradients and brightband detection are used to classify PR footprints as convective or stratiform (Awaka et al. 1997). TMI polarization corrected temperature at 85.5 and 37 GHz are calculated following Cecil et al. (2002) using data from product 1B11.

Table 1.

TRMM PR 2A23 rain type flags used to identify cloud types and a brief description of the characteristics of each cloud type.

TRMM PR 2A23 rain type flags used to identify cloud types and a brief description of the characteristics of each cloud type.
TRMM PR 2A23 rain type flags used to identify cloud types and a brief description of the characteristics of each cloud type.

The TRMM 3B42 product (Huffman et al. 2007) is primarily based on microwave and IR measurements with PR and rain gauge data used for calibration. This product has a temporal resolution of 3 h and a horizontal resolution 0.25° × 0.25°. To facilitate comparisons with the PR, 3B42 data were degraded to a 1.0° × 1.0° grid, the resolution the PR data was binned to.

b. Reanalysis data sources

The Interim European Centre for Medium Range Weather Forecasting (ECMWF) Re-Analysis (ERA-Interim; Dee et al. 2011) is used to characterize the large-scale convective environment. This reanalysis has a horizontal grid spacing of approximately 0.7° × 0.7° and 27 vertical levels between 1000 and 100 hPa. It is important to note that reanalyses can have substantial biases over tropical Africa and the east Atlantic due to the lack of observations and influence of the convection parameterization. However, for the purposes of this study, the advantages provided by a long-term homogeneous dataset outweigh these concerns. Furthermore, there is reason to suspect that the ERA-Interim reanalysis is sufficiently accurate for drawing qualitative conclusions. For example, Nicholls and Mohr (2010) performed an analysis of the environment of intense convective systems over West Africa using both radiosondes and ECMWF operational analyses and found that both datasets yielded qualitatively similar results.

c. Calculation of TRMM statistics

All statistics presented in this paper are computed over the period of July–September (JAS) 1998–2012. Similar to Liu et al. (2008), contiguous areas of measurable rain from the PR near-surface rain rate are used to define the boundaries of precipitation features. Following Liu et al. (2008), TMI data were collocated with the PR rain rates using a nearest-neighbor interpolation with parallax correction to account for the angle of the TMI measurements. The maximum echo-top height from the PR reflectivity and minimum PCT from the TMI were determined for each PF and used as a proxy for the intensity of the convective updrafts within the PFs. The six PF attributes used in this study (Table 2) have been used in numerous other studies (e.g., Zipser et al. 2006; Liu et al. 2007; Liu 2011). Most of the analyses using orbital data were calculated on a 1.0° × 1.0° grid by binning PR pixels falling within each grid box.

Table 2.

Source TRMM variables and products along with descriptions of each PF attribute.

Source TRMM variables and products along with descriptions of each PF attribute.
Source TRMM variables and products along with descriptions of each PF attribute.

Composites of TRMM PR statistics as a function of reanalysis variables (section 5) were created by first binning the PR orbital data to a daily 2.5° × 2.5° grid and degrading the reanalyses to the same resolution. Daily data were used to focus on the synoptic-scale variability and the horizontal resolution was degraded for computational expediency and since daily reanalysis fields tend to be dominated by spatial variability on larger scales. Next, PR statistics were binned according to the reanalysis values in matching grids. The boundaries of these bins were determined from the percentiles of reanalysis variables within several regional domains (Fig. 1a). Those regional domains are identified as: Atlantic (ATL), coast (CST), northern West Africa (NWA), central West Africa (CWA), southern West Africa (SWA), and East Africa (EAF). Five bins were created for each domain and reanalysis variable using the 20th, 40th, 60th, and 80th percentiles as the dividers.

Fig. 1.

Unconditional rain rate (mm h−1) from (a) TRMM 3B42 and (b) the TRMM PR. Terrain >750 m is contoured in black in this and subsequent figures. The unstippled region is used for calculating the PF size and intensity categories. The black boxes labeled in (a) are the boundaries of the six domains: Atlantic (ATL), coast (CST), northern West Africa (NWA), central West Africa (CWA), southern West Africa (SWA), and East Africa (EAF). Topographic features in (b) are the Guinea Highlands (G), Aïr Mountains (A), Jos Plateau (J), Cameroon Highlands (C), Darfur Highlands (D), Tondou Massif (T), Rift Valley Highlands (RV), and Ethiopian Highlands (E).

Fig. 1.

Unconditional rain rate (mm h−1) from (a) TRMM 3B42 and (b) the TRMM PR. Terrain >750 m is contoured in black in this and subsequent figures. The unstippled region is used for calculating the PF size and intensity categories. The black boxes labeled in (a) are the boundaries of the six domains: Atlantic (ATL), coast (CST), northern West Africa (NWA), central West Africa (CWA), southern West Africa (SWA), and East Africa (EAF). Topographic features in (b) are the Guinea Highlands (G), Aïr Mountains (A), Jos Plateau (J), Cameroon Highlands (C), Darfur Highlands (D), Tondou Massif (T), Rift Valley Highlands (RV), and Ethiopian Highlands (E).

For both the climatological composites of TRMM statistics and the previously mentioned composites as a function of reanalysis variables, the statistical robustness of the results was assessed using a bootstrap analysis (e.g., Iida et al. 2010; Barnes and Houze 2013). In each of the 1000 bootstrap simulations, TRMM statistics over a given regional domain are recalculated by drawing from daily binned data randomly with replacement using all available days. The number of times daily data is sampled in each simulation is equal to the number of days during JAS 1998–2012. The 95% uncertainty range (±2 standard deviations) was determined for each statistic in each regional domain using these 1000 simulations.

3. Regional variability of rain and reflectivity

This section examines regional differences in the properties of rain and reflectivity derived from the PR. Figure 1 shows the average rain rate from the 3-hourly TRMM 3B42 product and the PR near-surface rain rate. It demonstrates that, despite having approximately 24 times less temporal coverage, the PR is able to capture the spatial variability of rainfall. Over the continent, the highest PR rain rates tend to be collocated with topographic features such as the Guinea Highlands, Jos Plateau, Cameroon Highlands, Tondou Massif, Rift Valley Highlands, and Ethiopian Highlands (topographic features identified in Fig. 1b). Over the east Atlantic, the highest rain rates are found in the intertropical convergence zone (ITCZ) between 4° and 12°N. According to the PR, the highest rain rates in the entire region occur in the east Atlantic just west of the Guinea Highlands and in the northeast corner of the Gulf of Guinea near the Cameroon Highlands. Rain rates in these locations range from 0.5 to 0.8 mm h−1 or 1100 to 1760 mm during JAS.

Several domains are illustrated in Fig. 1. The unstippled region is used to group PFs by size and intensity (discussed in section 4). The black boxes denote the boundaries of six regional domains. These domains were chosen to examine land–ocean, north–south, and east–west contrasts in the properties of convection. Three rainfall statistics used throughout this study are shown in Table 3 for each regional domain. The unconditional rain rate (top rows) includes zeroes in the averaging while the conditional rain rate (bottom rows) does not. As a result, the conditional rain rate is a measure of the intensity of the rainfall when it is present. Rain coverage (middle rows) is simply the fraction of the time that rain is present. The unconditional rain rate can be modulated by changes in rain coverage and the conditional rain rate. The contrasts in regional statistics emphasized in this section are statistically significant with respect to sampling errors according to the bootstrap analysis introduced in section 2c.

Table 3.

Unconditional rain rate (mm h−1), rain coverage (%), and conditional rain rate (mm h−1) within each domain. Rain type abbreviations are total (T), stratiform (S), convective (C), shallow convective (SC), and deep convective (DC). The contribution (%) of rain types to the unconditional rain rate and rain coverage within each domain are shown in parentheses.

Unconditional rain rate (mm h−1), rain coverage (%), and conditional rain rate (mm h−1) within each domain. Rain type abbreviations are total (T), stratiform (S), convective (C), shallow convective (SC), and deep convective (DC). The contribution (%) of rain types to the unconditional rain rate and rain coverage within each domain are shown in parentheses.
Unconditional rain rate (mm h−1), rain coverage (%), and conditional rain rate (mm h−1) within each domain. Rain type abbreviations are total (T), stratiform (S), convective (C), shallow convective (SC), and deep convective (DC). The contribution (%) of rain types to the unconditional rain rate and rain coverage within each domain are shown in parentheses.

Figure 2 shows the contribution of shallow convective, deep convective, and stratiform rain to the total rainfall at each grid point. The contribution of different cloud types to the rainfall in each of the domains is also shown in parentheses in the top row of Table 3. Shallow convection accounts for <1% of the rainfall in the continental domains (NWA, CWA, and EAF; all domains are shown in Fig. 1), approximately 4%–5% in the coastal domains (CST and SWA), and 7.2% over the east Atlantic (ATL). The ratio of deep convective to stratiform rain is greatest over the arid parts of the continent (1.86 over NWA) and lowest over the east Atlantic (0.82 over ATL) and largely consistent with previous studies (e.g., Schumacher and Houze 2006).

Fig. 2.

Average (a) shallow convective, (b) deep convective, and (c) stratiform rain fraction (%) from the TRMM PR during JAS 1998–2012.

Fig. 2.

Average (a) shallow convective, (b) deep convective, and (c) stratiform rain fraction (%) from the TRMM PR during JAS 1998–2012.

Funk et al. (2013) suggest that pixels identified by rain type flags 140 and 152 should be reclassified from stratiform to convective because warm rain processes are evident in vertical reflectivity profiles of each. The ATL domain receives the largest fraction of its rainfall from pixels matching these rain type flags (2.9% from type 152 and 0.5% from type 140). The main effect of reclassifying these rain types as shallow and deep convective, respectively, would be an increase in the shallow convective rain fraction over the ocean and a reduction of the land–ocean contrast in stratiform fraction. However, reclassifying these rain types would not affect the overall conclusions of the paper and for this reason we use the original classifications.

Figure 3 shows the conditional average near-surface rain rate and near-surface reflectivity for deep convection and stratiform. The conditional deep convective rain rates (Fig. 3a) are greatest in the CST domain (12.24 mm h−1) and the stratiform rain rates (Fig. 3b) are greatest in the ATL domain (2.69 mm h−1) and decrease moving toward the subtropics. Over the ATL domain, shallow convection accounts for 12.3% of the total area or 46.2% of the convective area (Table 3). Because of the large amount of shallow convection in the ATL domain, conditional convective (7.1 mm h−1) and deep convective (11.31 mm h−1) rain rates differ significantly despite being very similar in the semiarid continental domains. This illustrates the importance of separating convective precipitation into its shallow and deep components.

Fig. 3.

Average (a) deep convective and (b) stratiform conditional rain rate (mm h−1) and (c) deep convective and (d) stratiform conditional near-surface reflectivity (dBZ) from the TRMM PR during JAS 1998–2012.

Fig. 3.

Average (a) deep convective and (b) stratiform conditional rain rate (mm h−1) and (c) deep convective and (d) stratiform conditional near-surface reflectivity (dBZ) from the TRMM PR during JAS 1998–2012.

In contrast to the rain rates, the conditional deep convective near-surface reflectivity (Fig. 3c) is greatest over the Sahel and the conditional stratiform near-surface reflectivity (Fig. 3d) is similar the over tropical land and ocean. The TRMM rain profiling algorithm uses attenuation estimates from the surface reference technique to adjust the drop size distribution (DSD) model by altering the parameter a in the Z–R relationship (Z = aRb) (Iguchi et al. 2000, 2009; Kozu et al. 2009). When reflectivity values are low and attenuation is weak this technique does not work and a default DSD model is used (Iguchi et al. 2009). Because stratiform clouds tend to have weaker reflectivity and attenuation, the DSD model is only adjusted in the most intense stratiform echoes. Kozu et al. (2009) showed that this algorithm estimates a and drop sizes to be smaller over the ocean than the land, which is qualitatively consistent with surface disdrometer observations. These adjustments to the Z–R relationship have large effects on the conditional rain rates. At near-surface reflectivity values within 0.5 dBZ of 40 dBZ, conditional deep convective rain rates are 10 and 18 mm h−1 over the NWA and ATL domains, respectively. This effect contributes to the discrepancy between the geographic variability of conditional deep convective rain rates shown here (Fig. 3a) and those of Schumacher and Houze (2006, see their Fig. 3g), who apply a uniform Z–R relationship to the 2-km reflectivity.

Several studies suggest that the PR near-surface rain rate underestimates intense rainfall over land (e.g., Kirstetter et al. 2013; Rasmussen et al. 2013). This underestimation may be due to the Z–R relationship, nonuniform beam filling effects, or the attenuation correction. Motivated by the analysis of Rasmussen et al. (2013), we consider how conditional rain rates from the CWA and ATL domains would differ if they were computed by applying Z–R relationships from the literature to the PR near-surface reflectivity. For the CWA domain, we examine two Z–R relationships that distinguish between convective and stratiform rainfall and one that does not (Table 4). Two of the Z–R relationships were derived using measurements from Africa (Sauvageot and Lacaux 1995; Gosset et al. 2010) and the other used measurements from an intense squall line over Mississippi (Uijlenhoet et al. 2003). Despite the difference in latitude of these two locations, the geographic variability of a determined from the PR suggests they have qualitatively similar DSDs (Kozu et al. 2009). For the ATL domain, we use the Z–R relationships of Cunning and Sax (1977) and Austin and Geotis (1979) derived from Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE) measurements and that of Tokay and Short (1996), which was produced using measurements from the tropical Pacific. We also examine the effect of applying the same Z–R relationship to both the CWA and ATL domains by applying the Z–R relationship used in Schumacher and Houze (2006).

Table 4.

Comparison between conditional deep convective and stratiform rain rates (mm h−1) derived from the PR near-surface reflectivity using Z–R relationships from the literature and the PR near-surface rain rate. The difference (%) between the rain rates estimated using the Z–R relationships and those in Table 3 are shown in parentheses.

Comparison between conditional deep convective and stratiform rain rates (mm h−1) derived from the PR near-surface reflectivity using Z–R relationships from the literature and the PR near-surface rain rate. The difference (%) between the rain rates estimated using the Z–R relationships and those in Table 3 are shown in parentheses.
Comparison between conditional deep convective and stratiform rain rates (mm h−1) derived from the PR near-surface reflectivity using Z–R relationships from the literature and the PR near-surface rain rate. The difference (%) between the rain rates estimated using the Z–R relationships and those in Table 3 are shown in parentheses.

Table 4 shows that the Z–R relationships from the literature produce higher deep convective conditional rain rates over the CWA domain than the PR near-surface rain rate. One explanation for this discrepancy is that the increase in a, which indicates large drop sizes and results in lower rain rates for a given reflectivity value, may be overdone by the TRMM algorithm. However, the Z–R relationship of Schumacher and Houze (2006) produces much higher conditional deep convective rain rates than the Z–R relationships derived using African measurements and may be inappropriate for this location.

In contrast, the Z–R relationships from the literature produce lower conditional stratiform rain rates than the PR algorithm over the CWA domain. Closer examination of the relationship between the near-surface rain rate and reflectivity (not shown) shows that when the stratiform reflectivity is high and the DSD model is altered, a increases and becomes more in line with Z–R relationships from the literature. The overestimation of rain rates relative to Z–R relationships from the literature may then be because the DSD model is only altered for the most intense stratiform echoes instead of all of them.

Over the ATL domain, some of the Z–R relationships yield higher conditional rain rates than the TRMM algorithm and others produce lower rain rates (Table 4), making it difficult to draw any firm conclusions. However, it must be stressed that applying the continental Z–R relationships to the ocean would yield dramatically lower rain rates than the PR near-surface rain rate and any of the oceanic Z–R relationships. For example, applying the Z–R relationship of Gosset et al. (2010) to the ATL domain would result in conditional deep convective and stratiform rain rates 41.4% and 37.4% lower than the PR near-surface rain rates.

To summarize, the conditional deep convective rain rates in Fig. 3a are likely underestimated over the Sahel. This result is similar to what Rasmussen et al. (2013) found in southern South America, another region that experiences intense convection. In contrast, conditional stratiform rain rates over the Sahel (Fig. 3b) may be overestimated. In addition, applying one Z–R relationship from the literature to all regions would ignore the real differences in DSD indicated by the PR and Z–R relationships from the literature. However, attempting to apply different Z–R relationships to different regions would be prohibitively difficult. Therefore, we use the TRMM PR near-surface rain rate for the remainder of the study while acknowledging its limitations.

To examine the vertical structure of convection in each region, Fig. 4 shows contoured frequency–altitude diagrams (CFADs) constructed from reflectivity profiles where deep convective or stratiform rain is detected. These CFADs show the frequency (% of observations) of reflectivity values at each level when near-surface rain is detected. The most striking difference between the deep convective CFADs is that echoes are detected above 12 km much more frequently in the continental domains (NWA, CWA, and EAF) than the coastal and oceanic domains (ATL, CST, and SWA) (Figs. 4a–f). In addition, the stratiform CFADs suggest a deeper layer of snow and ice in the NWA and CWA domains than the others (Figs. 4g–l). This is consistent with the increased depth of the deep convection in these domains (Figs. 4a–f). Similar to previous studies (Hirose and Nakamura 2004; Geerts and Dejene 2005; Fuentes et al. 2008), the continental domains also show a decrease in reflectivity values between the melting level (~4.5 km) and the surface in the stratiform CFADs. The mean reflectivity decreases 2.62 dBZ between 4 and 1.5 km over the NWA domain but increases by 2.27 dBZ over the ATL domain. This suggests that there is more evaporation taking place in the continental stratiform than the oceanic stratiform due to the lower mean humidity. Evaporation preferentially evaporates smaller drops, resulting in a larger mean drop size (Munchak et al. 2012), so these results are consistent with higher values of a in continental stratiform Z–R relationships (Table 4).

Fig. 4.

CFAD diagrams showing the reflectivity frequency (%) for (a)–(f) deep convective and (g)–(l) stratiform precipitation in each regional domain (Fig. 1a). Note that the x axes are different for the deep convective and stratiform CFADs. Data are from the TRMM PR during JAS 1998–2012.

Fig. 4.

CFAD diagrams showing the reflectivity frequency (%) for (a)–(f) deep convective and (g)–(l) stratiform precipitation in each regional domain (Fig. 1a). Note that the x axes are different for the deep convective and stratiform CFADs. Data are from the TRMM PR during JAS 1998–2012.

4. Regional variability of precipitation features

This section discusses regional differences in the properties of convection determined from statistical summaries of the PF attributes. As in section 3, we only focus on those regional differences that are statistically significant. Figure 5 shows the attribute value at which 50% of the rainfall comes from PFs with attribute values higher and lower than the displayed value. These figures provide a quick way of determining what types of convective systems matter most to the rainfall in a given region. This figure is similar to Figs. 35 from Liu (2011) except that the time period and PF attributes examined are slightly different.

Fig. 5.

At each grid point 50% of the rainfall comes from PF attribute values above and below those plotted. Attributes are PF (a) area (km2), (b) volumetric rain rate (km2 mm h−1), (c) maximum 18-dBZ echo top (km), (d) maximum 40-dBZ echo top (km), (e) minimum 85-GHz PCT (K), and (f) minimum 37-GHz PCT (K). In (d) regions of gray shading indicate areas where PFs with no 40-dBZ echoes account for ≥50% of the rainfall.

Fig. 5.

At each grid point 50% of the rainfall comes from PF attribute values above and below those plotted. Attributes are PF (a) area (km2), (b) volumetric rain rate (km2 mm h−1), (c) maximum 18-dBZ echo top (km), (d) maximum 40-dBZ echo top (km), (e) minimum 85-GHz PCT (K), and (f) minimum 37-GHz PCT (K). In (d) regions of gray shading indicate areas where PFs with no 40-dBZ echoes account for ≥50% of the rainfall.

Figure 5a shows that PFs with areas greater than 104 km2 account for 50% of the rainfall over much of tropical Africa and the east Atlantic. However, much of the South Atlantic (the white area) is characterized by PF areas less than 100 km2 or several pixels. The northern Sahara, high peaks of the Ethiopian and Rift Valley Highlands, and the coast north of the Gulf of Guinea (5°N, 5°W) are also characterized by especially small PFs. These results are similar to what has been found using geostationary IR (Machado et al. 1992). However, the TRMM PR can determine if a cold cloud is really composed of several small convective cells with a common anvil or if it is a contiguous area of precipitation. Consistent with previous studies (Zhou et al. 2013), the PF area (Fig. 5a) and volumetric rain rate (Fig. 5b) have a similar geographic variability, implying that these two quantities are closely related. For this reason we focus on PF area through the remainder of this study.

All four intensity metrics in Fig. 5 indicate that convective systems with deep or intense updrafts are especially important to rainfall in the continental domains (NWA, CWA, and EAF). To the north and south of 15°N, convective systems with weaker intensities are more important to the rainfall. Similar to Fig. 3, topographic features do not stand out prominently, indicating that topographic features increase the frequency of convective systems but do not modify PF intensity. Because of the similar geographic variability of the four intensity metrics, 40-dBZ echo tops will be used as a proxy for convective updraft intensity through the remainder of this study.

It is important to note that Fig. 5 says nothing about the total rain rate from large or intense convective systems. For example, convective systems of a given intensity could be more important to regions near 15°N but less frequent than at 10°N because the region to the south has more convective systems and rainfall overall. This issue will be explored in the following two subsections.

a. Convective systems of different sizes

To investigate the importance of convective systems of different sizes to the regional rainfall, this study follows an approach similar to that of Romatschke and Houze (2011a,b). PFs were separated into three size categories with each category accounting for one-third of the rainfall in the full domain (the unstippled region in Fig. 1). However, unlike Romatschke and Houze (2011a,b) we do not ignore the rainfall from very small PFs. As a result, our size categories account for all raining PR pixels. The area (A) and effective diameters of the three categories are: small (A < 4340 km2, d ≤ 74 km), medium (A = 4340–25 852 km2, d = 75–181 km), and large (A > 25 852 km2, d ≥ 182 km). Figure 6a shows the number of PFs and rain type makeup of each size category. Within individual size categories, the stratiform fraction increases from 27% in the small group, to 45% in the medium group, to 58% in the large group. Shallow convective rain only makes a meaningful contribution (12%) to the rain from small PFs.

Fig. 6.

(a) Number of small (D ≤ 74 km), medium (D = 75–181 km), and large (D ≥ 182 km) PFs and the rainfall contribution (%) of the total, stratiform, convective, shallow convective (SC), and deep convective (DC) rainfall from each category. (b) As in (a), but for low-top (no 40 dBZ and ≤5 km), medium-top (5.25–6.5 km), and high-top (≥6.75 km) PFs. Statistics are computed over the full domain (unstippled region in Fig. 1).

Fig. 6.

(a) Number of small (D ≤ 74 km), medium (D = 75–181 km), and large (D ≥ 182 km) PFs and the rainfall contribution (%) of the total, stratiform, convective, shallow convective (SC), and deep convective (DC) rainfall from each category. (b) As in (a), but for low-top (no 40 dBZ and ≤5 km), medium-top (5.25–6.5 km), and high-top (≥6.75 km) PFs. Statistics are computed over the full domain (unstippled region in Fig. 1).

Figure 7 shows PFs representative of the three size categories in geostationary IR imagery (left panels) and PR near-surface rain rate (right panels). The lower right panel illustrates the ability of the PR to distinguish between the convective and stratiform regions of organized convection. Generally speaking, the three PF size categories are associated with different morphologies. The small PFs include isolated deep convective cells over the Sahel and a mix of deep and shallow convective cells over the east Atlantic. The medium-sized PFs are small or average-sized MCSs; often their stratiform regions are not well separated from the convective regions. The large PFs are large MCSs with well-separated stratiform regions. Many of the large PFs over the continent would meet the criteria for mesoscale convective complexes (MCCs) (Maddox 1980; Laing et al. 1999).

Fig. 7.

Representative illustrations of the (top) small, (middle) medium, and (bottom) large PFs. Panels show (left) zoomed-out images of IR BT (K) and (right) zoomed-in images of TRMM PR rain rate (mm h−1). Event time stamps are shown in the right side panels. The white boxes in the left panels indicate the domain of the right panels.

Fig. 7.

Representative illustrations of the (top) small, (middle) medium, and (bottom) large PFs. Panels show (left) zoomed-out images of IR BT (K) and (right) zoomed-in images of TRMM PR rain rate (mm h−1). Event time stamps are shown in the right side panels. The white boxes in the left panels indicate the domain of the right panels.

Figure 8 shows the average rain rate (left panels) and the fraction of the total rain rate (right panels) at each 1.0° × 1.0° grid point from small, medium, and large PFs. Since all raining PR pixels are accounted for, the left panels sum to the rain rate in Fig. 1b. The regions that receive the largest amount of rainfall from small PFs are the major mountain ranges and the portion of the east Atlantic west of 35°W (Fig. 8a). However, the Sahara and South Atlantic receive the largest fraction of rainfall from small PFs (Fig. 8b). These regions also have a very low stratiform fraction (Fig. 2c), which is consistent with the fact that small systems tend to have little stratiform rainfall (Fig. 6a). The largest amounts of rainfall from medium PFs occur along the Guinea coast, Cameroon Highlands, and Ethiopian Highlands (Fig. 8c); this category accounts for 30%–50% of the rainfall over most of the African continent and the portion of the Atlantic between 5° and 15°N (Fig. 8d). The rain rate from large PFs is greatest along the eastern portion of the east Atlantic ITCZ (Fig. 8e). The regions that receive the greatest fraction of their rainfall from large PFs are the eastern portion of the east Atlantic ITCZ, the West African Sahel, and the western Congo Basin (4°S, 18°E) (Fig. 8f). Many other studies have noted the importance of MCSs to rainfall over the Sahel (Mathon et al. 2002).

Fig. 8.

Average (left) rain rate (mm h−1) and (right) rain fraction (%) from (a),(b) small, (c),(d) medium, and (e),(f) large PFs during JAS 1998–2012.

Fig. 8.

Average (left) rain rate (mm h−1) and (right) rain fraction (%) from (a),(b) small, (c),(d) medium, and (e),(f) large PFs during JAS 1998–2012.

Some inferences into the life cycle of the PFs can be made from Fig. 8. Over the continent, the mountain ranges receive the largest fraction of rainfall from small PFs and the plains of the Sahel receive the largest fraction of rainfall from large PFs. This suggests that convection is triggered over the mountains and grows into large PFs over the Sahel, consistent with Laing et al. (2008). This illustrates the insights into convective system formation and evolution that can be drawn from PF sizes. Further support for this inference is provided in section 6, which examines the diurnal cycle of convection.

Detailed analyses of the rain type and conditional rain rates within each size category were performed for each of the six domains and are briefly discussed. The domains with the highest and lowest stratiform fraction within large PFs were the ATL at 65% and the interior continental domains (NWA, CWA, and EAF) at 50%–52%. Furthermore, the stratiform fraction was highest for the large PFs and lowest for the small PFs in each of the six regional domains. Shallow convection accounted for 31% of the small PF rainfall in the ATL and 87% of the shallow convective rain in the ATL came from small PFs. We also found that conditional rain rates increase with the size of the PFs. In the CWA domain, small PFs have conditional deep convective and stratiform rain rates of 7.64 and 1.20 mm h−1; for the large PFs these values were 15.71 and 2.75 mm h−1 (irrespective of PF sizes these values are 10.96 and 2.24 mm h−1; Table 3). The relationship between PF size and conditional rain rates is illustrated further in section 5 where we examine how these statistics vary with the environmental moisture, conditional instability, and shear in each regional domain.

b. Convective systems of different intensities

Similar to what was done with the PF sizes, PFs with different maximum 40-dBZ echo tops were separated into three categories, each accounting for one-third of the rainfall in the full domain (Fig. 6b). The low-top category contains PFs with maximum 40-dBZ echo tops ≤ 5 km and PFs with no 40-dBZ echoes. PFs with no 40-dBZ echoes accounted for about one-third of the rainfall from the low-top category. In addition, this is the only category where shallow convection makes a meaningful contribution to the rainfall (12%). The medium-top category has maximum 40-dBZ echo tops between 5.25 and 6.5 km. The high-top category has maximum 40-dBZ echo tops ≥6.75 km. All of the medium- and high-top PFs have 40-dBZ echo tops located above the melting level, which is near 4.5 km. The high-top category receives the lowest fraction (34%) of its rainfall from stratiform precipitation. There are two reasons why the high-top PFs have the smallest stratiform fraction. First, maximum 40-dBZ echo tops are typically located within the convective region of precipitation systems, making this intensity metric more sensitive to the convective portion of the PFs. Second, the high-top PFs are mostly located over the continent where the stratiform rain fraction is lower (Fig. 2c).

Figure 9 shows the average rain rate (left panels) and the fraction of the total rain rate (right panels) at each 1.0° × 1.0° grid box from the low-, medium-, and high-top PFs defined above. The two regions with the largest amount of rainfall from low-top PFs are the east Atlantic between 5° and 10°N and the northeastern corner of the Gulf of Guinea (Fig. 9a). In contrast, the regions that receive the largest fraction of rainfall from low-top PFs are the Sahara and subtropical Atlantic (Fig. 9b). The medium-top PF category is important to rainfall in the eastern portion of the east Atlantic between 5° and 20°N and over the continent between the equator and 10°N (Fig. 9d). The high-top PFs dominate the rainfall poleward of this band in both the Sahel and over the Congo basin (Fig. 9f). Some parts of these regions receive over 80% of their rainfall from the high-top PFs. Because the PR near-surface rain rate tends to underestimate rainfall from intense convection (Rasmussen et al. 2013), this fraction would increase if the Z–R relationships in Table 4 were used to compute the rain rates.

Fig. 9.

Average (left) rain rate (mm h−1) and (right) rain fraction (%) from (a),(b) low-top, (c),(d) medium-top, and (e),(f) high-top PFs during JAS 1998–2012.

Fig. 9.

Average (left) rain rate (mm h−1) and (right) rain fraction (%) from (a),(b) low-top, (c),(d) medium-top, and (e),(f) high-top PFs during JAS 1998–2012.

A detailed analysis of the rain types and conditional rain rates for each of the PF intensity categories in each domain was also performed. As would be expected, meaningful amounts of shallow convective rain (4%–5%) are only observed in the low-top PFs over the ATL, CST, and SWA domains. In addition, higher conditional rain rates were found in the high-top PFs than the other intensity categories in each of the six domains. We also found that the high-top PFs in the ATL and CST domain had higher conditional rain rates than the high-top PFs in the other domains. However, this result is highly sensitive to the assumed DSD and Z–R relationship as illustrated in section 3.

5. Relationship between convection and its environment

To provide an overview of the convective environment, Fig. 10 shows the mean and standard deviation of daily column relative humidity (CRH), convective available potential energy (CAPE) taken from the ERA-Interim forecasts, and 925–600-hPa shear. Mean CRH values over 75% are found over much of the east Atlantic ITCZ, parts of coastal West Africa, and the Congo (Fig. 10a). Over the continent, an intense moisture gradient with a high standard deviation of CRH is found between 12° and 20°N between the Saharan air layer (SAL) and monsoon air mass. This region is also characterized by fairly high mean CAPE and 925–600-hPa shear (Figs. 10b,c). Lower values of CAPE and shear are found to the north and south of this region and over the ocean.

Fig. 10.

Mean (shaded) and standard deviation (contours) of CRH (%), (b) CAPE (J kg−1), and (c) 925–600-hPa shear (m s−1) with mean shear vectors. CAPE is taken from the ERA-Interim forecasts.

Fig. 10.

Mean (shaded) and standard deviation (contours) of CRH (%), (b) CAPE (J kg−1), and (c) 925–600-hPa shear (m s−1) with mean shear vectors. CAPE is taken from the ERA-Interim forecasts.

We also examined the vertical distribution of moisture in the six regional domains. Over the east Atlantic, the low levels are consistently moist and the variability of CRH is dominated by free-tropospheric variability; this is typical of oceanic regions (e.g., Brown and Zhang 1997; Holloway and Neelin 2009). Periods of dry free-tropospheric air here are associated with both SAL outbreaks (Carlson and Prospero 1972) and large-scale subsidence (Braun 2010). Over the continent, especially near the Sahara, dry air can be found near the surface and in the free troposphere.

It is important to stress that these three environmental variables (CRH, CAPE, and 925–600-hPa shear) are not independent and are related to each other differently in each of the regional domains. Over the ATL domain, CRH and CAPE are positively correlated. Changes in low-level temperatures are small so CAPE primarily responds to changes in low-level moisture. Under this definition of CAPE, entrainment due to midlevel dry air is not accounted for but is likely important (Holloway and Neelin 2009). CAPE and CRH are also positively correlated in the NWA domain. In this domain, low-level temperatures are anomalously cool (3.0 K below climatology) for the highest quintile of CAPE. This region is so moisture limited that, for CAPE to be present, cooler moister air must be advected from the south. In contrast, CAPE and CRH are negatively correlated in the CWA and EAF domains. In these locations, the highest quintile of CRH has anomalously cool low levels (2–3 K below climatology) and CAPE values are reduced by 200–300 J kg−1 relative to climatology. In contrast, the highest quintile of CAPE in these locations has anomalously warm low levels (2–3 K above climatology) and a CRH that is ~3% below climatology.

Figures 1115 show PR statistics as a function of these three reanalysis variables for all six regional domains except the CST domain, where both land and ocean are present and the results are more difficult to interpret. The 20th, 40th, 60th, and 80th percentiles of daily 2.5° × 2.5° reanalysis grids for each domain and variable were used to separate observations from the PR into five bins. We tried using more bins but this produced noisier results. The error bars shown in each figure were calculated using the bootstrap analysis discussed in section 2c.

Fig. 11.

Difference (%) between the unconditional total, shallow, deep convection, and stratiform rain rate within specific percentile ranges of reanalysis variables and the climatological rain rate for that rain type within each domain. The values displayed on the x axes are the (left) CRH (%), (middle) CAPE (J kg−1), and (right) 925–600-hPa shear (m s−1) minimum, 20th, 40th, 60th, and 80th percentiles, and the maximum. Error bars show the ±2 standard deviation estimated from the bootstrap analysis. Climatological rain rates are shown in Table 3. Shallow rain rates are negligible in the NWA, CWA, and EAF domains and are not plotted.

Fig. 11.

Difference (%) between the unconditional total, shallow, deep convection, and stratiform rain rate within specific percentile ranges of reanalysis variables and the climatological rain rate for that rain type within each domain. The values displayed on the x axes are the (left) CRH (%), (middle) CAPE (J kg−1), and (right) 925–600-hPa shear (m s−1) minimum, 20th, 40th, 60th, and 80th percentiles, and the maximum. Error bars show the ±2 standard deviation estimated from the bootstrap analysis. Climatological rain rates are shown in Table 3. Shallow rain rates are negligible in the NWA, CWA, and EAF domains and are not plotted.

Fig. 12.

As in Fig. 11, but for the fractional rain contribution of specific rain types.

Fig. 12.

As in Fig. 11, but for the fractional rain contribution of specific rain types.

Fig. 13.

As in Fig. 11, but for conditional rain rates.

Fig. 13.

As in Fig. 11, but for conditional rain rates.

Fig. 14.

As in Fig. 11, but for the fractional rain contribution of small, medium, and large PFs.

Fig. 14.

As in Fig. 11, but for the fractional rain contribution of small, medium, and large PFs.

Fig. 15.

As in Fig. 11, but for the fractional rain contribution of low-top, medium-top, and high-top PFs.

Fig. 15.

As in Fig. 11, but for the fractional rain contribution of low-top, medium-top, and high-top PFs.

Figure 11 shows the total, shallow, deep convective, and stratiform unconditional rain rates as a function of the three reanalysis variables. So that different domains and rain types can be compared, the rain rates are expressed as percentage differences from climatology for each rain type and domain. In each domain, the total, deep convective, and stratiform rain rates increase with CRH and the stratiform rain rates increase the most. In the ATL domain, the shallow convective rain rate increases slightly between the first and third quintiles of CRH and then decreases (Fig. 11a). In the SWA domain, changes in shallow convective rain are smaller than for the other rain types (Fig. 11j). These results suggest that, when the environment moistens, the shallow clouds develop vertically into deep convective clouds in addition to expanding in coverage.

In some regions (ATL, NWA, and SWA) (Figs. 11b,e,k) there is a significant increase in rain rate with CAPE while in other regions there is not (CWA and EAF) (Figs. 11h,n). The regions where rain rate does not respond to CAPE have the highest mean CAPE (Fig. 10b) and are likely not CAPE limited on many days. Furthermore, as mentioned above, there is a negative correlation between CAPE and CRH in the CWA and EAF domains. Therefore, the effects of increased CAPE may be negated by the fact that the air is drier. In addition, a weak but continuous and significant relationship between shear and rain rate is found in the ATL and SWA domains (Figs. 11c,l), which have lower mean shear and may be ‘shear limited’ (Fig. 10c).

Figure 12 shows how the rain type fractions are modulated by CRH, CAPE, and shear. Consistent with the results in Fig. 11, there is a significant decrease in the shallow convective rain fraction with increasing CRH in the ATL and SWA domains (Figs. 12a,j). Furthermore, in all five domains there is a significant increase in the stratiform rain fraction and decrease in the deep convective rain fraction with increasing CRH. It is not immediately clear why stratiform rainfall should be more sensitive to CRH than deep convective rainfall but other studies have found similar results (Zhou et al. 2013). The dependence of stratiform fraction on CRH may partly explain why the regional variability of the mean CRH (Fig. 10a) and mean stratiform fraction (Fig. 2c) are similar. This will be discussed further in section 7.

The relationship between the importance of each rain type and CAPE and shear is more domain dependent. In both the ATL and SWA domains, shallow convective rain fractions decrease with increasing CAPE (Figs. 12b,k). This suggests that increased CAPE also encourages shallow convection to develop into deep convection. In the NWA domain (Fig. 12e), the stratiform rain fraction increases and the deep convective rain fraction decreases with higher CAPE while the opposite is observed in the CWA and EAF domains (Figs. 12h,n). In addition, there is a small but statistically significant increase in the deep convective rain fraction with higher shear values in the CWA and SWA domains (Figs. 12i,l).

Figure 13 is similar to Fig. 11 but shows the response of conditional deep convective and stratiform rain rates to changes in the large-scale environment. Changes in the conditional rain rate are much smaller than the changes in the unconditional rain rate (Fig. 11) (note that the y axes are different). We also examined how rain area responds to reanalysis fields (not shown) and the changes were much larger than those seen in Fig. 13. This indicates that changes in unconditional rain rate are driven more by the amount of rain coverage than its intensity.

The relationship between CRH and conditional rain rates is strongest in the NWA domain (Fig. 13d). This is likely due to the fact that this is the most moisture-limited domain (Fig. 10a). Statistically significant relationships between the conditional deep convective rain rate and CAPE and shear are seen in all five domains while the enhancement of conditional stratiform rain rates is statistically significant in some domains and not others. In the CWA and EAF domains, there is a statistically significant increase in the conditional deep convective rain rate at higher CAPE and shear values even though there is no significant increase in the unconditional rate. The high CAPE and shear days see more intense rain rates but no increase in rain coverage.

Figures 14 and 15 show how the fractional rain contribution from the PF size and intensity categories changes as a function of the large-scale environment and clarifies how the changes in the conditional rain rate (Fig. 13) relate to changes in the convective morphology. The relative importance of small and large PFs has a very strong dependence on CRH in the ATL and NWA domains (Figs. 14a,d). Because rain coverage increases dramatically with CRH in these regions, the connectivity between the PFs increases and so do their typical sizes. In the CWA domain (Fig. 14g), there is a slight but significant shift toward smaller PFs during the driest periods while the other periods are nearly identical. Higher values of CAPE and shear are also associated with a shift toward more organized convection in most domains but the magnitude of this shift varies.

In the ATL domain, as the CRH increases the importance of medium-top PFs increases relative to the low-top PFs (Fig. 15a). This is consistent with the tendency for deep convective rain rates to increase relative to shallow convective rain rates (Fig. 12a). The CWA, SWA, and EAF domains have a decrease in the rain fraction from high-top PFs and an increase in the fraction from medium- and low-top PFs (Figs. 15g,j,m). In these regions, the rain rate from high-top PFs does not decrease in an absolute sense but does decrease relative to the medium- and low-top PFs.

In every continental domain, higher values of CAPE and shear are associated with a statistically significant increase in the fraction of the rainfall from high-top PFs. The increases in the rain fraction from high-echo-top PFs with CAPE and shear are consistent with the increases in conditional deep convective rain rates (Fig. 13). In the ATL domain, the fraction of the rainfall from medium-top PFs increases relative to that from low-top PFs (Figs. 15b,c). In many of these domains, the total rain rate is not elevated when the CAPE and shear are high (Fig. 11); therefore, it is the distribution of PF properties that responds to these fields. It has been argued in previous studies (Mohr and Thorncroft 2006; Nicholls and Mohr 2010) that the high mean frequency of intense convection over the Sahel (Figs. 9e,f) is related to the high mean conditional instability and shear (Figs. 10b,c); these analyses reinforce this finding.

Figure 16 shows how deep convective and stratiform CFADs change in response to the CRH in the ATL domain. On drier days the frequency of deep and intense deep convective echoes decreases (Fig. 16a). A similar shift is observed in the stratiform CFADs (Fig. 16c) and is consistent with the reduction in ice seeding that would occur if the deep convection were weaker. It is also possible that the stratiform is responding directly to the drier middle to upper-tropospheric humidity as suggested by (Del Genio et al. 2012). In contrast, more humid days see a shift to deeper and more intense deep convective and stratiform echoes (Figs. 16b,d), which is consistent with the changes in conditional rain rates (Fig. 13a) and the PF intensity categories (Fig. 15a). Similar changes are observed in CFADs constructed using the lowest and highest quintiles of CAPE in this region.

Fig. 16.

Climatological CFADs taken from Fig. 4 showing reflectivity frequency (contoured every 0.5%) and deviations from climatology (% shaded) for (a),(b) deep convective and (c),(d) stratiform precipitation in the ATL domain. Left (right) panels show deviations from climatology for the driest (most humid) quintile of CRH. The conditional rate rates and their deviation from climatology (see Table 3 for climatological values) are shown in the upper right for reference.

Fig. 16.

Climatological CFADs taken from Fig. 4 showing reflectivity frequency (contoured every 0.5%) and deviations from climatology (% shaded) for (a),(b) deep convective and (c),(d) stratiform precipitation in the ATL domain. Left (right) panels show deviations from climatology for the driest (most humid) quintile of CRH. The conditional rate rates and their deviation from climatology (see Table 3 for climatological values) are shown in the upper right for reference.

The relationship between CFADs and CRH in the CWA domain is somewhat more complicated (Fig. 17). On dry days, both the deep convective and stratiform CFADs have a clear increase in the frequency of high echo tops (Figs. 17a,c) while the opposite is observed on the humid days (Figs. 17b,d), consistent with Fig. 15g. These changes are likely not due to the changes in humidity but changes in CAPE. During the driest quintile the CAPE is 153 J kg−1 above the mean while it is 280 J kg−1 below the mean in the most humid quintile. Below the melting level (~4.5 km), the deviations from the climatological CFADs are much smaller. This suggests that differences in low-level evaporation between the low and high CRH days are acting to cancel out the changes occurring aloft. Similar changes are also observed for high and low CRH days in the EAF and SWA domains. The tendency for decreased conditional instability to accompany higher CRH and vice versa also helps explain why the changes in conditional rain rates as a function of CRH are so small in these regions (Figs. 13g,j,m).

Fig. 17.

As in Fig. 16, but for the driest and most humid quintile of CRH in the CWA domain.

Fig. 17.

As in Fig. 16, but for the driest and most humid quintile of CRH in the CWA domain.

6. Diurnal cycle

This section examines the diurnal cycle of rainfall over tropical Africa and parts of the east Atlantic. A major aim of this section is to provide a diurnal context for results from the previous sections, particularly those relating to PF sizes that are strongly modulated by the diurnal cycle. Less emphasis is placed on the physical causes of changes in the properties of convection throughout the diurnal cycle since they have been addressed in previous studies (e.g., Parker et al. 2005a; Laing et al. 2008).

Figure 18 shows the phase (left panels) and amplitude (right panels) of the first harmonic of the diurnal cycle for the TRMM 3B42 and PR rain rates. Because the TRMM 3B42 product has a higher sampling frequency than the PR it has a more coherent depiction of the diurnal cycle. Over most of the topographic features, rain rates peak between 1600 and 2000 LT (Figs. 18a,c). In addition, the topographic features tend to have the highest diurnal cycle amplitudes (Figs. 18b,d). However, this is partly because they have higher total rain rates (Fig. 1). Downstream of the topographic features, the highest rain rates occur in the early morning. Two of the more prominent early morning (0000–0600 LT) rain rate peaks are found near Niamey, Niger (12°N, 2°E), and in the valley between the Cameroon Highlands and the Jos Plateau (8°N, 10°E) (Figs. 18a,c). This picture of the diurnal cycle is consistent with previous studies based on geostationary IR (Laing et al. 2008), TRMM (Hirose et al. 2008; Yamamoto et al. 2008), and ground-based radar measurements (Rickenbach et al. 2009).

Fig. 18.

(a),(c) Phase and (b),(d) amplitude of the first harmonic of the diurnal cycle for the (top) TRMM 3B42 and (bottom) TRMM PR rain rates during JAS 1998–2012.

Fig. 18.

(a),(c) Phase and (b),(d) amplitude of the first harmonic of the diurnal cycle for the (top) TRMM 3B42 and (bottom) TRMM PR rain rates during JAS 1998–2012.

Over the ocean, near the western coast of Africa, the rain rate peaks at 0600–0800 LT and transitions to a 1200 LT peak farther from the coast (Figs. 18a,c). This suggests that convection is either forming or amplifying along the coast in the morning hours and then has moved away from the coast by 1200 LT. The oceanic regions away from the continent have a much weaker and less coherent diurnal cycle than the coastal zones.

To clarify the relationship between rainfall, convective morphology, and topographic features, Fig. 19 shows the diurnal cycle of the TRMM 3B42 and PR rain rates, rain rates for each PF size, and the stratiform rain fraction averaged from 5° to 15°N. Westward propagating diurnal signals, highlighted by dashed lines, are clearly visible in the TRMM 3B42 composite (Fig. 19a) and closely resemble those found by Laing et al. (2008) using geostationary IR imagery. These same westward propagating signals are also apparent in the TRMM PR composite but are considerably noisier (Fig. 19b). The easternmost signal is associated with the Ethiopian Highlands while the signals between 15° and 30°E are associated with the elevated terrain near the Darfur Highlands and Tondou Massif. Farther west, near 10°E, is the signal associated with the Cameroon Highlands and the Jos Plateau. The situation at the coast is more complicated. There is a clear signal associated with the afternoon convection forming over the Guinea Highlands, which has been found in previous studies (Hodges and Thorncroft 1997; Ventrice et al. 2012). However, as this signal moves westward over the ocean it reamplifies and quickly moves westward into the east Atlantic, peaking between 0900 and 1200 LT.

Fig. 19.

Longitude–diurnal cycle plots averaged between 5° and 15°N of the (a) TRMM 3B42, (b) TRMM PR, (d) small PF, (e) medium PF and (f) large PF rain rate (mm h−1), and (c) the stratiform rain fraction (%). Dashed lines indicate the propagation of coherent diurnal cycle signals in the TRMM 3B42 rain rate.

Fig. 19.

Longitude–diurnal cycle plots averaged between 5° and 15°N of the (a) TRMM 3B42, (b) TRMM PR, (d) small PF, (e) medium PF and (f) large PF rain rate (mm h−1), and (c) the stratiform rain fraction (%). Dashed lines indicate the propagation of coherent diurnal cycle signals in the TRMM 3B42 rain rate.

Over land, the rain rate from small PFs is greatest during the afternoon (1500–1800 LT) and tends to peak near topographic features, the source region of the coherent propagating signals (Fig. 19d). This is also shown in Figs. 20a–d, which show the diurnal cycle of small PFs averaged over 6-h blocks. In contrast, the diurnal cycle of rain from small PFs is flat over the ocean. The medium-sized PFs also tend to be concentrated near topographic features in the afternoon (Figs. 19e and 20h) but unlike the small PFs they are still abundant at 0000 LT (Figs. 19e and 20e).

Fig. 20.

Average unconditional rain rate (mm h−1) from (a)–(d) small, (e)–(h) medium, and (i)–(l) large PFs at (a),(e),(i) 0000, (b),(f),(j) 0600, (c),(g),(k) 1200, and (d),(h),(l) 1800 LT during JAS 1998–2012.

Fig. 20.

Average unconditional rain rate (mm h−1) from (a)–(d) small, (e)–(h) medium, and (i)–(l) large PFs at (a),(e),(i) 0000, (b),(f),(j) 0600, (c),(g),(k) 1200, and (d),(h),(l) 1800 LT during JAS 1998–2012.

The rainfall from large PFs is clearly enhanced between 0000 and 0600 LT along the tail ends of the diurnal convective signals propagating away from the Ethiopian highlands at 35°E and the Darfur Highlands and Tondou Massif near 20°E (Fig. 19f). However, this is less evident for the diurnal signal propagating away from the Cameroon Highlands and Jos Plateau near 10°E. It is also difficult to observe these signals in the plan maps (Figs. 20i–l). These subtle features of the diurnal cycle may be too weak for the limited sampling rate of the PR to resolve. Over the Sahel, the high frequency of large PFs in the early morning has been attributed to the fact that shear peaks at this time in association with the development of a nocturnal low-level jet, which supports the development of organized convection (e.g., Parker et al. 2005a; Laing et al. 2008).

The most intense peak in the large PF rain rate is centered at 0900 LT and 15°W (just off the coast of West Africa) (Fig. 19f). Figures 20j and 20k show that this rain rate is concentrated in a strip just off the coast at 0600 LT that moves westward to 18°W by 1200 LT. The fact that this early morning signal is dominated by large PFs and is not associated with a peak in small PFs (Fig. 19d) suggests that convection is being reinvigorated as it moves over the ocean. Convective systems crossing from the land to the ocean in the evening and early morning could be reinvigorated by radiative cooling of anvils or an increase in surface temperature and moisture associated with the warm sea surface.

Over the continent, the stratiform fraction is greatest at 0900 LT (~50%–60%) and lowest at 1500 LT (~30%–35%) (Fig. 19c). This is consistent with the fact that, during the morning, rainfall is dominated by large PFs, which are more stratiform, while during the afternoon rainfall comes from small PFs that are more convective (Fig. 6a). We also examined the diurnal cycle of the three PF intensity categories and found it was much weaker and less coherent than the diurnal cycle of PF sizes. Of the three PF intensity categories, the high-echo-top PFs had the most coherent diurnal cycle. The rain rate from this category peaks between 1800 and 0000 LT over the continent and has a minimum between 0600 and 1200 LT. Other studies have found a similar diurnal cycle of intense convection over land whether they used echo tops or microwave ice scattering signatures (e.g., Nesbitt and Zipser 2003; Liu 2011). The afternoon to early evening peak in intense convection can be attributed to the fact that the low levels are warm and unstable at this time and cooler and more stable in the morning.

7. Discussion and conclusions

This study used data from the TRMM satellite to examine regional differences in the frequency of different cloud types and conditional rain intensity (section 3) as well as the size of convective systems and the intensity of their updrafts based on echo tops and PCTs (section 4). We also examined how the properties of convection are modulated by moisture, conditional instability, and shear over the east Atlantic and different parts of tropical Africa (section 5). The effect of the diurnal cycle on the life cycle and properties of convection was also analyzed (section 6). In this section we discuss possible connections between the findings in these different sections and the relationship with previous studies. We also offer some conclusions.

Similar to previous studies (Schumacher and Houze 2003b, 2006), we found that stratiform fractions were greater over the ocean than the land and highest in the regions just north of the equator that had the highest unconditional rain rates (Fig. 2c). The contrast in stratiform fractions between the equatorial regions and the subtropics is partly due to differences in the degree of convective organization and differences in the relative amounts of shallow and deep convection over the ocean (Figs. 2a,b). Large organized convective systems have higher stratiform fractions than small convective systems (Fig. 6a) due to the presence of mesoscale circulations in the large systems that support stratiform development (Houze 2004). In addition, because shallow convection does not extend above the melting level it cannot produce the ice that contributes to the development of stratiform clouds.

The analyses in section 5 provide insights into the large-scale environmental conditions responsible for stratiform variability. Over the east Atlantic ITCZ (ATL), days with low CRH had large amounts of shallow convection and small convective systems, which are unfavorable for stratiform rain development. These dry days resemble the climatological conditions over the subtropics. While previous studies have found that, over the tropical oceans, entrainment limits the depth of deep convection when free-troposphere humidity is low (e.g., Brown and Zhang 1997; Bretherton et al. 2004; Holloway and Neelin 2009), the connection between this and stratiform rain has not been emphasized. In addition, changes in the depth and intensity of echoes above the melting level as a function of CRH (Fig. 16) raises the possibility that depositional growth or sublimation can also be modified by the middle-tropospheric humidity (e.g., Schumacher and Houze 2006; Del Genio et al. 2012). Taken together, these results suggest a link between the increased free-troposphere humidity in the deep tropics and high stratiform fractions. However, it should be stressed that the ultimate source of this moisture is likely the convection itself.

Over the Sahel (CWA), days with high CRH also had higher stratiform fractions. One possibility is that increased access to low-level moisture or a reduction in downdrafts due to moister midlevels increases the “convective sustainability,” resulting in longer-lived convective systems that have more time to produce stratiform cloud. However, days with high CRH in this region also had cooler low levels, decreased conditional instability, and lower echo tops (Figs. 15 and 17). The increased stability on days with high CRH may be more detrimental to deep convection than to stratiform regions. Similar to previous studies (Hirose and Nakamura 2004; Geerts and Dejene 2005; Fuentes et al. 2008), our results also suggests that low-level humidity can influence the amount of evaporation taking place in the stratiform rain.

According to the PR near-surface rain rate, conditional deep convective and stratiform rain rates are greatest off the coast of West Africa, west of the Guinea Highlands (Figs. 3a,b). However, similar to results of Rasmussen et al. (2013), conditional deep convective rain rates over the Sahel calculated using Z–R relationships from the literature were much higher than the PR near-surface rain rate (Table 4). This implies that the PR precipitation algorithm underestimates conditional deep convective rain rates over the Sahel due to an overestimation of drop sizes. It should be stressed that this does not negate previous results (e.g., Kozu et al. 2009) suggesting that drop sizes are larger over the Sahel than the east Atlantic. Our results merely suggest that this difference is overestimated. Despite the high degree of uncertainty in conditional rain rates owing to DSD assumptions, some conclusions about the relationship between the geographic distribution of conditional rain rates and the large-scale environment can be drawn from our study.

Over the Sahel, conditional deep convective and stratiform rain rates from large PFs (large MCSs) were over twice as high as those from small PFs (convective cells). Similar differences were observed in the other regions and in a global analysis by Hirose et al. (2009). We also found that conditional rain rates in PFs with high 40-dBZ echo tops were greater than in PFs with lower echo tops. Both these results are fairly intuitive. Systems with higher 40-dBZ echo tops have stronger updrafts and are able to condense larger amounts of water vapor. In addition, MCSs possess mesoscale circulations that can draw upon the water vapor over a large area as well as contribute to the strength of the updrafts (Houze 2004). It is also possible that the sizes and echo-top heights of the PFs are correlated as found in Liu (2011).

A large fraction of the rainfall that falls off the coast of West Africa and over the Sahel comes from large convective systems. The presence of high conditional rain rates in these regions may be related to fact that such a large portion of the rainfall in this region comes from organized convection. The Sahel, which receives an especially large portion of its rainfall from high-echo-top PFs (Fig. 9f), has only slightly higher conditional deep convective rain rates than the surrounding regions according to the PR near-surface rain rate (Fig. 3a). However, as discussed in section 3, conditional rain rates may be underestimated in this region.

Over the Sahel and other continental regions, days with high CAPE and shear received a much larger fraction of their rain from high-echo-top PFs than typically occurs (Fig. 15). In many regions there was also an increase in the typical size of convective systems but this increase was smaller and not statistically significant in every region (Fig. 14). Conditional deep convective rain rates were also significantly higher than climatology on these days (Fig. 13), which is consistent with the increased organization and vertical development of the convection. However, in the CWA and EAF domains there was no corresponding increase in the unconditional rain rate (Fig. 11) or rain coverage, indicating that the number of systems did not increase when the CAPE or shear was elevated. Our results are consistent with those of Nicholls and Mohr (2010), who compared the environment of convective systems of varying intensity, determined from 85-GHz ice scattering signatures, over West Africa and found that the more intense convective systems had higher values of surface θe and low-level shear. Our analyses also suggest strengthen the finding in previous studies (e.g., Hodges and Thorncroft 1997; Parker et al. 2005b; Mohr and Thorncroft 2006) that the intensity of convection over the Sahel is linked to the favorable shear and instability found south of the SAL near the AEJ.

Nguyen and Duvel (2008) examined the diurnal cycle over equatorial Africa during convectively suppressed and active periods associated with convectively coupled Kelvin waves. During the active periods, large early morning convective systems were enhanced more than small afternoon systems, which resulted in a relatively flat diurnal cycle. In contrast, suppressed periods were dominated by afternoon convection from small convective cells. It is reasonable to expect that the modification of convective systems sizes associated with changes in moisture, conditional instability, and shear shown in Fig. 14 would result in similar changes in the diurnal cycle.

Future work will examine how AEWs modulate the convective environment (e.g., moisture, conditional instability, and shear) and the properties of convection according to TRMM using similar metrics. We also plan to examine how the diurnal cycle is influenced by changes in the convective regime.

Acknowledgments

The authors thank Courtney Schumacher and the two anonymous reviewers for their insightful comments which greatly improved the manuscript. TRMM data used in this study were acquired as part of NASA’s Earth-Sun System Division and archived and distributed by the Goddard Earth Sciences (GES) Data and Information Services Center (DISC). The ECMWF reanalysis was obtained from the Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory. This research was supported by National Science Foundation Grants ATM-0732255 and AGS-1321568.

REFERENCES

REFERENCES
Austin
,
P. M.
, and
S. G.
Geotis
,
1979
:
Raindrop sizes and related parameters for GATE
.
J. Appl. Meteor.
,
18
,
569
575
, doi:.
Awaka
,
J.
,
T.
Iguchi
,
H.
Kumagai
, and
K.
Okamoto
,
1997
: Rain type classification algorithm for TRMM precipitation radar. Proc. IGARSS ‘97, Vol. 4, Singapore, IEEE,
1633
1635
, doi:.
Barnes
,
G. M.
, and
K.
Sieckman
,
1984
:
The environment of fast- and slow-moving tropical mesoscale convective cloud lines
.
Mon. Wea. Rev.
,
112
,
1782
1794
, doi:.
Barnes
,
H. C.
, and
R. A.
Houze
,
2013
:
The precipitating cloud population of the Madden–Julian oscillation over the Indian and west Pacific Oceans
.
J. Geophys. Res. Atmos.
,
118
,
6996
7023
, doi:.
Berg
,
W.
,
T.
L’Ecuyer
, and
C.
Kummerow
,
2006
:
Rainfall climate regimes: The relationship of regional TRMM rainfall biases to the environment
.
J. Appl. Meteor.
,
45
,
434
454
, doi:.
Berry
,
G. J.
, and
C. D.
Thorncroft
,
2005
: Case study of an intense African easterly wave. Mon. Wea. Rev.,133, 752–766, doi:.
Berry
,
G. J.
, and
C. D.
Thorncroft
,
2012
: African easterly wave dynamics in a mesoscale numerical model: The upscale role of convection. J. Atmos. Sci.,69, 1267–1283, doi:.
Braun
,
S. A.
,
2010
:
Reevaluating the role of the Saharan air layer in Atlantic tropical cyclogenesis and evolution
.
Mon. Wea. Rev.
,
138
,
2007
2037
, doi:.
Bretherton
,
C. S.
,
M. E.
Peters
, and
L. E.
Back
,
2004
:
Relationships between water vapor path and precipitation over the tropical oceans
.
J. Climate
,
17
,
1517
1528
, doi:.
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
, doi:.
Carlson
,
T. N.
, and
J. M.
Prospero
,
1972
: The large-scale movement of Saharan air outbreaks over the northern equatorial Atlantic. J. Appl. Meteor.,11, 283–297, doi:.
Cecil
,
D. J.
,
E. J.
Zipser
, and
S. W.
Nesbitt
,
2002
:
Reflectivity, ice scattering, and lightning characteristics of hurricane eyewalls and rainbands. Part I: Quantitative description
.
Mon. Wea. Rev.
,
130
,
769
784
, doi:.
Cunning
,
J. B.
, and
R. I.
Sax
,
1977
:
A Z–R relationship for the GATE B-scale array
.
Mon. Wea. Rev.
,
105
,
1330
1336
, doi:.
Dee
,
D. P.
, and Coauthors
,
2011
:
The ERA-Interim reanalysis: Configuration and performance of the data assimilation system
.
Quart. J. Roy. Meteor. Soc.
,
137
,
553
597
, doi:.
Del Genio
,
A. D.
,
J.
Wu
, and
Y.
Chen
,
2012
:
Characteristics of mesoscale organization in WRF simulations of convection during TWP-ICE
.
J. Climate
,
25
,
5666
5688
, doi:.
Fortune
,
M.
,
1980
:
Properties of African squall lines inferred from time-lapse satellite imagery
.
Mon. Wea. Rev.
,
108
,
153
168
, doi:.
Fuentes
,
J. D.
,
B.
Geerts
,
T.
Dejene
,
P.
D’Odorico
, and
E.
Joseph
,
2008
:
Vertical attributes of precipitation systems in West Africa and adjacent Atlantic Ocean
.
Theor. Appl. Climatol.
,
92
,
181
193
, doi:.
Funk
,
A.
,
C.
Schumacher
, and
J.
Awaka
,
2013
:
Analysis of rain classifications over the tropics by version 7 of the TRMM PR 2A23 algorithm
.
J. Meteor. Soc. Japan
,
91
,
257
271
, doi:.
Geerts
,
B.
, and
T.
Dejene
,
2005
:
Regional and diurnal variability of the vertical structure of precipitation systems in Africa based on spaceborne radar data
.
J. Climate
,
18
,
893
916
, doi:.
Gosset
,
M.
,
E.-P.
Zahiri
, and
S.
Moumouni
,
2010
:
Rain drop size distribution variability and impact on X-band polarimetric radar retrieval: Results from the AMMA campaign in Benin
.
Quart. J. Roy. Meteor. Soc.
,
136
,
243
256
, doi:.
Guy
,
N.
, and
S. A.
Rutledge
,
2012
:
Regional comparison of West African convective characteristics: A TRMM-based climatology
.
Quart. J. Roy. Meteor. Soc.
, 138, 1179–1195, doi:.
Hagos
,
S.
, and
C.
Zhang
,
2010
:
Diabatic heating, divergent circulation and moisture transport in the African monsoon system
.
Quart. J. Roy. Meteor. Soc.
,
136
,
411
425
, doi:.
Hamilton
,
R. A.
,
J. W.
Archbold
, and
C. K. M.
Douglas
,
1945
:
Meteorology of Nigeria and adjacent territory
.
Quart. J. Roy. Meteor. Soc.
,
71
,
231
264
, doi:.
Hirose
,
M.
, and
K.
Nakamura
,
2004
:
Spatiotemporal variation of the vertical gradient of rainfall rate observed by the TRMM Precipitation Radar
.
J. Climate
,
17
,
3378
3397
, doi:.
Hirose
,
M.
,
R.
Oki
,
S.
Shimizu
,
M.
Kachi
, and
T.
Higashiuwatoko
,
2008
:
Finescale diurnal rainfall statistics refined from eight years of TRMM PR data
.
J. Appl. Meteor. Climatol.
,
47
,
544
561
, doi:.
Hirose
,
M.
,
R.
Oki
,
D. A.
Short
, and
K.
Nakamura
,
2009
:
Regional characteristics of scale-based precipitation systems from ten years of TRMM PR data
. J. Meteor. Soc. Japan,
87A
,
353
368
, doi:.
Hodges
,
K. I.
, and
C. D.
Thorncroft
,
1997
:
Distribution and statistics of African mesoscale convective weather systems based on the ISCCP Meteosat imagery
.
Mon. Wea. Rev.
,
125
,
2821
2837
, doi:.
Holloway
,
C. E.
, and
J. D.
Neelin
,
2009
:
Moisture vertical structure, column water vapor, and tropical deep convection
.
J. Atmos. Sci.
,
66
,
1665
1683
, doi:.
Houze
,
R. A.
,
1989
:
Observed structure of mesoscale convective systems and implications for large-scale heating
.
Quart. J. Roy. Meteor. Soc.
,
115
,
425
461
, doi:.
Houze
,
R. A.
,
1997
:
Stratiform precipitation in regions of convection: A meteorological paradox?
Bull. Amer. Meteor. Soc.
,
78
,
2179
2196
, doi:.
Houze
,
R. A.
,
2004
:
Mesoscale convective systems
.
Rev. Geophys.
,
42
,
RG4003
, doi:.
Huffman
,
G. J.
, and Coauthors
,
2007
: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor.,8, 38–55, doi:.
Iguchi
,
T.
,
T.
Kozu
,
R.
Meneghini
,
J.
Awaka
, and
K.
Okamoto
,
2000
: Rain-profiling algorithm for the TRMM Precipitation Radar. J. Appl. Meteor.,39, 2038–2052, doi:.
Iguchi
,
T.
,
T.
Kozu
,
J.
Kwiatowski
,
R.
Meneghini
,
J.
Awaka
, and
K.
Okamoto
,
2009
:
Uncertainties in the rain profiling algorithm for the TRMM Precipitation Radar
.
J. Meteor. Soc. Japan
,
87A
,
1
30
, doi:.
Iida
,
Y.
,
T.
Kubota
,
T.
Iguchi
, and
R.
Oki
,
2010
:
Evaluating sampling error in TRMM/PR rainfall products by the bootstrap method: Estimation of the sampling error and its application to a trend analysis
.
J. Geophys. Res.
,
115
, D22119, doi:.
Janiga
,
M. A.
, and
C. D.
Thorncroft
,
2013
:
Regional differences in the kinematic and thermodynamic structure of African easterly waves
.
Quart. J. Roy. Meteor. Soc.
,
139
,
1598
1614
, doi:.
Kirstetter
,
P.-E.
,
Y.
Hong
,
J. J.
Gourley
,
M.
Schwaller
,
W.
Petersen
, and
J.
Zhang
,
2013
:
Comparison of TRMM 2A25 products, version 6 and version 7, with NOAA/NSSL ground radar–based National Mosaic QPE
.
J. Hydrometeor.
,
14
,
661
669
, doi:.
Kozu
,
T.
,
T.
Iguchi
,
T.
Kubota
,
N.
Yoshida
,
S.
Seto
,
J.
Kwiatowski
, and
Y. N.
Takayabu
,
2009
:
Feasibility of raindrop size distribution parameter estimation with TRMM Precipitation Radar
.
J. Meteor. Soc. Japan
,
87A
,
53
66
, doi:.
Kummerow
,
C.
,
W.
Barnes
,
T.
Kozu
,
J.
Shiue
, and
J.
Simpson
,
1998
:
The Tropical Rainfall Measuring Mission (TRMM) sensor package
.
J. Atmos. Oceanic Technol.
,
15
,
809
817
, doi:.
Laing
,
A. G.
,
J. M.
Fritsch
, and
A. J.
Negri
,
1999
:
Contribution of mesoscale convective complexes to rainfall in Sahelian Africa: Estimates from geostationary infrared and passive microwave data
.
J. Appl. Meteor.
,
38
,
957
964
, doi:.
Laing
,
A. G.
,
R.
Carbone
,
V.
Levizzani
, and
J.
Tuttle
,
2008
:
The propagation and diurnal cycles of deep convection in northern tropical Africa
.
Quart. J. Roy. Meteor. Soc.
,
134
,
93
109
, doi:.
Liu
,
C.
,
2011
:
Rainfall contributions from precipitation systems with different sizes, convective intensities, and durations over the tropics and subtropics
.
J. Hydrometeor.
,
12
,
394
412
, doi:.
Liu
,
C.
,
E. J.
Zipser
, and
S. W.
Nesbitt
,
2007
:
Global distribution of tropical deep convection: Different perspectives from TRMM infrared and radar data
.
J. Climate
,
20
,
489
503
, doi:.
Liu
,
C.
,
E. J.
Zipser
,
D. J.
Cecil
,
S. W.
Nesbitt
, and
S.
Sherwood
,
2008
:
A cloud and precipitation feature database from nine years of TRMM observations
.
J. Appl. Meteor.
,
47
,
2712
2728
, doi:.
Machado
,
L. A. T.
,
M.
Desbois
, and
J.-P.
Duvel
,
1992
:
Structural characteristics of deep convective systems over tropical Africa and the Atlantic Ocean
.
Mon. Wea. Rev.
,
120
,
392
406
, doi:.
Maddox
,
R. A.
,
1980
:
Mesoscale convective complexes
.
Bull. Amer. Meteor. Soc.
,
61
,
1374
1387
, doi:.
Mathon
,
V.
, and
H.
Laurent
,
2001
:
Life cycle of Sahelian mesoscale convective cloud systems
.
Quart. J. Roy. Meteor. Soc.
,
127
,
377
406
, doi:.
Mathon
,
V.
,
H.
Laurent
, and
T.
Lebel
,
2002
:
Mesoscale convective system rainfall in the Sahel
.
J. Appl. Meteor.
,
41
,
1081
1092
, doi:.
Mohr
,
K. I.
, and
C. D.
Thorncroft
,
2006
:
Intense convective systems in West Africa and their relationship to the African easterly jet
.
Quart. J. Roy. Meteor. Soc.
,
132
,
163
176
, doi:.
Munchak
,
S. J.
,
C. D.
Kummerow
, and
G.
Elsaesser
,
2012
:
Relationships between the raindrop size distribution and properties of the environment and clouds inferred from TRMM
.
J. Climate
,
25
,
2963
2978
, doi:.
Nesbitt
,
S. W.
, and
E. J.
Zipser
,
2003
: The diurnal cycle of rainfall and convective intensity according to three years of TRMM measurements. J. Climate,16, 1456–1475.
Nesbitt
,
S. W.
,
E. J.
Zipser
, and
D. J.
Cecil
,
2000
:
A census of precipitation features in the tropics using TRMM: Radar, ice scattering, and lightning observations
.
J. Climate
,
13
,
4087
4106
, doi:.
Nguyen
,
H.
, and
J.-P.
Duvel
,
2008
: Synoptic wave perturbations and convective systems over equatorial Africa. J. Climate,21, 6372–6388, doi:.
Nicholls
,
S. D.
, and
K. I.
Mohr
,
2010
:
An analysis of the environments of intense convective systems in West Africa in 2003
.
Mon. Wea. Rev.
,
138
,
3721
3739
, doi:.
Parker
,
D. J.
, and Coauthors
,
2005a
:
The diurnal cycle of the West African monsoon circulation
.
Quart. J. Roy. Meteor. Soc.
,
131
,
2839
2860
, doi:.
Parker
,
D. J.
,
C. D.
Thorncroft
,
R. R.
Burton
, and
A.
Diongue-Niang
,
2005b
:
Analysis of the African easterly jet, using aircraft observations from the JET2000 experiment
.
Quart. J. Roy. Meteor. Soc.
,
131
,
1461
1482
, doi:.
Pearson
,
K. J.
,
G. M. S.
Lister
,
C. E.
Birch
,
R. P.
Allan
,
R. J.
Gogan
, and
S. J.
Woolnough
,
2013
: Modelling the diurnal cycle of tropical convection across the ‘grey zone.’ Quart. J. Roy. Meteor. Soc., 140, 491–499, doi:.
Rasmussen
,
K. L.
,
S. L.
Choi
,
M. D.
Zuluaga
, and
R. A.
Houze
,
2013
:
TRMM precipitation bias in extreme storms in South America
.
Geophys. Res. Lett.
,
40
,
3457
3461
, doi:.
Rickenbach
,
T.
,
R.
Nieto Ferreira
,
N.
Guy
, and
E.
Williams
,
2009
:
Radar-observed squall line propagation and the diurnal cycle of convection in Niamey, Niger, during the 2006 African Monsoon and Multidisciplinary Analysis intensive observing period
.
J. Geophys. Res.
,
114
, D03107, doi:.
Romatschke
,
U.
, and
R. A.
Houze
,
2011a
:
Characteristics of precipitating convective systems in the premonsoon season of South Asia
.
J. Hydrometeor.
,
12
,
157
180
, doi:.
Romatschke
,
U.
, and
R. A.
Houze
,
2011b
:
Characteristics of precipitating convective systems in the South Asian monsoon
.
J. Hydrometeor.
,
12
,
3
26
, doi:.
Rotunno
,
R.
,
J. B.
Klemp
, and
M. L.
Weisman
,
1988
:
A theory for strong, long-lived squall lines
.
J. Atmos. Sci.
,
45
,
463
485
, doi:.
Sauvageot
,
H.
, and
J.-P.
Lacaux
,
1995
: The shape of averaged drop size distributions. J. Atmos. Sci.,52, 1070–1083, doi:.
Schumacher
,
C.
, and
R. A.
Houze
,
2003a
:
The TRMM Precipitation Radar’s view of shallow, isolated rain
.
J. Appl. Meteor.
,
42
,
1519
1524
, doi:.
Schumacher
,
C.
, and
R. A.
Houze
,
2003b
:
Stratiform rain in the tropics as seen by the TRMM Precipitation Radar
.
J. Climate
,
16
,
1739
1756
, doi:.
Schumacher
,
C.
, and
R. A.
Houze
,
2006
:
Stratiform precipitation production over sub-Saharan Africa and the tropical east Atlantic as observed by TRMM
.
Quart. J. Roy. Meteor. Soc.
,
132
,
2235
2255
, doi:.
Schumacher
,
C.
,
M. H.
Zhang
, and
P. E.
Ciesielski
,
2007
:
Heating structures of the TRMM field campaigns
.
J. Atmos. Sci.
,
64
,
2593
2610
, doi:.
Shimizu
,
S.
,
R.
Oki
,
T.
Tagawa
,
T.
Iguchi
, and
M.
Hirose
,
2009
:
Evaluation of the effects of the orbit boost of the TRMM satellite on PR rain estimates
.
J. Meteor. Soc. Japan
,
87A
,
83
92
, doi:.
Short
,
D. A.
, and
K.
Nakamura
,
2000
:
TRMM radar observations of shallow precipitation over the tropical oceans
.
J. Climate
,
13
,
4107
4124
, doi:.
Tokay
,
A.
, and
D. A.
Short
,
1996
: Evidence from tropical raindrop spectra of the origin of rain from stratiform versus convective clouds. J. Appl. Meteor.,35, 355–371, doi:.
TRMM Precipitation Radar Team
,
2011
: Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar Algorithm—Instruction manual for version 7. [Available online at http://www.eorc.jaxa.jp/TRMM/documents/PR_algorithm_product_information/pr_manual/PR_Instruction_Manual_V7_L1.pdf.]
Uijlenhoet
,
R.
,
M.
Steiner
, and
J. A.
Smith
,
2003
:
Variability of raindrop size distributions in a squall line and implications for radar rainfall estimation
.
J. Hydrometeor.
,
4
,
43
61
, doi:.
Ventrice
,
M. J.
,
C. D.
Thorncroft
, and
M. A.
Janiga
,
2012
:
Atlantic tropical cyclogenesis: A three-way interaction between an African easterly wave, diurnally varying convection, and a convectively coupled atmospheric Kelvin wave
.
Mon. Wea. Rev.
,
140
,
1108
1124
, doi:.
Weisman
,
M. L.
, and
R.
Rotunno
,
2004
:
“A theory for strong long-lived squall lines” revisited
.
J. Atmos. Sci.
,
61
,
361
382
, doi:.
Yamamoto
,
M. K.
,
F. A.
Furuzawa
,
A.
Higuchi
, and
K.
Nakamura
,
2008
:
Comparison of diurnal variations in precipitation systems observed by TRMM PR, TMI, and VIRS
.
J. Climate
,
21
,
4011
4028
, doi:.
Yuter
,
S. E.
, and
R. A.
Houze
,
1998
:
The natural variability of precipitating clouds over the western Pacific warm pool
.
Quart. J. Roy. Meteor. Soc.
,
124
,
53
99
, doi:.
Zhou
,
Y.
,
W. K. M.
Lau
, and
C.
Liu
,
2013
:
Rain characteristics and large-scale environments of precipitation objects with extreme rain volumes from TRMM observations
.
J. Geophys. Res. Atmos.
,
118
,
9673
9689
, doi:.
Zipser
,
E. J.
,
C.
Liu
,
D. J.
Cecil
,
S. W.
Nesbitt
, and
D. P.
Yorty
,
2006
:
Where are the most intense thunderstorms on Earth?
Bull. Amer. Meteor. Soc.
,
87
,
1057
1071
, doi:.