Abstract

Seasonal and spatial variation of the vertical gradient of rainfall rate was investigated using global precipitation data observed by the Precipitation Radar (PR) on the Tropical Rainfall Measuring Mission (TRMM) satellite. The vertical gradient was rendered by features of downward decreasing (DD) or downward increasing (DI) rainfall rate in the lower part of the profile. The DD profiles dominated tropical interior landmasses such as Africa and the Brazilian Plateau in summer. The DI profiles were observed over land in winter and over ocean except for regions with very little rainfall. In addition, DI profiles appeared during the height of the wet season even over the tropical landmasses, such as the mature monsoon period over inland India and over the Amazon River basin. Individual precipitation systems were also investigated in terms of their areally averaged DD and DI characteristics mainly over India. Deep (shallow) profiles tended to be DD (DI) for all seasons except the premonsoon season. As the rain area increased, the vertical gradient of rainfall rate decreased (DD tendency). Embedded in the dominant DD signature for deep storms, deep but significant DI profiles were observed in every month. They characterized the precipitation in the premonsoon season. More than half of the mesoscale/ synoptic-scale systems (rain areas >104 km2) having the significant DD or DI regions had both of them as part of their slant cores. The vertical gradients for these systems had a similar trend for both their stratiform and convective parts. During the mature period of the southwest monsoon, the number of small systems that were DI and widespread systems with moderate vertical gradient increased.

1. Introduction

The study and continuous measurement of precipitation systems is indispensable for understanding the climate system and the global water cycle. Distinguishing the local variation from the general characteristics of precipitation systems would be a good step toward understanding the dynamical and thermodynamical mechanisms of the systems and microphysical processes therein (e.g., Houze 1981; Zipser and Lutz 1994). These “precipitation regimes” include the easterly and westerly flow regimes in the Amazon area, onset and mature phases of the monsoon, and the well-known El Niño and La Niña episodes (e.g., Petersen and Rutledge 2001; Rickenbach et al. 2002).

During recent years, rain-type classification and rainfall rate estimates have been much more sophisticated as a result of the first spaceborne radar, the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) (Awaka et al. 1997; Iguchi et al. 2000). The active measurements have provided homogeneous three-dimensional datasets of various precipitation systems. Several kinds of precipitation climatology have been reported using the TRMM PR data. Liu and Fu (2001) and Fu and Liu (2001) investigated the principal modes and variability of rainfall-rate profiles over the Tropics and inferred the microphysics therein. Takayabu (2002) prescribed individual rain profiles in a spectral representation. These approaches for dealing with large sets of vertical rainfall-rate structure data made it possible to understand what the statistics or instantaneous snapshots indicate. Composite analyses of vertical and horizontal patterns are a good way to understand the impact of individual systems on precipitation climatology (e.g., Stano et al. 2002). To classify precipitation systems, storm height, near-surface rainfall rate, and the existence of a bright band have been the main characteristics used to index the rain type. Precipitation-system structure can also be identified by the stratiform rain fraction, bright-band height, and rainfall at the bright band, in addition to the aforementioned factors (Shige et al. 2004; Schumacher and Houze 2003a; Tao et al. 1993). Hirose and Nakamura (2002; hereafter HN02) investigated the characteristics of vertical gradients of rainfall rates at low levels, which were not well studied due to the lack of data on vertical hydrometeor profiles. They showed that most of the maximum rainfall rates aloft appeared around 4 km, and they divided vertical profiles of rainfall rate into two types: those either with or without rainfall rate peaks aloft, in other words, downward decreasing or increasing rainfall rates at lower levels. They introduced an index of vertical gradient (IVG), to characterize the vertical profile of rainfall rate. The index clarified the difference between tropical interior continental and oceanic precipitation systems as well as a precipitation regime that appeared in the southwest monsoon-affected region over inland India. Maximum hydrometeor contents aloft have been commonly seen in snapshots of radar observations and in model representations (e.g., Rutledge and Hobbs 1984; Rutledge and Houze 1987; Westcott and Kennedy 1989; Williams et al. 1989). However, the interpretation of these peaks shown in profiles averaged over large regions has remained uncertain since statistical analyses of the vertical rainfall-rate profiles over large areas have been difficult. Large-scale and seasonal mapping of this type of index was impossible before the TRMM PR.

The global characteristics of the vertical gradient of rainfall rate have not been studied. A downward increase in mass flux indicates precipitating hydrometeors growth due to coalescence (Liu and Fu 2001). Convection with core height about 2 km could be another possible reason for downward increasing rainfall rates. On the other hand, one interpretation for a downward decrease in water flux is evaporation (e.g., Fu et al. 2003; Adeyewa and Nakamura 2003). In several models, the rate of change of rainwater mixing ratio (qr) due to evaporation of raindrops is derived from the magnitude of rainwater mixing ratio, temperature, pressure, subsaturation, and Reynolds number (e.g., Syono and Takeda 1962; Tripoli and Cotton 1980). Katzfey and Ryan (1997) reported that evaporation in the unsaturated subcloud layer near the surface reduced the rainfall percentage significantly for stratiform rain. Rosenfeld and Mintz (1988) investigated the downward decrease in rainfall under cloud base, around 1–2.2 km, for rain shafts in summer convective cells in a semiarid region. The effects of evaporation strongly depend on precipitation type, which is associated with the vertical velocity, rainfall intensity (Takemi 1999), and the surrounding environment. There are other possible factors with regard to a downward decreasing rainfall rate. Hydrometeors suspended aloft can be influenced by strong updrafts and lateral transport by the horizontal flow. Suspended hydrometeors in the initial stages of precipitation can generate a slight maximum rainfall rate aloft even though it is generally weak. Differences in raindrop fall velocities in strong updrafts and downdrafts or differences in hydrometeors (e.g., graupels) can influence rainfall rate retrieval through ZR relationships (e.g., Atlas et al. 1973; Battan 1976; Dotzek and Beheng 2001; Dotzek and Fehr 2003; Matejka et al. 1980; Rutledge and Hobbs 1984; Salles and Creutin 2003).

Mesoscale convective systems (MCSs) or storms with significant vertical shear show clear vertical tilt or a nondownward flux of hydrometeors (e.g., Klemp and Wilhelmson 1978; McAnelly et al. 1997; Parker and Johnson 2000; Rutledge and Houze 1987). The combined effect of lateral and upward flux on water loading means negative buoyancy may be nonnegligible for average vertical profiles. Precipitation events often occur in complex combinations of stratiform and convective patterns. Some insight could be gained by examining a large data volume showing the dynamic variation of precipitation-system structure. Further possible explanations can be obtained them from observations of individual precipitation systems with various scales in time and space. The global climatology of the vertical gradient of rainfall rate can characterize precipitation-system structure more concretely. This paper describes the seasonal and spatial variation of the vertical gradient of monthly rainfall rate and qualitative analyses of individual rainfall-rate profiles and individual precipitation systems.

The precipitation datasets and method of analysis are described in section 2. The definition and validity of IVG are described in section 3. In section 4, a global map of the index and the characteristics are depicted. Section 5 presents several parameters for individual rain profiles. Section 6 explores the population of individual systems over distinct geographical regions. The findings are summarized and discussed in section 7.

2. Precipitation datasets

In this study, three years (1998–2000) of rainfall rate data from the TRMM PR (2A25 with 250-m vertical resolution) are used to investigate three-dimensional precipitation-system structure. Version 5 of the TRMM PR algorithms was used. Rainfall rate is calculated accounting for the differences in stratiform and convective patterns, ice and water mixture, nonuniform beam filling effects and the pressure-modified terminal velocity in stagnant air (Foote and Toit 1969). A considerable amount of validation has been performed to certify that the estimated near-surface rainfall rate is reliable (e.g., Kummerow et al. 2000). Here, “near surface” means the lowest level free from ground clutter, which is around 500 m above the surface at nadir. However, some problems still remain in validating the three-dimensional structure of precipitation. One such problem is associated with the bright band. Conventional knowledge is not yet good enough to physically validate rainfall rates around the bright band derived from current algorithms. Rainfall profiles show spikes around the bright band and have less accuracy. The second problem is the technical difficulty in producing fully homogeneous three-dimensional datasets. Detection of the bright band needs high vertical resolution. However, at the edge of the PR swath, height smearing is intolerable, being more than 1 km. Data near the scan edge contains factors that deteriorate the vertical profiles such as height smearing, a slanting incident mainlobe, and masked regions at low levels from sidelobes in the nadir.

To mitigate the aforementioned factors, the rainfall rate around the bright band was replaced with linearly interpolated data using data ±750 m from the brightband peak. And secondly, only data with a swath angle within 7° from nadir was used. The number of the available field of views (FOVs) in the cross track was 21 out of a total of 49. The analyzed width being around 86 km out of the available 215-km width of the TRMM PR for each cross-track scan. Awaka (1998) reported that 80% of bright bands are detected correctly within 7° of nadir.

The aforementioned procedure was used throughout the study except in estimating rain area. Rain flag data was used in all angle bins to estimate the size of a precipitation system more accurately. The rain flag is defined when significant rain signal exists within the FOV. The minimum detectable radar reflectivity of the PR is 18 dBZ (∼0.5 mm h−1) where the signal-to-noise (S/N) ratio is nearly 0 dB at storm top. Values around 13 dBZ corresponding to weak rainfall rates are also observable though it depends on the system noise level.

The area of each precipitation system and the fraction of convective area (individual as well as total) were calculated as follows: First, a flag is checked on a pixel where rain is certain to exist. Next, nearby rain pixels are used to identify a common precipitation system using neighboring flags within about 10 km. The neighbors within the 10 km are recognized as being part of the same system as are the pixels adjacent to them. Contiguous systems are determined through iteration. The “rain area” is calculated approximately by multiplying the number of connecting pixels by 15 km2 as the footprint size is about 4.3 km at nadir and increases only by 0.8% 7° off nadir. Areas of individual convective rain are derived from a similar procedure using a “convective” flag instead of a rain flag. The total convective rain area is defined as the summation of all embedded convective pixels in the same precipitation system. The area of large systems will, however, be underestimated due to the TRMM PR swath (Nesbitt et al. 2000).

Also, caution must be taken in interpreting the downward decrease in rainfall rate below the maximum as attenuation is significant and present in TRMM PR data (e.g., Liu and Fu 2001). However, the attenuation correction for reflectivity profiles was validated by a ground-based radar (e.g., Bolen and Chandrasekar 2000) and is assumed to be well corrected in this study.

3. Index of vertical gradient

In this paragraph, the terminologies used by HN02 are reviewed. IVG is used to express the vertical gradient of rainfall rate at low levels. It is simply defined as the rainfall rate at 2 km minus the rate at 3.5 km divided by the interval of 1.5 km. The units are mm h−1 km−1. Negative values indicate a downward decreasing pattern characterized by a maximum rainfall rate aloft and positive ones a downward increasing pattern. Hereafter, the terms DD and DI refer to a downward decrease and downward increase in rainfall rate at low levels, respectively. The 2- and 3.5-km altitudes were chosen based on the number of samples and levels around and beneath the peak concentrations. Data were excluded when the orographic features affected the sample number by more than 10% in a grid at 2 km. In studying the IVG over different regions, “inland” is selected in such a way that the orographic effects are minimal. The “coastal” (inland) region is defined as having land within (not within) 1° of an oceanic grid. In the coastal upwelling region, mountains such as the Western Ghats obstruct the large moist inflow and generate copious orographic DI rain. Storm height indicates the top altitude of the highest three consecutive range bins with significant echoes (>17 dBZ). Though 2A23 data contain storm height, only 2A25 data were used to simplify the analyses. The term storm height is not used in the conventional sense of the height of the cloud or convection but rather as the maximum height of precipitation echo observed by the TRMM PR.

Investigation of the large number of vertical profiles revealed that DD contained most of the significant modal structures of rainfall rate. Figure 1 shows monthly accumulated rainfall-rate peak and storm-height histograms over inland India (5°–25°N, 65°–90°E). A rainfall rate peak was simply defined as the center point between three consecutive range bins (±750 m) of positive and negative vertical gradient of rainfall rate. Rainfall rate peaks are concentrated around 4 km for stratiform rain (Fig. 1a). However, the peak at 4 km in the stratiform histogram does not indicate a rainfall-rate maximum in this dataset. The peak generally appears at 750 m below the brightband peak since it tends to be an inflection point induced by the interpolated data after removing the original data near the bright band. The concentration of rainfall rate peaks at 4 km shows that the 3.5-km level is not affected by the profile modification for the bright band.

Fig. 1.

Peak height distribution over inland India from Apr to Oct. Thick lines are rainfall-rate peak histograms, broken lines are peaks recognized as DD, and thin lines are storm height histograms

Fig. 1.

Peak height distribution over inland India from Apr to Oct. Thick lines are rainfall-rate peak histograms, broken lines are peaks recognized as DD, and thin lines are storm height histograms

The histogram for convective rain (Fig. 1b) shows large variations with height but with a constant peak magnitude. The peaks in frequency levels varied several hundred meters in phase with the storm heights; however, the peaks disappeared in winter (not shown in Fig. 1b). It seems reasonable to use a constant level of 3.5 km as it is around or beneath the central level of peaks. This study did not consider the low-level peaks that appeared around 2 km, associated with shallow convection, since robust statistics are needed with sufficient samples for the most significant peaks. The fact that IVG in general does not cover the shallow modal structure is not a serious problem since the intensity of shallow modal rainfall and its impact on the IVG are small.

As indicated in Table 1, about 15% of all profiles had peaks as defined above, and most of the peaks were seen above 3 km (87% for stratiform and 77% for convective rain). The DD trend seems to be significant. More than about 70% of profiles with the peak above 3 km exhibited DD signature. Most of the remaining profiles had a DI signature with peaks appearing at a high (∼5 km) altitude (shown in Fig. 1). For the present study, such profiles were not considered. For example, it was found in the GATE experiment that preexisting overhang and thick anvil could reach 30 dBZ (e.g., Szoke and Zipser 1986). Most of the peaks at marginally low levels (3 ∼ 3.5 km) were DD in nature. The sensitivity of the upper level in the definition of the IVG was checked by mapping of the IVG over the globe and comparing against peaks in the histogram profiles for various upper levels (not shown). The DI profiles with marginally low-level peaks were embedded sporadically within large rainy regions such as the western Pacific and the highly wet Amazon throughout the year. The adjustment of the upper level for the marginally low-level peaks has less impact on final statistics. Most of the gradients with an upper threshold of 4 km increased positively and lessened the number of DD cases. Thus, a level of 3.5 km was retained in order not to deteriorate the detection of peak signals. The IVG is reasonable and proper as an index for a concise but significant characteristic of the vertical gradient of rainfall profiles.

Table 1.

Ratio of the number of profiles having peaks to all profiles (peak ratio), ratio of the number of profiles having peaks higher than 3 km to those having peaks (high peak), and ratio of the number of peak profiles characterized as DD to those having peaks higher than 3 km (DD peak). The weighted average is shown in the last line. The units are in percentages

Ratio of the number of profiles having peaks to all profiles (peak ratio), ratio of the number of profiles having peaks higher than 3 km to those having peaks (high peak), and ratio of the number of peak profiles characterized as DD to those having peaks higher than 3 km (DD peak). The weighted average is shown in the last line. The units are in percentages
Ratio of the number of profiles having peaks to all profiles (peak ratio), ratio of the number of profiles having peaks higher than 3 km to those having peaks (high peak), and ratio of the number of peak profiles characterized as DD to those having peaks higher than 3 km (DD peak). The weighted average is shown in the last line. The units are in percentages

4. Seasonal variation of IVG over the globe

a. Spatiotemporal distribution of IVG derived from monthly mean profiles

The previous investigation on the seasonal distribution of IVG over and around India is now extended beyond the global Tropics to 37°S and N. In this section, the climatological characteristics of IVG are described. Figure 2 shows the variation of IVG derived from monthly averaged rainfall-rate profiles with 1° × 1° spatial resolution. In general, a significant amount of DD rainfall rates appear over interior landmasses in the summer season. Downward increase rates occur in the winter, in midlatitudes, and over oceans except in regions with very little rainfall. Over Africa, DD rates dominated throughout the year and are in phase with the rainfall (migrating with the rain between the Northern and Southern Hemisphere). The high frequency of DD rates indicates that tropical interior convective systems have a tendency to be DD in summer. Over India, the northward migration of DD characteristics during the onset of the monsoon and southward withdrawal during the retrogressing periods was observed as reported by HN02. Furthermore, DD rates seem to appear frequently between “dry” and “highly wet” regions. This introduces a hypothesis to be considered later. Downward decrease rates appear frequently over the Amazon River basin corresponding to the southward migration (August–October) of rainy regions. This is prior to or around the transition period (e.g., September–November) when the convection shows much more continental-like characteristics exhibiting a greater frequency of deep radar reflectivity cores and more frequent lightning (e.g., Petersen and Rutledge 2001; Rickenbach et al. 2002). The IVG distribution over this wet region consists of oceanic and continental elements reflecting westerly and easterly flow regimes with time scales of several days. On the other hand, DD rates widely appear over the Brazilian Plateau. Other regions with significant DD areas are: northern Australia, the Arabian Peninsula, southern plains of the United States, and the Indochina Peninsula. Downward decrease profiles over the ocean with very little rainfall (e.g., the eastern Atlantic) had a moderate IVG signature of −0.1 to 0 mm h−1 km−1.

Fig. 2.

Monthly IVG map over the global Tropics. The spatial resolution is 1° × 1°. Dark (light) stipple indicates significant DI: 0.6 < IVG (moderate DI: 0.3 < IVG < 0.6). Dark (light) shading represents significant DD: IVG < −0.3 (moderate DD: −0.3 < IVG < 0).> >

Fig. 2.

Monthly IVG map over the global Tropics. The spatial resolution is 1° × 1°. Dark (light) stipple indicates significant DI: 0.6 < IVG (moderate DI: 0.3 < IVG < 0.6). Dark (light) shading represents significant DD: IVG < −0.3 (moderate DD: −0.3 < IVG < 0).> >

The structural differences between significant DD regions over land were examined. The focus being on the latitudinal change of monthly rainfall, rainfall rate, and the frequency of occurrence of DD profiles. The following regions were highlighted: the African continent having a typical continental-like pattern, India characterized by abundant monsoon rainfall, and South America including the wet Amazon region and the wet or dry Brazilian Plateau. Figure 3 shows the latitudinal and temporal variation of monthly rainfall, strong rainfall rate (>4 mm h−1), and DD occurrence over the regions described above. A significant fluctuation in rainfall north and south of the equator with an amplitude from 10° to 15° in latitude is observed over Africa (Fig. 3a). The significant DD region does not overlap the maximum rainfall region. The horseshoelike pattern in Fig. 3b is almost identical to the one in HN02 but only for convective rain. Downward decrease characteristics are slightly reduced in the highly wet region and, instead, coincide with the zonally averaged rainfall contours of around 100 mm month−1. The strongest zonally averaged rainfall rates were observed over India in the premonsoon season and in the northern part of India and Pakistan in mid to late summer. Most of the significant DD pixels correspond to relatively strong rainfall rates (>3 mm h−1, see Fig. 9b in HN02). There are similarities between the rainfall over Africa and India. The highest DD occurrence is located on the fringe of the large rainfall amounts. The amplitude of the sine-curve pattern of DD occurrence is larger than that for the rainfall amount. The occurrence of DD profiles over India is synchronized with the pattern over Africa. There are, however, several significant differences between them. The most significant one being that the pattern over India does not have a maximum frequency of occurrence of DD profiles around 15°N in the middle of summer that overlaps with the mature phase of the monsoon. The rainfall over Africa can be classified as “continental” considering the sine-curve pattern of DD characteristics that correspond to the maximum heating by solar insolation. In addition, the variation of moisture inflow from the ocean can be crucial in changing the structure of tropical interior convection over India.

Fig. 3.

Latitudinal time section illustrating the relationship between monthly total rainfall (100 and 200 mm month−1, thin and thick lines), strong rainfall rate (4 mm h−1, dotted regions) and strong DD occurrence of IVG < −0.5, gray shading (10%–20%, 20%–30%, and more than 30%) over (a) inland Africa (20°W–50°E), (b) India (60°–90°E), and (c) South America (80°–40°W)

Fig. 3.

Latitudinal time section illustrating the relationship between monthly total rainfall (100 and 200 mm month−1, thin and thick lines), strong rainfall rate (4 mm h−1, dotted regions) and strong DD occurrence of IVG < −0.5, gray shading (10%–20%, 20%–30%, and more than 30%) over (a) inland Africa (20°W–50°E), (b) India (60°–90°E), and (c) South America (80°–40°W)

Rainfall over South America, mainly Brazil (Fig. 3c), also has such similarities with the other rainy regions. One of the main characteristics is the northward migration of extensive rainfall regions with DI profiles around the Amazon River basin. These oceanic characteristics are consistent with earlier results (e.g., Garstang et al. 1994; Schumacher and Houze 2003a). Downward decrease profiles appear widely before and around the transition period of the South American rainy season as mentioned above. The region from 10° to 20°S (mainly over the Brazilian Plateau) was generally characterized as DD. The strongest rainfall rates, on average, were found particularly around 30°S corresponding to the Pampa Humeda.

b. Variation of IVG in differing moisture environments

The variation of IVG with atmospheric conditions is examined in this section. Monthly and latitudinal variation of the frequency of occurrence of DD profiles over India showed a one-to-one correspondence with storm height (HN02), the mean number of thunderstorm days, and the mean maximum surface wet-bulb temperature (Manohar et al. 1999). In addition to the systematic changes in the equatorial convective zones, dry and wet regimes are also of interest. Trenberth (1999) showed relationships between wind, precipitation, and atmospheric water vapor with a recycling ratio and reported that precipitation associated with transported water vapor dominates in the monsoon onset period, while precipitation associated with evaporated and recycled water vapor dominates in the late or postmonsoon season. Several climatological elements such as moisture inflow, surface soil moisture, and evapotranspiration from vegetation do not correspond with DD occurrences that show relatively symmetric succession between the onset and the retrogressing periods, while variations in the surface boundary may affect significantly subsequent convective systems.

To explain the change in precipitation-system structure associated with wet and dry environmental conditions, knowledge of the amount of atmospheric moisture is helpful. Figure 4 shows the effect of total water vapor on IVG over India, northern Africa, the Amazon River basin, and the Brazilian Plateau. The area of these locations is approximately 1.1 × 106, 8.8 × 106, 1.3 × 106, and 3.5 × 106 km2, respectively. The inland regions were selected as either DD-significant regions or highly wet regions affected by moist maritime air. The abscissa shows precipitable water derived from National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data (Kalnay et al. 1996; Trenberth and Guillemot 1998) and the ordinate average rainfall rate derived from the TRMM PR with a 2.5° resolution.

Fig. 4.

Regional difference relationships between DD and DI, total column water vapor, and rainfall rate at 2.5° spatial resolution. Circles and dark points represent DD and DI. Shown are central and southern India (10°–25°N, 65°–90°E), northern Africa (10°–25°N, 20°W–50°E), the Amazon River basin (10°S–5°N, 80°–50°W, altitudes of 0–100 m), and the Brazilian Plateau (30°–5°S, 60°–35°W, altitudes ranging from 200 to 1000 m)

Fig. 4.

Regional difference relationships between DD and DI, total column water vapor, and rainfall rate at 2.5° spatial resolution. Circles and dark points represent DD and DI. Shown are central and southern India (10°–25°N, 65°–90°E), northern Africa (10°–25°N, 20°W–50°E), the Amazon River basin (10°S–5°N, 80°–50°W, altitudes of 0–100 m), and the Brazilian Plateau (30°–5°S, 60°–35°W, altitudes ranging from 200 to 1000 m)

Comparing India (5°–25°N, 65°–90°E) and northern Africa (10°–25°N, 20°W–50°E) shows the difference in water vapor amount and the affected precipitation-system structure in IVG. A DD signature dominates in summer over these regions as expected. In other words, DD profiles appear from April or May to October when precipitable water is below 50 mm. On the other hand, DI profiles appear during winter when precipitable water is relatively low. In midsummer when atmospheric water vapor was around or more than 50 mm, the rainfall rate weakened and had DI characteristics over India. Since water vapor amount is a maximum in August, the lower atmosphere becomes saturated (e.g., Ninomiya and Kobayashi 1998; Lim and Kim 2002). The relative humidity in mid to late summer is ≥about 80% at 850 hPa (not shown). At the height of the wet season, precipitation (e.g., HN02) and cloud (e.g., Laing and Fritsch 1993b) are at a maximum, reducing the solar insolation at the surface. The abundance of water vapor turns the continental precipitation regime into a quasi-oceanic one. In other words, the monsoon brings highly moist air into inland India. As a result, DD profiles, which typically border the monsoon rainfall regimes, are deemed to be located where there is a more moderate amount of water vapor.

The study area over northern Africa includes two distinct precipitation regimes: the northern desert/steppe region and the tropical savanna region. This is evident in the distribution of precipitation as many large systems are concentrated around 10°–15°N with few systems around and north of 20°N (not shown). Downward decrease profiles with weak rainfall rates and low atmospheric moisture in the summer months are mostly observed north of 15°N. All regions over northern Africa had lower atmospheric moisture than over India and showed a DD tendency during mature summer. The elevation of the inland Amazon River basin is less than 100 m (10°S–5°N, 80°–50°W). The precipitation regime is similar to that over India during the wettest part of the summer. The monsoon/tropical rainforest has abundant water vapor (40–55 mm) resulting in DI profiles from January to June, predominantly in April and May. In contrast, atmospheric moisture over the Brazilian Plateau (30°–5°S, 60°–35°W, elevation of 200–1000 m) is less. The DI signatures associated with abundant atmospheric moisture are less. Most of the region remains DD during the rainy season. Even in winter, the atmospheric moisture and DD fraction is higher than those over India and Africa. These results suggest that the IVG in dry winter, wet summer, and highly wet summer is DI, DD, and moderately DI, respectively.

5. Vertical characteristics of individual rainshafts

Individual rainfall-rate profiles can be characterized by the storm height, the rainfall rate near the surface, and the vertical gradient at low levels. The main area of interest is India where the IVG transitions through a change of dry, wet, and highly wet seasons. The rainfall over inland India has been characterized in terms of IVG and rainfall rate (see Fig. 10 in HN02 and Fig. 3b in this paper). Figure 5 shows a schematic diagram of precipitation over inland India in terms of four phases: the winter period (phase A), the hot season prior to the monsoon (phase B) with significant DI profiles (a majority of the IVG greater than 1.5), a DD-salient phase (phase C) surrounding the highly wet monsoon region, and the mature phase of the Indian monsoon (phase D) dominated by moderate DI profiles (IVG between 0 and 1.5). The IVG shows significant variation with time in the region (i.e., south of 25°N). HN02 showed that DD profiles generally represent deep storms and DI profiles shallow storms. Further, it was found that the general relationship between IVG and storm height for the rainy season may not be valid for the premonsoon season when significant DI profiles were present with high storm heights and strong rainfall rates. These variations suggested the necessity for further investigations.

Fig. 5.

Schematic diagram for seasonal and zonal mean precipitation structure over inland India. Abscissa and ordinate are as in Fig. 3b; A, B, C, and D indicate the phase characterized by IVG and rainfall rate

Fig. 5.

Schematic diagram for seasonal and zonal mean precipitation structure over inland India. Abscissa and ordinate are as in Fig. 3b; A, B, C, and D indicate the phase characterized by IVG and rainfall rate

Figure 6 shows the relationship between near-surface rainfall rate and storm height for DD and DI stratiform and convective rain profiles in each phase from A to D as in Fig. 5. IVG is averaged over each grid [50 (rainfall) × 80 (height) intervals] and the number of profiles in the grid are shown as contours and shading, respectively. Crosses (circles) indicate individual profiles with a significant DD (DI) (i.e., IVG > ±30). The stratiform rain patterns show that moderately high storms having moderate DD profiles and low storms having moderate DI profiles generally form the monthly IVG. Below 5 km, DI profiles dominate the winter. As the season progresses, moderately high storms with moderate DD profiles increase and reach a maximum number in summer. Simultaneously, DI profiles associated with low storm heights of a few kilometers appear mostly in weak rainfall rates. Downward increase profiles in low storm heights also existed in the case of convective rain profiles. A striking difference between the two types is the difference in magnitude of the significant IVG. In winter (phase A), most of the precipitation is associated with moderate DI signatures and some moderate DD profiles for midlevel heights. In the premonsoon season (phase B), heavy rainfall (>50 mm h−1) that is mainly DI dominates. At the start of the rainy season (phase C), deep convective storms reaching up to 15 km with significant DD signatures and strong rainfall rates are noticeable. Monthly averaged rain profiles in this phase are DD every year in spite of the existence of some significant DI profiles. Similar features are present in the postmonsoon season. At the height of the wet season (phase D), precipitation associated with a high IVG, strong rainfall rates, and deep storms disappears. This mode seems to be suppressed and changed into a monsoonal rain with moderately low storm heights, weak rainfall rates, and moderate IVG.

Fig. 6.

Relationship between rainfall rate, storm height, and IVG for individual rain profiles over inland India. Rain in Jan, Mar, May, and Aug is depicted as typical of phases A to D in Fig. 5 for stratiform (left) and convective (right) rain, respectively. The shading indicates sample numbers of 1, 5, 10, and 50 over each grid that fall into 50 (rainfall) × 80 (height) intervals in a rainfall rate range from 0.1 to 200 mm and heights between 0 and 20 km. The solid (dashed) lines indicate DD (DI). The contours indicate an IVG of −5, −1, −0.1, 0.1, 1, 5 mm h−1 km−1 with a simple spatial average. Crosses and circles show profiles with intense DD (IVG < −30) and DI (IVG > 30) rates, respectively

Fig. 6.

Relationship between rainfall rate, storm height, and IVG for individual rain profiles over inland India. Rain in Jan, Mar, May, and Aug is depicted as typical of phases A to D in Fig. 5 for stratiform (left) and convective (right) rain, respectively. The shading indicates sample numbers of 1, 5, 10, and 50 over each grid that fall into 50 (rainfall) × 80 (height) intervals in a rainfall rate range from 0.1 to 200 mm and heights between 0 and 20 km. The solid (dashed) lines indicate DD (DI). The contours indicate an IVG of −5, −1, −0.1, 0.1, 1, 5 mm h−1 km−1 with a simple spatial average. Crosses and circles show profiles with intense DD (IVG < −30) and DI (IVG > 30) rates, respectively

As a whole, shallow precipitation noticeably increases in summer. Storms higher than about 5 km have a tendency to be DD with the deepest ones (>15 km) strongly DD (IVG < −10 mm h−1 km−1). Rainfall rate, storm height, and IVG are well correlated in DI profiles; the stronger or deeper the storm, the stronger the DI gradient. Rainfall exceeding 50 mm h−1 was often associated with significant DI profiles (IVG > 30 mm h−1 km−1). It is plausible since an IVG of 30 mm h−1 km−1 is equivalent to a rainfall rate change of 45 mm h−1 between 2 and 3.5 km. However, there were a few exceptional cases with weak rainfall having significant DI profiles. For example, there was a profile with a near-surface rainfall rate of 9.0 mm h−1, a storm height of 12.75 km, and an IVG of 30.5 mm h−1 km−1 (convective rain in May). After careful inspection of the horizontal pattern (not shown), it was located at the edge of a mesosale convective system and was part of an organized three-dimensional structure and will be discussed later. The profile had a maximum of 74.1 mm h−1 at 2.5 km and a sharp decline from 2 to 1 km.

The above procedure was used to study IVG characteristics in other regions as shown in Fig. 7. The first panel (top left) is for convective rain over northern Africa (10°–25°N, 20°W–50°E) in August, which corresponds to the most significant DD region. It shows significant IVG characteristics more clearly than those observed over India and Africa in May (not shown). The number of significant DD profiles was higher for those with storm heights of 15 km and a near-surface rainfall rate of 50 mm h−1. Deep storms with strong rainfall rates had significant DD profiles accompanied by significant DI profiles. The pattern in March over India did not have significant DD profiles but only significant DI ones. Such precipitation regimes shown in phase B, did not appear over northern Africa. This will be examined later by looking at precipitation systems as a whole. The second panel (top right) is for convective rain around the Amazon River basin in September during which DD signatures dominated. The observed pattern is similar to the one over India during phase D with a few significant IVG cases. The fraction of DI is less.

Fig. 7.

As in Fig. 6 but for convective rain over northern Africa (AFR_conv_AUG, 10°–25°N, 20°W–50°E) in Aug, convective rain over the Amazon River basin (AMZ_conv_SEP, 5°S–0°, 70°–50°W) in Sep, convective rain over the South China Sea (SCS_conv_AUG, 10°–20°N, 110°–120°E) in Aug, and oceanic rain over the eastern part of the South Atlantic (DrySA_all_3yr, off Angola, 20°–10°S, 0°–10°E) for all 3 yr

Fig. 7.

As in Fig. 6 but for convective rain over northern Africa (AFR_conv_AUG, 10°–25°N, 20°W–50°E) in Aug, convective rain over the Amazon River basin (AMZ_conv_SEP, 5°S–0°, 70°–50°W) in Sep, convective rain over the South China Sea (SCS_conv_AUG, 10°–20°N, 110°–120°E) in Aug, and oceanic rain over the eastern part of the South Atlantic (DrySA_all_3yr, off Angola, 20°–10°S, 0°–10°E) for all 3 yr

The lower two panels in Fig. 7 show oceanic cases. In general, monthly profiles of oceanic rainfall are characterized by low storm heights, weak rainfall rates, and DI profiles. Rainy regions that contain deep storms have a smaller IVG but are still DI (HN02). The third panel (bottom left) shows convective rain over the South China Sea in August and corresponds to the active regions of rainfall. The fraction of shallow storms that are moderate DI is larger than the storms that are moderate DD, in contrast to the cases over land shown in the top panels. Despite the similarity in the relationship between near-surface rainfall rate and storm height, characteristics of the IVG are in striking contrast for these different precipitation regimes. The last panel (bottom right) is for all precipitation over the eastern part of the South Atlantic Ocean (off the coast of the Republic of Angola) and is typical of a region with very little rainfall (6.8 mm yr−1 near surface) over the ocean. This region is located on the eastern edge of the Atlantic high and over the cold Benguela Current with strong upwelling. Storms were separated into two groups based on altitude. Atmospheric subsidence appeared to be closely associated with the shallow storms around and below 2 km. Short and Nakamura (2000) reported that this is one of the regions where, on average, the shallowest precipitation falls. The small number of DI profiles is an important characteristic of this type of light, oceanic rainfall. The second group appears at and above 5 km and has DD characteristics. Isolated DD profiles constantly appear throughout the year. The bimodal structure or capped height around 2 km and 5 km is widely seen over oceans (Short and Nakamura 2000). However, the rainfall and the fraction of DI profiles are smaller than in other oceanic cases having bimodal structure.

In order to understand the aforementioned profile characteristics, analyses for each system as a whole will be done over landmasses in the following section. Here, the horizontal extent of rain area as defined in section 2 is introduced to understand the oceanic profiles with very little rainfall. There were 7758 systems in the region during the three years, and the vast majority of them (97%) were small with rain areas <100 km2. Sixty-six percent of the 4567 DD profiles and 79% of the 6070 DI profiles were embedded in these isolated small precipitation systems. During the 3-yr period, the largest observed system was 9090 km2 and contained 60 DD profiles and 27 DI profiles. The small number of DD and DI profiles was mainly due to the restriction of the calculation of IVG in a narrower swath and the existence of other types. Over other regions, large systems were dominant. For example, 68% of the DD and 49% of the DI profiles were embedded in a few (3%) but large (>104 km2) systems over the South China Sea. Weak precipitation systems over the ocean with very little rainfall are most likely related to the climatological conditions there. Precipitation over the ocean with very little rainfall can be classified into two types based on IVG, that is, isolated shallow and weak storms with DI profiles and moderately high but isolated and weak storms that are DD. However, the actual number of small weak events does not give an accurate proportion (section 2). The 2A23 version 5 algorithm classified almost all of this rain as “stratiform.” These isolated rainfall systems should be interpreted as shallow, warm convective rain (Heymsfield et al. 2000; Short and Nakamura 2000; Schumacher and Houze 2003b).

6. Horizontal and vertical extent of individual precipitation systems

a. IVG for individual precipitation systems

Numerous rainfall-rate profiles for inland India were examined to assess how DD profiles are embedded in individual precipitation systems and how the characteristics of those profiles alter the monthly average. Several different rain patterns based on IVG emerged: widespread moderately DI stratiform rain, significant DD profiles embedded in deep storms, significant DI profiles in strong convergence zones such as convection at the leading edge of linear mesosale convective systems, moderately DD trailing (or leading) stratiform rain, and relatively small DD convective regions embedded in large stratiform rain areas with moderate DI profiles. Figure 8 shows a precipitation system with very heavy rainfall rates (exceeding 100 mm h−1) and high storm heights (∼15 km) that contains significant DD and DI profiles simultaneously in the slanting convection. The mean flow obtained from NCEP–NCAR reanalysis data was easterly with weak vertical shear at the closest time and location. The trailing stratiform rain around 79.5°E has DD characteristics. The horizontal extent of the precipitation system should be simultaneously investigated with the vertical extent. The objective of this section is to specify the characteristics of IVG more concretely in light of the climatological features depicted in the previous section. Inland India was chosen for the present analysis, mainly to develop the shape characteristics depicted by HN02. Several other factors were also examined: rain type (convective/stratiform), storm height, IVG, near-surface rainfall rate, and the areas of the entire precipitation system, the individual convective elements, and the aggregate convective portion within a system.

Fig. 8.

An example of a snapshot of a large convective system with significant IVG characteristics. Shading shows rainfall rate over central India at 2023 LT 31 Oct 1998. The cross section is along 16.8°N. Two lines indicate altitudes of 2 km and 3.5 km. The lower panel shows the corresponding IVG. Circles and points indicate DD and DI, respectively

Fig. 8.

An example of a snapshot of a large convective system with significant IVG characteristics. Shading shows rainfall rate over central India at 2023 LT 31 Oct 1998. The cross section is along 16.8°N. Two lines indicate altitudes of 2 km and 3.5 km. The lower panel shows the corresponding IVG. Circles and points indicate DD and DI, respectively

b. Relationship between rain area and IVG

In this section, the role of the convective area within precipitation systems is investigated. Three years of observations reveal that 99% of the data with intense gradients (IVG < −10 or >10 mm h−1 km−1) is associated with convective regions. In addition to the significant evaporation in severe convective storms with subsidence warming (Takemi 1999), the effect of strong updrafts is also important (e.g., Dotzek and Fehr 2003). System-dependent characteristics of IVG are examined along with the horizontal area of individual precipitation systems.

Figure 9 shows the convective fraction of systems with significant DD (IVG < −10 mm h−1 km−1) and DI (IVG > 10) profiles. The total convective area of a system is the summation of all the individual convective areas embedded in the system. Figure 9 is for May over inland India. It is reasonable to assume that larger precipitation systems had larger convective regions and that those were the main components of the total convective area in those systems. Gray points on large black points in Fig. 9a indicate that most large convective regions are embedded in large systems and include both significant DD and DI profiles. This is more clear in the individual convective area versus rain area plot shown in Fig. 9b. A large convective region on the scale of an MCS (e.g., larger than 104 km2) often includes both DD and DI profiles, whereas in small convective regions (∼100 km2) significant IVG is absent. Significant DD and DI profiles were also observed in separate convective regions in the same system. A few cases with significant IVG were found in small systems. Many of the significant DD and DI profiles were adjacent in the same large convective region as in Fig. 8. These systems, that have both significant DD and DI profiles, are referred to as slant cores for clarity. The proportion of these systems was 1.7% by number and 22.9% by area over India during the three years. About one-fourth of all rainshafts observed over inland India fall in this category. The systems have a tendency to be DD. Ninety percent of these systems are larger than 103 km2. It was observed that 57% of large systems (>104 km2) with significant IVG have slant cores. The interpretation may link the general understanding of oft-studied individual precipitation systems such as a developing cell containing several convective cells or airmass thunderstorms (e.g., Houze 1993). The lateral transport of water would be important in interpreting the structure, for example, as reported by Rutledge and Houze (1987) that 20% of the surface precipitation in the stratiform region is due to the influx of hydrometeors from the convective line.

Fig. 9.

Relationships between rain area and (a) total and (b) individual convective rain areas with significant DD (gray points) and DI profiles (large points) over inland India in May. Small dots represent precipitation systems without significant vertical gradients. Thin lines indicate the convective area fraction

Fig. 9.

Relationships between rain area and (a) total and (b) individual convective rain areas with significant DD (gray points) and DI profiles (large points) over inland India in May. Small dots represent precipitation systems without significant vertical gradients. Thin lines indicate the convective area fraction

The fractional area of convective rain was on the order of several tens of percent and had a scale dependence (Fig. 9a). There was a limitation as to the extent of the convective regions [thick line, area of convective rain = 2.1 (area of rain)0.85]. This boundary was tested for all rain events during the three years and found to be valid (not shown). The area is approximately 70% in the case of 100 pixels (1.5 × 103 km2) and 50% for 1000 pixels (1.5 × 104 km2). It strongly depends on the thresholds used in a classification method called the “H method” where a convective core must be significant when compared with the background rain intensity (Awaka et al. 1997; Steiner et al. 1995). Rain-type fraction is closely related to atmospheric heating whose heterogeneity affects dynamic systems (Schumacher and Houze 2003a). A simplified representation for a conglomerate of systems must have information on precipitation systems of various scales of (Fabry 1996; Uijlenhoet 2002).

The relationship between rain area and areally averaged IVG was investigated for different climatological regions. Figure 10 shows seasonal variations of the relationship between rain area and IVG over India, northern Africa, the Amazon River basin, and the Brazilian Plateau. Northern Africa (10°–25°N) was chosen as an example of a tropical landmass with both wet and dry conditions. Generally, over India and Africa in summer, DI profiles decreased as the area increased. This is similar to previous results using rain area density (HN02). Over India, the average IVG line (downward sloping white line in Fig. 10) shifts from DD to DI and back to DD over the period May to October with DI signatures in small, shallow storms and DD signatures in large, deep storms. Many moderate DD and a few significant DD storms characterized May. The IVG of larger systems was close to zero, whereas smaller systems were mostly DI in summer. Laing and Fritsch (1993b) found that the monthly distribution of mesoscale convective complexes (MCCs) over India peaks in mid to late summer. They explained the late summer peak by the maximum spatial extent of the monsoon. The increase and enlargement of large systems was consistent with the general MCC population. However, these large systems did not show a clear DD signature. The rain area was an order of magnitude smaller than that of general MCCs due to the difference between rain, cold cloud shields, and the narrow TRMM PR swath. The major rain mode over northern Africa was DD throughout the year. Precipitation systems appeared most frequently in August, which is consistent with the highest occurrence of MCCs around 10°N as reported by Laing and Fritsch (1993a). Over these two continental regions, there was a consistent increase in the large systems but the occurrence of significant IVG was inconsistent reflecting the environmental differences in the generation of quasi-oceanic and continental mesoscale systems there.

Fig. 10.

Seasonal variation of the relationships between area and IVG averaged for each system over India, northern Africa, the Amazon River basin, and the Brazilian Plateau. These regions are same as in Fig. 4. The line shows average IVG for 20 log intervals in rain area from 20 to 105 km2

Fig. 10.

Seasonal variation of the relationships between area and IVG averaged for each system over India, northern Africa, the Amazon River basin, and the Brazilian Plateau. These regions are same as in Fig. 4. The line shows average IVG for 20 log intervals in rain area from 20 to 105 km2

The lower two panels of Fig. 10 show examples over the moist Amazon River basin and the wet and dry Brazilian Plateau. The same downward sloping IVG relationship seen over India and Africa occurs in these dry, wet, and highly wet regimes. Over the Amazon River basin, most of the systems are DI except in seasons with relatively less rainfall (i.e., August–November). Small DI systems and large DD systems were seen throughout the year. Over the Brazilian Plateau, large systems were systematically DD in the austral summer. The DD trend was not as clear as the one over Africa.

c. Precipitation-system structure during the mature monsoon

Changes in individual systems from the onset to the mature phase of the monsoon over inland India are examined. The number of total, small (<100 km2), and large (>104 km2) systems is shown in Table 2. An area of 100 km2 represents the “cumulus scale” according to the nomenclature defined by Houze and Cheng (1977) with their minimum detectable signal echo thresholds. The number of profiles of significant IVG (<−10 or >10 mm h−1 km−1) is also given in Table 2. The number of all systems and their average area in August was 2.8 times more and 13% larger, respectively, than in May. The number of FOVs and rain coverage increased from May to August; however, significant IVG profiles decreased. Though the number of small systems make up about half of the total number of systems (48% in May and 54% in August), the rain coverage by these cumulus-scale systems is far less (3% in May and 4% in August). The number of DI profiles in small systems was 5.4 times greater in August than in May. With the rapid increase in the number of moderate DI systems from May to August, the DD fraction decreased from 48% to 19%. On the other hand, large systems in August were 1.9 times more in number and 28% larger in area than in May. In comparison with all systems, rain coverage was about 50%, whereas, the number of systems decreased from 4% to 3% due to the rapid increase of small systems. The DD fraction in large systems changed from 65% to 54% from May to August. Furthermore, the number of significant DD (DI) profiles in August was just 53% (32%) of that in May in spite of the increase in the number and area of precipitation systems. The standard deviation of IVG was reduced from 3.7 to 1.8 mm h−1 km−1. The number of mesoscale and synoptic-scale systems that were moderate DI increased.

Table 2.

The number of systems (No.) and embedded DD and DI profiles (FOV_DD, FOV_DI) in May and Aug. Small and large systems are defined as ones with areas less than 100 km2 and greater than 104 km2 , respectively. The bracket indicates the number of profiles with significant IVG (<−10 or >10 mm h−1 km−1 )

The number of systems (No.) and embedded DD and DI profiles (FOV_DD, FOV_DI) in May and Aug. Small and large systems are defined as ones with areas less than 100 km2 and greater than 104 km2 , respectively. The bracket indicates the number of profiles with significant IVG (<−10 or >10 mm h−1 km−1 )
The number of systems (No.) and embedded DD and DI profiles (FOV_DD, FOV_DI) in May and Aug. Small and large systems are defined as ones with areas less than 100 km2 and greater than 104 km2 , respectively. The bracket indicates the number of profiles with significant IVG (<−10 or >10 mm h−1 km−1 )

The role of significant IVG in a system was investigated. A precipitation system with significant DI profiles (IVG > 10 mm h−1 km−1) and no significant DD profiles (IVG < −10) is termed a significant-DI system. This type of system appeared constantly, 7 systems per month on average, throughout the year except in January and December when precipitation events were few. On the other hand, significant-DD systems, those having significant DD profiles (IVG < −10) alone, were observed from May to October (∼23 systems per month). In the hot season (phase B in section 5), significant-DD systems were few in comparison with significant-DI systems. The aforementioned slant cores which include both significant DD and DI profiles appeared in summer (∼13 systems month−1) and in particular during the onset of the monsoon. Figure 11 shows the relationship between IVG in the convective region with that in the stratiform region in May and August. The stratiform part of these systems shows system-dependent variation, even though these systems are classified based on intense gradients that mostly appear in the convective part. The absolute value of IVG in the convective region is larger than that in the stratiform region as expected. In general, they seem to have a linear relationship (sloping upward from left to right). Slant cores are DD especially in the stratiform region in May with the number of slant-core DD systems reduced by half in August. Most of the IVG in the stratiform portion of significant-DI systems changes from DD to DI. Convective cores aloft in significant-DD systems become moderate, and the DI fraction in the stratiform region increases.

Fig. 11.

IVG of areally averaged rainfall profiles in each category for convective (abscissa) and stratiform (ordinate) regions. They were obtained for each individual system over India in May and Aug. Precipitation systems are organized into all systems, slant cores, significant-DI ones, and significant-DD ones, respectively, from the top

Fig. 11.

IVG of areally averaged rainfall profiles in each category for convective (abscissa) and stratiform (ordinate) regions. They were obtained for each individual system over India in May and Aug. Precipitation systems are organized into all systems, slant cores, significant-DI ones, and significant-DD ones, respectively, from the top

7. Conclusions and discussion

The main objective was to study the spatial and temporal variation of the index of vertical gradient (IVG) of rainfall rate using data collected by the first spaceborne radar. The vertical gradient of rainfall rate was used to represent the structural differences of the vertical profiles. Whether the rainfall rate profile had a peak aloft or not was determined by the IVG being either negative (DD) or positive (DI) at low levels.

Profiles with rainfall rate peaks aloft often appeared over tropical interior landmasses in summer, corresponding to the seasonal variation of rainfall in the equatorial convective zones. The DD profiles broadly appeared as a continental pattern over Africa and the Brazilian Plateau in summer or as a horseshoelike pattern bordering the height of the wet season over India. The environmental differences between dry, wet, and highly wet regimes characterized the degree of IVG (i.e., the tendency to be DI, DD, and moderately DI, respectively). Atmospheric moisture was related to precipitation-system structure as reflected in the vertical gradient. In an extremely moist atmosphere during mid to late summer over India, the number of shallow, isolated, and weak DI precipitation systems and widespread systems with a moderate IVG increased.

In general, the IVG for individual systems or profiles decreased (DD tendency) as rain area increased or as storms developed vertically. Unlike relatively weak DD profiles in stratiform rain, the physical meaning of significant DD profiles is deemed to be due to other possible factors such as strong up-and downdrafts and slant cores apart from evaporation. Suspended water drops may imply the existence of strong updrafts. These systems would require strong forcing from the surface to develop deeply. Well-developed systems in May result from the large amount of latent and sensible heat fluxes from the surface and the moist environment in the middle troposphere (e.g., Shinoda and Uyeda 2002). On the other hand, storms with significant DI convective cores alone were also observed in every month over India. Significant-DI systems became a significant minority in the premonsoon season when they made up the greatest portion of severe precipitation systems. These systems may imply the existence of significant low-level convergence. Precipitation systems in the hot season (phase B in section 5) were characterized as deep storms with strong rainfall rates and significant DI profiles. In other words, only significant DI systems with strong low-level convergence might be able to develop under strong atmospheric subsidence in the Hadley cell.

Slant cores were also observed frequently with adjacent profiles that were significantly DD with high storm heights and significantly DI with strong rainfall rates in large convective areas. Slant cores may be consistent with rearward tilting updraft structures in a mature MCS (e.g., McAnelly et al. 1997; Lang et al. 2003). Most of them appeared in large systems (90% were larger than 103 km2) and were noticeably observed in May over India. These systems were specified by the existence of significant IVG and showed system-dependent links in the stratiform and convective patterns. The relationship between these patterns did not show that DD and DI were complementary. They also exhibited a wet to highly wet regime shift from May to August. Abundant moisture makes the regime quasi-oceanic by generating widespread precipitation or cloud shields. The appearance of widespread systems with moderate vertical gradients could be explained by an increase in stratiform rain fraction during the fully developed monsoon due to a more widespread warm, moist boundary layer and more uniform buoyancy (Schumacher and Houze 2003a). Precipitation or cloud systems may be able to change their own regime by reducing the solar insolation under highly wet conditions.

The fuller study of rainfall rate distribution at the lowest levels lies outside the scope of this study due to the uncertainty in the sampling below 2 km. However, investigation of the shallow peaks is also important, for example, to improve the retrieval of rainfall at the surface. Such studies are needed to know the concentration of numerous shallow precipitation systems with echo top around 2 km and to consider the topographic effect on the sampling over land.

In order to interpret the characteristics of the vertical profiles, understanding the type of precipitation system and the atmospheric conditions (dry/wet) was important. Further pattern recognition would make the interpretation of the stratiform and convective mixture more meaningful. This suggests that the investigation of average precipitation regimes, statistical analyses, or spectral representations would be more informative if precipitation regimes were recognized as being congregations of further parameterized precipitation-system types.

Acknowledgments

The data used in this paper were provided by Japan Aerospace Exploration Agency (previously known as the National Space Development Agency of Japan). The authors would like to express their gratitude to the members of the TRMM project, especially in the Earth Observation Research and Application Center. Helpful advice and constructive comments on the manuscript from Dr. A. Higuchi and Dr. H. Fujinami are gratefully acknowledged. Special thanks are due to Dr. T. N. Rao for his fruitful discussion and comments. Thanks are also extended to anonymous reviewers for detailed and valuable comments.

REFERENCES

REFERENCES
Adeyewa
,
Z. D.
, and
K.
Nakamura
,
2003
:
Preliminary study of rainfall and storm structure over Africa with TRMM precipitation radar data.
Meteor. Z
,
12
,
197
202
.
Awaka
,
J.
,
1998
:
Algorithm 2a23—Rain type classification.
Proc. Symp. on Precipitation Observation from a Non-Sun Synchronous Orbit, Nagoya, Japan, THAS and ESTO, 215–220
.
Awaka
,
J.
,
T.
Iguchi
,
H.
Kumagai
, and
K.
Okamoto
,
1997
:
Rain type classification algorithm for TRMM precipitation radar.
Proc. IGARSS'97, Singapore, IEEE, 1633–1635
.
Battan
,
L. J.
,
1976
:
Vertical air motions and Z–R relation.
J. Appl. Meteor
,
15
,
1120
1121
.
Bolen
,
S. M.
, and
V.
Chandrasekar
,
2000
:
Quantitative cross validation of space-based and ground-based radar observations.
J. Appl. Meteor
,
39
,
2071
2079
.
Dotzek
,
N.
, and
K. D.
Beheng
,
2001
:
The influence of deep convective motions on the variability of Z–R relations.
Atmos. Res
,
59–60
,
15
39
.
Dotzek
,
N.
, and
T.
Fehr
,
2003
:
Relationship between precipitation rates at the ground and aloft—A modeling study.
J. Appl. Meteor
,
42
,
1285
1301
.
Fabry
,
F.
,
1996
:
On the determination of scale ranges for precipitation fields.
J. Geophys. Res
,
101
,
12819
12826
.
Foote
,
G. B.
, and
P. S. D.
Toit
,
1969
:
Terminal velocity of raindrops aloft.
J. Appl. Meteor
,
8
,
249
253
.
Fu
,
Y.
, and
G.
Liu
,
2001
:
The variability of tropical precipitation profiles and its impact on microwave brightness temperatures as inferred from TRMM data.
J. Appl. Meteor
,
40
,
2130
2143
.
Fu
,
Y.
,
Y.
Lin
,
G.
Liu
, and
Q.
Wang
,
2003
:
Seasonal characteristics of precipitation in 1998 over East Asia as derived from TRMM PR.
Adv. Atmos. Sci
,
20
,
511
529
.
Garstang
,
M.
,
H. L.
Massie
,
J.
Halverson
,
S.
Greco
, and
J.
Scala
,
1994
:
Amazon coastal squall lines. Part I: Structure and kinematics.
Mon. Wea. Rev
,
122
,
608
622
.
Heymsfield
,
G. M.
,
B.
Geerts
, and
L.
Tian
,
2000
:
TRMM precipitation radar reflectivity profiles as compared with high-resolution airborne and ground-based radar measurements.
J. Appl. Meteor
,
39
,
2080
2102
.
Hirose
,
M.
, and
K.
Nakamura
,
2002
:
Spatial and seasonal variation of rain profiles over Asia observed by spaceborne precipitation radar.
J. Climate
,
15
,
3443
3458
.
Houze
,
R. A,Jr
,
1981
:
Structures of atmospheric precipitation systems—A global survey.
Radio Sci
,
16
,
671
689
.
Houze
,
R. A,Jr
,
1993
:
Cloud Dynamics.
Academic Press, 573 pp
.
Houze
,
R. A,Jr
, and
C. P.
Cheng
,
1977
:
Radar characteristics of tropical convection observed during GATE: Mean properties and trends over the summer season.
Mon. Wea. Rev
,
105
,
964
980
.
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
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis project.
Bull. Amer. Meteor. Soc
,
77
,
437
471
.
Katzfey
,
J. J.
, and
B. F.
Ryan
,
1997
:
Modification of the thermodynamic structure of the lower troposphere by the evaporation of precipitation: A GEWEX cloud system study.
Mon. Wea. Rev
,
125
,
1431
1446
.
Klemp
,
J. B.
, and
R. B.
Wilhelmson
,
1978
:
The simulation of three-dimensional convective storm dynamics.
J. Atmos. Sci
,
35
,
1070
1096
.
Kummerow
,
C.
, and
Coauthors
,
2000
:
The status of the Tropical Rainfall Measuring Mission (TRMM) after two years in orbit.
J. Appl. Meteor
,
39
,
1965
1982
.
Laing
,
A. G.
, and
J. M.
Fritsch
,
1993a
:
Mesoscale convective complexes in Africa.
Mon. Wea. Rev
,
121
,
2254
2263
.
Laing
,
A. G.
, and
J. M.
Fritsch
,
1993b
:
Mesoscale convective complexes over the Indian monsoon region.
J. Climate
,
6
,
911
919
.
Lang
,
S.
,
W. K.
Tao
,
J.
Simpson
, and
B.
Ferrier
,
2003
:
Modeling of convective–stratiform precipitation processes: Sensitivity to partitioning methods.
J. Appl. Meteor
,
42
,
505
527
.
Lim
,
Y-K.
, and
K-Y.
Kim
,
2002
:
Temporal and spatial evolution of the Asian summer monsoon in the seasonal cycle of synoptic fields.
J. Climate
,
15
,
3630
3644
.
Liu
,
G.
, and
Y.
Fu
,
2001
:
The characteristics of tropical precipitation profiles as inferred from satellite radar measurements.
J. Meteor. Soc. Japan
,
79
,
131
143
.
Manohar
,
G. K.
,
S. S.
Kandalgaonkar
, and
M. I. R.
Tinmaker
,
1999
:
Thunderstorm activity over India and the Indian southwest monsoon.
J. Geophys. Res
,
104
,
4169
4188
.
Matejka
,
T. J.
,
R. A.
Houze Jr
, and
P. V.
Hobbs
,
1980
:
Microphysics and dynamics of clouds associated with mesoscale rainbands in extratropical cyclones.
Quart. J. Roy. Meteor. Soc
,
106
,
29
56
.
McAnelly
,
R. L.
,
J. E.
Nachamkin
,
W. R.
Cotton
, and
M. E.
Nicholls
,
1997
:
Upscale evolution of MCSs: Doppler radar analysis and analytical investigation.
Mon. Wea. Rev
,
125
,
1083
1110
.
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
.
Ninomiya
,
K.
, and
C.
Kobayashi
,
1998
:
Precipitation and moisture balance of the Asian summer monsoon in 1991. Part I: Precipitation and major circulation systems.
J. Meteor. Soc. Japan
,
76
,
855
877
.
Parker
,
M. D.
, and
R. H.
Johnson
,
2000
:
Organizational modes of midlatitude mesoscale convective systems.
Mon. Wea. Rev
,
128
,
3413
3436
.
Petersen
,
W. A.
, and
S. A.
Rutledge
,
2001
:
Regional variability in tropical convection: Observations from TRMM.
J. Climate
,
14
,
3566
3586
.
Rickenbach
,
T. M.
,
R.
Nieto-Ferreira
,
J.
Halverson
,
D. L.
Herdies
, and
M. A. F.
Silva Dias
,
2002
:
Modulation of convection in the southwestern Amazon basin by extratropical stationary fronts.
J. Geophys. Res., 107, 8040, doi:10.1029/2000JD000263
.
Rosenfeld
,
D.
, and
Y.
Mintz
,
1988
:
Evaporation of rain falling from convective clouds as derived from radar measurements.
J. Appl. Meteor
,
27
,
209
215
.
Rutledge
,
S. A.
, and
P. V.
Hobbs
,
1984
:
The mesoscale and microscale structure and organization of clouds and precipitation in midlatitude cyclones. XII: A diagnostic modeling study of precipitation development in narrow cold-frontal rainbands.
J. Atmos. Sci
,
41
,
2949
2972
.
Rutledge
,
S. A.
, and
R. A.
Houze
Jr.
,
1987
:
A diagnostic modeling study of the trailing stratiform region of a midlatitude squall line.
J. Atmos. Sci
,
44
,
2640
2656
.
Salles
,
C.
, and
J-D.
Creutin
,
2003
:
Instrumental uncertainties in Z– R relationships and raindrop fall velocities.
J. Appl. Meteor
,
42
,
279
290
.
Schumacher
,
C.
, and
R. A.
Houze
Jr.
,
2003a
:
Stratiform rain in the Tropics as seen by the TRMM Precipitation Radar.
J. Climate
,
16
,
1739
1756
.
Schumacher
,
C.
, and
R. A.
Houze
Jr.
,
2003b
:
The TRMM Precipitation Radar's view of shallow, isolated rain.
J. Appl. Meteor
,
42
,
1519
1524
.
Shige
,
S.
,
Y. N.
Takayabu
,
W. K.
Tao
, and
D. E.
Johnson
,
2004
:
Spectral retrieval of latent heating profiles from TRMM PR data. Part I: Development of a model-based algorithm.
J. Appl. Meteor
,
43
,
1095
1113
.
Shinoda
,
T.
, and
H.
Uyeda
,
2002
:
Effective factors in the development of deep convective clouds over the wet region of eastern China during the summer monsoon season.
J. Meteor. Soc. Japan
,
80
,
1395
1414
.
Short
,
D. A.
, and
K.
Nakamura
,
2000
:
TRMM radar observations of shallow precipitation over the tropical oceans.
J. Climate
,
13
,
4107
4124
.
Stano
,
G.
,
T. N.
Krishnamurti
,
T. S. V. V.
Kumar
, and
A.
Chakraborty
,
2002
:
Hydrometeor structure of a composite monsoon depression using the TRMM radar.
Tellus
,
54
,
370
381
.
Steiner
,
M.
,
R. A.
Houze
Jr.
, and
S. E.
Yuter
,
1995
:
Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data.
J. Appl. Meteor
,
34
,
1978
2007
.
Syono
,
S.
, and
T.
Takeda
,
1962
:
On the evaporation of raindrops in a sub-cloud layer.
J. Meteor. Soc. Japan
,
40
,
245
265
.
Szoke
,
E. J.
, and
E. J.
Zipser
,
1986
:
A radar study of convective cells in mesoscale systems in GATE. Part II: Life cycles of convective cells.
J. Atmos. Sci
,
43
,
199
218
.
Takayabu
,
Y. N.
,
2002
:
Spectral representation of rain profiles and diurnal variations observed with TRMM PR over the equatorial area.
Geophys. Res. Lett., 29, 1584, doi:10.1029/2001GL014113
.
Takemi
,
T.
,
1999
:
Evaporation of rain falling below a cloud base through a deep atmospheric boundary layer over an arid region.
J. Meteor. Soc. Japan
,
77
,
387
397
.
Tao
,
W. K.
,
S.
Lang
,
J.
Simpson
, and
R.
Adler
,
1993
:
Retrieval algorithms for estimating the vertical profiles of latent heat release: Their applications for TRMM.
J. Meteor. Soc. Japan
,
71
,
685
700
.
Trenberth
,
K. E.
,
1999
:
Atmospheric moisture recycling: Role of advection and local evaporation.
J. Climate
,
12
,
1368
1381
.
Trenberth
,
K. E.
, and
C. J.
Guillemot
,
1998
:
Evaluation of the atmospheric moisture and hydrological cycle in the NCEP/NCAR reanalyses.
Climate Dyn
,
14
,
213
231
.
Tripoli
,
G. J.
, and
W. R.
Cotton
,
1980
:
A numerical investigation of several factors contributing to the observed variable intensity of deep convection over South Florida.
J. Appl. Meteor
,
19
,
1037
1063
.
Uijlenhoet
,
R.
,
2002
:
Parameterization of rainfall microstructure for radar meteorology and hydrology.
Ph.D thesis, Wageningen University, 279 pp
.
Westcott
,
N. E.
, and
P. C.
Kennedy
,
1989
:
Cell development and merger in an Illinois thunderstorm observed by Doppler radar.
J. Atmos. Sci
,
46
,
117
131
.
Williams
,
E. R.
,
M. E.
Weber
, and
R. E.
Orville
,
1989
:
The relationship between lightning type and convective state of thunderclouds.
J. Geophys. Res
,
94
,
13213
13220
.
Zipser
,
E.
, and
K.
Lutz
,
1994
:
The vertical profile of radar reflectivity of convective cells: A strong indicator of storm intensity and lightning probability?
Mon. Wea. Rev
,
122
,
1751
1759
.

Footnotes

Corresponding author address: Masafumi Hirose, Earth Observation Research and Application Center, Office Tower 22F, Harumi Island Triton Square, 1-8-10, Harumi, Chuo-ko, Tokyo 104-6023, Japan. Email: hirose@eorc.jaxa.jp