1. Introduction
Tropical deep convection plays an important role in regulating the tropical circulation and energy and water budgets. The mature stage of a tropical convective cloud cluster consists of four parts (Houze 1993): 1) deep precipitating convective towers characterized by vigorous updrafts; 2) stratiform precipitating cloud connected to the deep convection exhibiting weaker, mesoscale vertical motions; 3) nonprecipitating thick anvils attached to either the stratiform or convective precipitating areas; and 4) cirrus—optically thin cloud detrained from any of the three previous cloud types. This paper will focus on the third category—nonprecipitating thick anvils associated with deep convection.
Studies that have discussed tropical and subtropical anvil properties (Webster and Stephens 1980; Machado and Rossow 1993; Schumacher and Houze 2006; Frederick and Schumacher 2008; Rickenbach et al. 2008; Cetrone and Houze 2009; Theisen et al. 2009) suggest that anvils can play an important role in the water budget of deep convective cloud systems. Webster and Stephens (1980) first highlighted the large areal coverage of anvils during the Winter Monsoon Experiment (WMONEX) in Indonesia and the South China Sea. Machado and Rossow (1993) used International Satellite Cloud Climatology Project (ISCCP) data to look at the tropics-wide distribution of convective systems and the associated mesoscale anvil cloud area. However, these studies were limited in their ability to measure anvil vertical structure and did not separate the raining (i.e., stratiform rain region) from the nonraining areas, leaving large gaps in our understanding of tropical anvils.
The Tropical Rainfall Measuring Mission (TRMM) satellite was launched in November 1997 and the Precipitation Radar (PR) aboard it has the capability to detect the vertical structure of rainfall and thick anvils. The TRMM PR data have been widely used to investigate the macro properties of tropical convective systems (e.g., Nesbitt et al. 2000; Petersen and Rutledge 2001; Schumacher and Houze 2003). Based on PR observations over the east Atlantic and West Africa, Schumacher and Houze (2006) argued that stratiform rain regions may grow at the expense of anvils in situations of low upper-level wind shear and high low-level humidity. However, the TRMM PR has a low sensitivity (∼17 dBZ), which limits its ability to detect the part of the anvil composed of smaller particles. By combining TRMM PR and CloudSat Cloud Profiling Radar (CPR) observations, Cetrone and Houze (2009) investigated anvils over West Africa, the Maritime Continent, and the Bay of Bengal, and suggested that differences in thick anvil echo structure occur because West African anvils are more closely linked to its (graupel producing) convective core, while oceanic anvils tend to extend out from large stratiform rain areas.
Frederick and Schumacher (2008) used measurements from a C-band polarimetric Doppler Radar (C-POL), which has a sensitivity ∼0 dBZ, to investigate anvil characteristics observed during the Tropical Warm Pool International Cloud Experiment (TWP-ICE) in Darwin during the Australian monsoon. They showed that the ratio of anvil area to rain area was 0.5 averaged throughout the field campaign and that anvils often lasted several hours after the parent convection had expired. In addition, they separated anvils into ice (i.e., echo base >6 km) and mixed (i.e., echo base between 3 and 6 km) regions since the radiative properties of both anvil types can be quite different (Webster and Stephens 1980). The average thickness of ice and mixed anvils were 2.8 and 6.7 km, respectively, and anvil-top heights ranged from 10 to 17 km. Mixed anvils usually formed in association with stratiform rain, and its maximum area coverage typically preceded the peak in ice anvils.
Rickenbach et al. (2008) and Theisen et al. (2009) used weather radar, satellite, and aircraft observations from the Cirrus Regional Study of Tropical Anvils and Cirrus Layers Florida Area Cirrus Experiment (CRYSTAL-FACE) to further relate convective and anvil properties. Rickenbach et al. (2008) analyzed small convective systems from 23 July 2002 and found that anvils reached thier maximum area 1–2 h after the parent convection had collapsed, with longer lag times associated with increased convective rainfall and system size. Theisen et al. (2009) analyzed 19 single and multicell storms from July 2002 and found that anvils appeared 10 min after the maximum convective reflectivity and attained their maximum height 15 min later. In addition, the anvil’s mean particle size increased with time and decreased with altitude; larger particle sizes and concentrations were associated with stronger convection.
Yuan and Houze (2010) developed an objective method to distinguish mesoscale convective systems (MCS) and anvils by jointly using Moderate Resolution Imaging Spectroradiometer (MODIS) brightness temperatures, Advanced Microwave Scanning Radiometer (AMSR) rain fields, and CloudSat radar profiles. They showed that the modal thickness of CloudSat-observed anvils is ∼4.5 km and that anvil areal coverage ranges from 1.5 to 5 times the rain area radii. Regions of preferred anvil occurrence were dependent on MCS type—for example, small separated MCSs have maximum anvil coverage over tropical Africa, the Maritime Continent, and Amazon basin, while connected MCSs have maximum anvil coverage over the Indian Ocean and west Pacific warm pool.
Anvil properties and occurrence have potentially significant impacts on climate. One of the major uncertainties in simulations of climate change is cloud feedbacks (e.g., changes in cloud height, cover, and optical thickness in a warming atmosphere; Colman 2003; Stephens 2005; Clement and Soden 2005). Better understanding of the climatological variation of anvils and thier relationship to deep convection can potentially improve relevant general circulation model (GCM) parameterizations. For instance, Randall et al. (1989) investigated cloud radiative forcing (CRF) and its interaction with large-scale dynamics in a GCM and highlighted the importance of assessing the GCM’s cloud generation processes with observations. Krueger et al. (1995) stressed the importance of realistic simulation of anvil properties (e.g., extent and other microphysical properties) because of the radiative significance of anvils. Zender and Kiehl (1997) tested the climate sensitivity of anvil representation in the National Center for Atmospheric Research (NCAR) Community Climate Model (CCM) and found that the upper-tropospheric temperature structure, Hadley circulation, tropical deep convection, and Northern Hemisphere wintertime flow field were all sensitive to anvil and cirrus characteristics; therefore, they also stressed the importance of anvil representation in GCM cloud parameterizations. Yao and Del Genio (1999) compared the climate feedback obtained from doubled CO2 experiments with different parameterizations of large-scale clouds and moist convection by using the Goddard Institute for Space Studies (GISS) GCM. They showed that the presence of optically thick anvil clouds reduces the climate sensitivity by about 0.6°C. However, it is very difficult to represent anvil formation processes in GCMs. Therefore, they suggested using a semi-empirical approach based on statistical relationships derived from satellite data and cloud resolving models (CRMs) to reduce uncertainty. Lopez et al. (2009) compared the tropical convective clouds simulated by a CRM with observations from MODIS and AMSR. They found that the CRM underestimated anvils (defined as cloud with optical thickness between 4 and 32 and tops colder than 245 K) by a factor of 4 and, because of this underestimation, the model failed to simulate accurate longwave [outgoing longwave radiation (OLR)] and shortwave (albedo) radiative patterns compared to the observations.
Despite these studies, the tropics-wide occurrence and vertical extent of anvils remain relatively unknown, and the relationship of anvils to their parent convection and the large-scale environment is far from understood. This study aims to further investigate anvil properties and links between anvils, their parent convection, and environmental variables across the whole tropics using long-term satellite and reanalysis datasets. The 3 datasets involved in this research are 1) 10 years (1998–2007) of TRMM PR reflectivity to quantify the 3D structure of deep convective cloud systems (i.e., convective rain, stratiform rain, and anvils), 2) National Centers for Environmental Prediction (NCEP)–NCAR reanalysis relative humidity and winds to explore how the large-scale environment is associated with anvil production, and 3) coincident TRMM PR–CloudSat-CPR overpasses in order to quantify the horizontal and vertical extent of anvils that the PR is missing.
2. TRMM Precipitation Radar data and anvil definition
The TRMM PR was the first precipitation radar launched in space and started collecting data December 1997 (Kummerow et al. 1998; Kozu et al. 2001). The PR has a frequency of 13.8 GHz and operates at Ku band (2.17-cm wavelength) with a sensitivity ∼17 dBZ (18 dBZ after the August 2001 orbit boost). The horizontal and vertical resolution of the PR is 4.3 km (5 km postboost) and 250 m at nadir, which is sufficient to capture individual convective features and their vertical structure. This study uses TRMM product 2A25 (version 6) orbital data from 1 January 1998 to 31 December 2007, which includes the attenuation-corrected reflectivity and a copy of the rain type from TRMM product 2A23. The dataset covers the whole tropics and part of the subtropics (35°S–35°N).
TRMM product 2A23 takes into account both the vertical and horizontal variability in the observed reflectivity field to classify radar echo as convective rain, stratiform rain, or other (Awaka et al. 1997). The “other” category may include either noise or echo aloft not reaching the surface of the earth. Some of this echo aloft is attached to convective or stratiform rain echo (Schumacher and Houze 2006).
In this study, anvils are defined as echo with a 2A23 rain type (TRMM Precipitation Radar Team 2005): equal to 160—“Maybe stratiform, but rain hardly expected near surface. Bright band may exist but is not detected; 170—“Maybe stratiform, but rain hardly expected near surface. Bright band hardly expected. Maybe cloud only”—; or 300—“Other”. In addition, anvil reflectivity must be greater than 17 dBZ and have an echo base higher than 3 km. An example anvil image is shown in Fig. 1. The orbital data are grouped into 2.5° × 2.5° daily grid boxes within a 35°S–35°N domain. At least 20 anvil pixels in a grid box are required before statistics are calculated to guarantee that the selected anvil is not noise.
We subset the anvils into ice and mixed ice–water categories (e.g., Fig. 1) because ice and mixed-phase anvils have significantly different radiative heating profiles (Webster and Stephens 1980; Ackerman et al. 1988) and may represent different anvil evolution processes. For example, ice-anvil hydrometeors (which generally have lower reflectivity values, suggesting smaller particles) likely fall more slowly than mixed-anvil hydrometeors, thus affecting anvil lifetime and areal distribution. In addition, ice- and mixed-anvil populations may represent different responses to large-scale environmental factors such as upper-level wind shear and low-level evaporation. The ice- and mixed-anvil definitions are based on the study of Frederick and Schumacher (2008). Ice anvils have an echo base ≥6 km (∼−8°C) to help guarantee an all-ice composition, while mixed anvils have an echo base between 3 and 6 km (this threshold is based on the assumption that PR-observed echo bases are approximately the same as observed by C-POL since sedimentation leads to the largest particles at cloud base) and an echo top ≥6 km to guarantee that part of the anvil is composed of ice. It should be stressed that only anvils composed of large enough hydrometeors can be detected by the PR because of its low sensitivity (∼17 dBZ). Therefore, anvils in this study refer to the thick anvils observable by the PR.
3. Anvil echo top and extent comparisons with CloudSat
Because of the low sensitivity of the TRMM PR, PR-observed anvils will be underestimated in both the vertical and horizontal. The CPR aboard CloudSat is a 94-GHz, nadir-viewing radar with a sensitivity ∼−26 dBZ (Stephens et al. 2002). Its footprint is ∼1.3 km (across track) × 1.7 km (along track) and it has a vertical resolution of 250 m, providing high-resolution observations of cloud structure.
This study utilizes the 2D-CloudSat-TRMM product (v1.0) from August 2006 to December 2008 to investigate the missing part of anvils viewed by the TRMM PR. During the available data period, there were about 6061 files with a TRMM–CloudSat observation between 20°N and 20°S. The files include reflectivity from both radars as well as the rain type from TRMM product 2A23, making it possible to find the PR-derived anvil position in CPR observations. Out of 6061 ing files, there were 332 cases (5.5%) that included PR-defined anvils. After marking the anvils on both radar tracks (an example overpass image is shown in Fig. 2), the difference of the echo top between the CloudSat CPR and TRMM PR was calculated. The PR echo top in these files is the height of the 20-dBZ echo (or “storm height”) and the CPR echo top is the height of the −24-dBZ echo (or “cloud top”). Statistics show that the PR underestimates anvil tops from 1 to 10 km with an average of ∼5 km compared to the CPR (Fig. 3a). Casey et al. (2007) investigated the cloud-top difference between the TRMM PR and Geoscience Laser Altimeter System (GLAS) aboard the Ice, Cloud, and Land Elevation Satellite (ICESat), and an analysis of the cloud-top difference in anvils based on his dataset also shows that the PR underestimates cloud tops by ∼5 km compared to the ICESat lidar. For another consistency check, optical depths from CloudSat product 2B-TAU were calculated for the PR-defined anvils; values were consistently large, with a peak near 35.
In addition, the PR will miss anvils in the horizontal direction. Of the 332 coincident PR–CPR images with PR-defined anvils, the most robust cases (229 images) were used to subjectively compare the difference in anvil horizontal extent between CloudSat and the TRMM PR. Anvils observed by CloudSat were defined as echo with a base between 3 and 8 km and at least 3 km thick, primarily based on the CloudSat classifications in Riley and Mapes (2009). The horizontal area factor being missed by the TRMM PR is defined as PR-missing–PR-identified anvils. In the CPR dataset, clouds are sometimes defined as anvils that are not obviously associated with deep convection, so they are excluded from the comparison since TRMM PR anvils not connected with convective systems are mostly just noise. Figure 3b shows that the PR underestimates anvil area by a factor of 3 to 6 depending on echo-base height, with an average factor of ∼4 compared to the CPR. The rest of the results will not take these corrections into account, but the reader should remain aware that the PR sees only the thickest portion of the discussed anvil and that extrapolation will be necessary to fully discuss the potential role anvils have on the radiative heating and water budget of deep convective cloud systems.
4. Anvil occurrence
Anvil occurrence is defined as days with anvils within a 2.5° × 2.5° grid divided by the total number of days that the PR reported observations during 1998–2007. Anvils observable by the PR occurs around 5% of the time across the tropics with distinct geographical variability (Fig. 4). In the tropics, anvils most commonly occur over Africa, the Maritime Continent, and Panama (i.e., about 9% of the time). The frequent occurrence of anvils over these regions is consistent with Webster’s (1983) definition of these regions as semipermanent equatorial convective zones that are not as affected by seasonal shifts in convection, unlike monsoon and intertropical convergence zone (ITCZ) regions that have lower anvil occurrence (i.e., about 3%–4% of the time). Anvils also frequently occur at higher latitudes, in part because of the more frequent sampling of the TRMM satellite (i.e., more possible orbit visits in a daily grid), but also because anvils can be prevalent in midlatitude frontal systems.
5. Anvil areal coverage
Anvil areal coverage is defined as the number of anvil columns observed by the TRMM PR divided by the total number of PR columns (both with and without echo) in the grid box, times 100. Conditional values represent the average percent a grid box is covered by anvils when anvils are present, and unconditional values represent the average percent a grid box is covered by anvils regardless of whether anvils are present. Figure 5a shows that the conditional anvil areal coverage is larger over land and smaller over ocean. For example, anvil areal coverage is 2%–3% over Africa and South America, but only about 1% over the west Pacific warm pool. The high areal coverage of anvils over Africa, northern Australia, and parts of Central and South America suggests that these regions are effective in producing large amounts of anvils per unit convection. The opposite is true over much of the tropical oceans. The efficiency of convection in producing anvils will be further explored in section 8.
Figures 5b,c show the geographical distribution of conditional anvil areal coverage separated into ice and mixed subtypes. Ice anvils account for about a third of the total anvil area and have higher areal coverage over land (0.5%–1%) than ocean (0%–0.5%). Similarly, mixed anvils have higher areal coverage over land (1%–2%) than ocean (0.5%–1%). In the tropical belt, ice- and mixed-anvil areal coverage is maximum over Africa where convection has been observed to be the deepest and most intense in the tropics (Petersen and Rutledge 2001; Alcala and Dessler 2002; Schumacher and Houze 2003; Zipser et al. 2006). The lower anvil amounts over the tropical oceans (e.g., the west Pacific) may also be due to the preference of stratiform rain to form at the expense of anvils since the ocean boundary layer is warm and humid with a weak diurnal cycle, which is favorable for the sustainability of convection and stratiform rain production (Schumacher and Houze 2003; Schumacher and Houze 2006). Section 8a directly compares stratiform rain area to anvil area over key locations in the tropics.
The unconditional anvil areal coverage geographical distribution is shown in Fig. 6, since it can be used to calculate the tropics-wide anvil radiative heating field. Maxima of more than 0.3% occur over central Africa and Panama. Secondary maxima of about 0.2% occur over West Africa, the Maritime Continent, and the Amazon basin, while ocean values are closer to 0.1%. Ice- and mixed-anvil patterns are similar, although more mixed anvils are evident in the subtropics.
6. Anvil vertical extent
Grid-averaged-anvil 17-dBZ echo tops range from 8 to 9.5 km (Fig. 7a) with a tropics-wide average of 8.5 km (Table 1). Frederick and Schumacher (2008) found that average anvil echo tops ranged between 10 and 17 km during the Australian monsoon based on more sensitive C-POL observations, in agreement with the CloudSat comparisons in section 3. In addition, anvils were generally higher over land than ocean by 0.5 km. Cetrone and Houze (2009) found that anvils observed by CloudSat tended to be higher over the ocean. However, the reverse is true when only considering CloudSat reflectivities greater than 0 dBZ, which is more consistent with the thick anvils observable by the TRMM PR. Cetrone and Houze (2009) also showed that TRMM PR convective and stratiform echoes reach higher heights over land than ocean because of greater buoyancy. Thus, land convection appears to produce larger hydrometeors at higher heights that then detrain into nearby anvils.
When separated into subtypes, ice-anvil echo tops range from 9 to 10 km (Fig. 7b) with a tropics-wide average of 9.3 km (Table 1), while mixed-anvil echo tops are lower, ranging from 7.7 to 9 km (Fig. 7c) with a tropics-wide average of 8.3 km (Table 1). Land–ocean echo-top differences within the ice and mixed subtypes are also ∼0.5 km. The lower mixed-anvil echo tops likely result from faster sedimentation of the larger mixed-anvil particles, thus rapidly leaving aloft particles too small for the TRMM PR to see. In addition, anvils can either be connected to the convective core or extrude from the stratiform rain region. Cetrone and Houze (2009) showed that anvils preferentially form closer to the active convective regions over land and extends out from large stratiform rain regions over the ocean. Ice anvils would be higher than mixed anvils over land if ice anvils were directly tied to the deeper convective cores.
Grid-averaged anvils are 2.4–3.1 km thick (Fig. 8) with a tropics-wide average of 2.7 km (Table 1). Anvils are also thicker over land than ocean by 0.3 km. While ice anvils are higher than mixed anvils, they are also thinner (Table 1). Climatologically, PR-observed ice anvils are 1.4 km thick, while PR-observed mixed anvils are 3.2 km thick. Again, land–ocean thickness differences within the ice and mixed subtypes are similar to the overall anvil thickness difference of 0.3 km. The vertical extent of anvils is likely determined by the vertical extent and intensity of the parent convection. In addition, large-scale upper-level wind shear and midlevel moisture also may play an important role. These topics will be discussed in section 8.
7. Temporal variability in thick anvils
a. Seasonal variability
The ITCZ experiences a seasonal north–south migration, and monsoon regions—by definition—experience significant seasonal variability. Since tropical anvils result from deep convection, it is useful to investigate the significance of seasonal variability in anvil macroproperties. Figure 9 shows that anvil echo tops experience significant seasonal variability over South Asia, northern Australia, and Mexico, all of which are affected by monsoon circulations. Anvil areal coverage (not shown) shows a noisier pattern with heightened variability on the order of 1% over Africa in addition to parts of South Asia, northern Australia, and Central and South America. Anvil thickness (not shown) shows very little seasonal variability.
b. Interannual variability
Since thick anvils are closely associated with tropical convection, which is shown to respond to different phases of the El Niño–Southern Oscillation (ENSO; Rasmusson and Wallace 1983, among many other studies), an ENSO signal in anvil occurrence would be expected. Based on the National Oceanic and Atmospheric Administration (NOAA) multivariate ENSO index (MEI), four abnormally warm and four abnormally cold ENSO events were identified (Table 2). The average anvil occurrence map for the cold events was then subtracted from the average anvil occurrence map for the warm events (Fig. 10a). Anvils occur 2%–8% more often in the central and eastern Pacific and 5% less often in the far west Pacific during El Niño compared to La Niña. This pattern is consistent with the variability of deep convection (in this case, convection with echo tops >7 km) occurrence during ENSO (Fig. 10b). The conditional anvil area anomaly (not shown) did not indicate an ENSO signal; therefore, it is speculated that anvil occurrence (through increased or decreased occurrence of convection) is more sensitive to SST changes than anvil areal coverage (cf. Lindzen et al. 2001, which argues the opposite).
8. Factors related to anvil generation and evolution
Anvils, by definition, are associated with deep convection, but very little work has been done to quantify the relationship between anvils and their parent convection or how the large-scale environment affects anvil formation and evolution. Investigation of these issues will help to better understand the processes necessary for anvil production, which also provides useful information for model evaluation and parameterization improvement.
a. Relationship between characteristics of convection and thick anvils
Convective rain viewed by the TRMM PR occurs about 35% of the time in daily grids of 2.5° resolution (not shown). Some of these convective cells may have thick anvils associated with them, while some will not because of factors such as the height of the convection, the convective intensity, or the size of the convective system. Figure 11 shows that convection occurs in conjunction with thick anvils occurs ∼18% of the time across the tropics within the 2.5° daily grid boxes, but varies geographically. Values are typically ∼9% over ocean and 21% over land. If only convection with echo tops >7 km is considered (not shown), the land–ocean percentages are 28% and 21%, respectively. Percentages are greatest over Africa, the southern United States and Mexico, and southern South America, while other land regions, such as the Maritime Continent and the Amazon basin, are notably closer to the tropics-wide mean. Note that the analysis in this section assumes that convection and anvils occur together because of the fairly large size of the grid box (2.5°) and because the PR will have difficulty observing the weaker reflectivities of remnant anvils that are left after the parent convection no longer exists.
To investigate the relationship between convective height and anvil occurrence, the average 17-dBZ echo top of convective rain in the presence of anvils (Fig. 12a) and in the absence of anvils (Fig. 12b) was calculated. Convection associated with thick anvils is usually, on average, at least 7 km high, which is logical since anvils require sufficient ice from the convective region to form and grow. Convection is even deeper over land when anvils are present, with average echo tops 8–10 km. Convection without anvils has average echo tops closer to 6 km. To further highlight the difference in convective echo tops when anvils ares present and not, Fig. 13 shows the histogram of the daily mean grid values in each situation. When convective echo tops average 5 km, anvils are not that common, whereas at 6 km there is an equal chance for anvils to occur or not to occur. Anvils become much more likely once convective echo tops average 7 km.
The difference between Figs. 12a and 12b (shown in Fig. 12c) indicates that convection associated with anvils is 1–3 km deeper than convection that is not. Regions where deep convection is prevalent, such as Africa, the west Pacific, and South America, show smaller echo-top differences than regions where deep convection is less common, such as the western Indian Ocean, northwest tropical Pacific, and Caribbean. This result suggests that variations in the large-scale environment may play a more important role in anvil variability in regions where convection regularly averages heights greater than 7 km. For regions where convection is less deep, variations in the depth of the convection and the large-scale environment likely contribute more equally to anvil occurrence variability.
Convective strength also appears to be associated with thick anvil occurrence. As shown in Fig. 14, more intense convection tends to produce thick anvils. For example, the mean reflectivity at 2 km for convection that occurs with anvils is 34–35 dBZ over Africa (Fig. 14a), but is 32–33 dBZ for convection that does not (Fig. 14b). The differences over Africa are the largest in the tropics (Fig. 14c). In general, reflectivity differences between convection that does and does not occur with anvils are closer to 1 dB. This relationship is further highlighted in the histograms in Fig. 15. When 32 dBZ is observed at 2-km height in a convective cell, the convection is equally likely to produce an anvil as not; however, when 34 dBZ is observed, anvils are much more likely to occur. In addition, convection associated with anvils over land is 1–2 dB stronger than over ocean (Fig. 14a). The same plots were created with convective reflectivity at 6 km and the patterns were very similar, but not as statistically significant. Since radar reflectivity is related to both hydrometeor size and number density, it is hard to say whether larger or more hydrometeors are produced at higher reflectivities. However, it is highly likely that stronger updrafts (especially over land) loft larger and/or more hydrometeors to higher heights, providing more fuel for the anvil region. It can also be speculated that updraft strength may change the hydrometeor density or size distribution in an anvil, leading to changes in the anvil radiative heating profile.
The size of the convective rain area (i.e., in a 2.5° × 2.5° grid box) may also be linked to anvil areal coverage. Anvils often form in the mature stage of tropical convective systems (Houze 1982; Machado and Rossow 1993); therefore, one would expect higher anvil areal coverage when more (or larger) convective cells are present. Convection associated with anvils (Fig. 16a) has conditional areal coverage ranging from 3% to 5%, whereas convection that is not associated with anvils (Fig. 16b) has conditional areal coverage closer to 2%. The difference between Figs. 16a and 16b (shown in Fig. 16c) is similar to the echo-top difference map (Fig. 14c) in that larger differences are required to form anvils in regions where deep convection can be marginal or occur infrequently.
To refine the relationship between the convective and anvil characteristics discussed above, multiple linear regression was done between anvil areal coverage and 17-dBZ echo-top heights and convective areal coverage, echo tops, and 2-km reflectivity. Figure 17 shows that convective areal coverage is by far the best predictor for anvil area and echo tops. When excluding convective area, convective echo-top height is the next most important predictor.
Figure 18 shows the frequency distribution of anvil area as a function of convective rain area over Africa (land area in 5°S–15°N, 30°W–30°E), tropical South America (land area in 15°S–10°N, 40°–90°W), and the west Pacific warm pool (ocean area in 15°S–10°N, 90°E–180°). Over Africa, the deep and intense mesoscale convective systems that generate anvils usually occur during the active monsoon period (Fortune 1980; Hodges and Thorncroft 1997; Schumacher and Houze 2006; Cetrone and Houze 2009). In contrast, the seasonal variation of convection over the tropical ocean is weak. Seasonal variability of convection and rainfall also exists over near-equatorial South America (Vera et al. 2006). In particular, the warm and moist environment during the wet season (similar to the tropical ocean) provides favorable conditions for the formation of convection. Therefore, South America generally lies somewhere between Africa and the west Pacific in convective and anvil properties. With this in mind, Fig. 18 reveals the following signatures:
Land exhibits a wider spectrum of areal coverage than that over ocean. For example, Africa has convective rain areas ranging from 0% to 8% and anvil areas ranging from 0% to 4%, whereas the west Pacific is more tightly distributed with convective rain areas ranging from 1% to 6% and anvil areas ranging from 0% to 2.5%.
For the same convective rain area, more anvils occur over land (
While it is easier to understand that smaller convective rain areas are related to less anvil area, the reason for very large convective rain areas corresponding to less anvil area is hard to determine. One possible reason is if the convective rain area is very large or highly organized, stratiform rain may form in preference to anvils (Schumacher and Houze 2006). Alternatively, the system may be too large for the PR swath to capture the statistics of the whole storm.
Figure 19 shows the areal coverage of thick anvils as a function of stratiform rain area. Similar to the convective area plots, the spread of the frequency distribution decreases as one goes from Africa, to South America, to the west Pacific. However, the stratiform rain area median increases in that same order while the convective rain area median was stable between regions. In addition, the slope of the fit line is slightly positive over Africa, near zero over South America, and slightly negative over the west Pacific. Therefore, stratiform rain area does not appear to be a good predictor for anvil production. Schumacher and Houze (2006) had suggested that stratiform rain may grow at the expense of anvils in situations of low upper-level wind shear and high low-level humidity, but this argument would only hold over the west Pacific based on these statistics.
b. Large-scale features associated with anvil occurrence
In tropical convective systems, stratiform precipitation accounts for 30%–70% of the total rain (Schumacher and Houze 2003) and using the C-POL observations from Darwin, Australia, Frederick and Schumacher (2008) showed that anvils typically formed after stratiform rain area peaked. In addition, upper-level shear can be a factor in the formation of stratiform rain (Houze 1993, 1997). Therefore, this study compares the zonal wind profile during situations when stratiform rain and thick anvils coexist and when only stratiform rain occurs to differentiate the features of upper-level shear during the occurrence of stratiform rain and anvils.
Figure 20a shows the zonal wind profile under the situations of no stratiform rain present, stratiform rain without anvils and stratiform rain with anvils over Africa, the west Pacific warm pool, and South America for 1998–2007 based on daily NCEP reanalysis data. The zonal wind over tropical Africa shows strengthening of the near-surface westerlies and upper-level easterlies when stratiform rain is present and even stronger reinforcement of the wind profile when stratiform rain and anvils coexist. In fact, central and West Africa are under the influence of three circulations at different heights:
African easterly jet (AEJ) is a prominent midtropospheric zonal jet resulting from strong meridional soil moisture gradients (Burpee 1972; Cook 1999). The AEJ is often prominent over Africa at approximately 600–700 hPa.
Tropical easterly jet (TEJ) is a prominent upper-level jet at about 200 hPa. Since the TEJ originates over the Tibetan Plateau, it becomes most pronounced during boreal summer and extends from Southeast Asia to Africa.
Monsoonal flow occurs near the surface of the earth resulting from the differential heating of ocean and land, which is manifested as the increase of westerly winds during the wet season.
Schumacher and Houze (2006) speculated that strengthening of the TEJ over Africa favors the formation of stratiform rain. Further strengthening of the TEJ may then enhance the detrainment of hydrometeors from the stratiform rain region to form anvils. As shown in Fig. 20a, enhancement of the upper-level TEJ may positively impact the formation of anvils over Africa.
The west Pacific warm pool experiences weak tropical easterlies through all levels. The zonal wind profile shows the weakening of the easterlies (or strengthening of the westerlies) below 300 hPa and strengthening of the easterlies above. However, compared to tropical Africa, the difference between the anvils and no-anvils situation is small. Therefore, it appears that upper-level shear plays less of a role in anvil production over the west Pacific. The MCSs over this region are larger and live longer than over many other tropical regions (Chen et al. 1996), such that convective rain area coverage may be more relevant.
South America is generally influenced by low-level easterly winds and weak upper-level westerly winds; opposite of Africa during the monsoon season. However, the change of zonal wind shear is as in Africa, such that the zonal wind profile difference (i.e., enhancement of the easterlies) between 500 and 200 hPa appears to be conducive to anvil production.
Figure 20b is similar to Fig. 20a but shows the relative humidity profile for each region. Over Africa, the entire atmospheric column is more humid by ∼6% when anvils are present, while humidity increases are close to 0 at low levels and rise to 6% above 600 hPa over the west Pacific and South America in the presence of anvils. It is unclear whether this increase is the result of anvils moistening the atmosphere or the more humid environment assisting in anvil generation. However, Sobel et al. (2004) suggested that stratiform rain regions moisten the mid- and upper troposphere based on lag correlations over Kwajalein in the central west Pacific, and the PR-observed stratiform rain regions are larger over these regions when anvils are present (14% versus 5%). Thus, the more robust stratiform rain regions may account for the moister environment when anvils are present, or anvils may similarly act to moisten the mid- and upper troposphere.
Based on the results from this section, the occurrence of anvils in the presence of stratiform rain may be separated into two stages. In the first stage, a certain amount of upper-level wind shear enhances hydrometeor detrainment from deep convection, thus feeding and increasing the stratiform rain area, which is accompanied by increased environmental humidity in the mid- and upper troposphere. When the wind shear strengthens further, the top part of the stratiform rain area is advected away to form anvils. The geographical variation of anvil vertical extent (i.e., higher and thicker anvils over land) implies that there will be a deeper layer of upper-level wind shear over land through which deep convection extends and from which material is blown off to form anvils. Over ocean, the lower and less thick anvils suggests upper-level wind shear is less effective in advecting hydrometeors from the less deep parent convection.
9. Conclusions
Studies on the climatological properties of anvils (i.e., thick, nonprecipitating cloud associated with deep convection) have been limited because of the lack of long-term, geographically diverse observations, although anvils are an important component of deep convective cloud systems. This study aims to fill this gap by describing the climatology of anvil properties across the tropics and subtropics using 10 years (1998–2007) of observations from the TRMM PR.
While the PR can observe the three-dimensional structure of deep convective systems—including thick anvils—it underestimates the horizontal and vertical extent of weak echo (i.e., <∼17 dBZ) because of its inability to sense small water and ice hydrometeors. Based on comparisons with CloudSat, the PR missed on average 5 km from the anvil tops and a factor of 4 of anvil horizontal extent. The results in this study do not take into account any echo-top or area corrections related to the TRMM PR’s lack of sensitivity. However, these corrections should be taken into account when estimating the radiative effect of anvils.
Across the tropics, PR-observed anvils occur ∼5% of the time and have a conditional areal coverage of 1.5%. Unconditional areal coverage ranges from 0.1% to 0.3%. Average anvil 17-dBZ echo tops range from 8 to 9.5 km with a tropics-wide average of 8.5 km. Average anvil thicknesses range from 1.3 to 3.5 km with a tropics-wide average of 2.7 km.
Anvils over land are usually higher and thicker than over ocean. Anvils occur most frequently over Africa, the Maritime Continent, and Panama; but have the largest conditional areal coverage over Africa, making Africa a hot spot of anvil production. When anvils are separated into ice and mixed subtypes (with mixed anvils indicating the possibility of both water and ice hydrometeors being present), further variations in these anvil properties are evident (e.g., mixed anvils are thicker and have lower echo tops than ice anvils and account for about two-thirds of the total anvil areal coverage).
Anvil properties also experience seasonal and interannual variability. Since thick anvils are created by deep convection, this temporal variability is likely due to the variability of the parent deep convection and/or the large-scale environment. For example, anvil echo tops show strong seasonal variability over monsoon regions, and anvil occurrence shows strong interannual variability over the central Pacific during ENSO events.
Not all convection is associated with thick anvils. Convection that occurs with anvils is deeper, stronger, and covers more area (as would be found in the mature stage of a tropical convective system), suggesting that each of these factors plays a role in producing and sustaining thick anvils. Multiple regression analysis shows that convective areal coverage is more important than convective echo tops or reflectivity in predicting anvil areal coverage and echo-top heights. Anvil-to-convective rain area ratios range from ¼ to 
Stronger wind shear is associated with the detrainment from the top of the parent convective and stratiform rain regions. The stronger the upper-level shear after stratiform rain forms, the more likely anvils are to occur. The upper-level TEJ may partially explain why more anvils occur over Africa compared to other tropical land and ocean locations. The mid- to upper troposphere has higher relative humidity (by ∼6%) in the presence of anvils, but this is likely a result rather than a cause of anvils, extrapolating from the results in Sobel et al. (2004).
Anvils (with a variety of possible definitions) have been shown to have large radiative signatures in the vertical with potentially significant impacts on the large-scale circulation (Webster and Stephens 1980; Hartmann et al. 1984; Slingo and Slingo 1991; Sherwood et al. 1994). However, even after an areal coverage correction based on the CloudSat comparison, thick anvils observed by the TRMM PR do not cover large areas and thus likely do not have a large-scale radiative importance. However, PR-observed anvils can cover large areas on smaller time and space scales, and are likely still very important to local water budget processes of the convective systems themselves.
Acknowledgments
This research was supported by ARM-DOE Grant DE-FG02-06ER64174. We appreciate Sean Casey’s assistance with the ICESat comparisons. We also appreciate the valuable comments of Ed Zipser and an anonymous reviewer.
REFERENCES
Ackerman, T. P., K.-N. Liou, F. P. J. Valero, and L. Pfister, 1988: Heating rates in tropical anvils. J. Atmos. Sci., 45 , 1606–1623.
Alcala, C. M., and A. E. Dessler, 2002: Observations of deep convection in the tropics using the Tropical Rainfall Measuring Mission (TRMM) precipitation radar. J. Geophys. Res., 107 , 4792. doi:10.1029/2002JD002457.
Awaka, J., T. Iguchi, H. Kumagai, and K. Okamoto, 1997: Rain type classification algorithm for TRMM precipitation radar. Proc. Int. Geoscience and Remote Sensing Symp., Suntec City, Singapore, IEEE, 1633–1635.
Burpee, R. W., 1972: The origin and structure of easterly waves in the lower troposphere of North Africa. J. Atmos. Sci., 29 , 77–90.
Casey, S. P. F., A. E. Dessler, and C. Schumacher, 2007: Frequency of tropical precipitating clouds as observed by the Tropical Rainfall Measuring Mission Precipitation Radar and ICESat/Geoscience Laser Altimeter System. J. Geophys. Res., 112 , D14215. doi:10.1029/2007JD008468.
Cetrone, J., and R. A. Houze Jr., 2009: Anvil clouds of tropical mesoscale convective systems in monsoon regions: Satellite radar climatology. Quart. J. Roy. Meteor. Soc., 135 , 305–317.
Chen, S. S., R. A. Houze Jr., and B. E. Mapes, 1996: Multiscale variability of deep convection in relation to large-scale circulation in TOGA COARE. J. Atmos. Sci., 53 , 1380–1409.
Clement, A. C., and B. Soden, 2005: The sensitivity of the tropical-mean radiation budget. J. Climate, 18 , 3189–3203.
Colman, R., 2003: A comparison of climate feedbacks in general circulation models. Climate Dyn., 20 , 865–873.
Cook, K. H., 1999: Generation of the African easterly jet and its role in determining West African precipitation. J. Climate, 12 , 1165–1184.
Fortune, M., 1980: Properties of African squall lines inferred from time-lapse satellite imagery. Mon. Wea. Rev., 108 , 153–168.
Frederick, K., and C. Schumacher, 2008: Anvil characteristics as seen by C-POL during the Tropical Warm Pool International Cloud Experiment (TWP-ICE). Mon. Wea. Rev., 136 , 206–222.
Hartmann, D. L., H. H. Hendon, and R. A. Houze Jr., 1984: Some implications of the mesoscale circulations in tropical cloud clusters for large-scale dynamics and climate. J. Atmos. Sci., 41 , 113–121.
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.
Houze Jr., R. A., 1982: Cloud clusters and large-scale vertical motions in the tropics. J. Meteor. Soc. Japan, 60 , 396–410.
Houze Jr., R. A., 1993: Cloud Dynamics. International Geophysics Series, Vol. 53, Academic Press, 573 pp.
Houze Jr., R. A., 1997: Stratiform precipitation in regions of convection: A meteorological paradox? Bull. Amer. Meteor. Soc., 78 , 2179–2196.
Kozu, T., and Coauthors, 2001: Development of precipitation radar onboard the Tropical Rainfall Measuring Mission (TRMM) satellite. IEEE Trans. Geosci. Remote Sens., 39 , 102–116.
Krueger, S. K., Q. Fu, K. N. Liou, and H.-N. S. Chin, 1995: Improvements of an ice-phase microphysics parameterization for use in numerical simulations of tropical convection. J. Appl. Meteor., 34 , 281–287.
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.
Lindzen, R. S., M. D. Chou, and A. Y. Hou, 2001: Does the Earth have an adaptative infrared iris? Bull. Amer. Meteor. Soc., 82 , 417–432.
Lopez, M. A., D. L. Hartmann, P. N. Blossey, R. Wood, C. S. Bretherton, and T. L. Kubar, 2009: A test of the simulation of tropical convective cloudiness by a cloud-resolving model. J. Climate, 22 , 2834–2849.
Machado, L. A. T., and W. B. Rossow, 1993: Structural characteristics and radiative properties of tropical cloud clusters. Mon. Wea. Rev., 121 , 3234–3260.
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.
Petersen, W. A., and S. A. Rutledge, 2001: Regional variability in tropical convection: Observations from TRMM. J. Climate, 14 , 3566–3586.
Randall, D. A., Harshvardhan, D. A. Dazlich, and T. G. Corsetti, 1989: Interactions among radiation, convection, and large-scale dynamics in a general circulation model. J. Atmos. Sci., 46 , 1943–1970.
Rasmussen, E. M., and J. M. Wallace, 1983: Meteorological aspects of the El Niño/Southern Oscillation. Science, 222 , 1195–1202.
Rickenbach, T., P. Kucera, M. Gentry, L. Carey, A. Lare, R.-F. Lin, B. Demoz, and D. O. Starr, 2008: The relationship between anvil clouds and convective cells: A case study in south Florida during CRYSTAL-FACE. Mon. Wea. Rev., 136 , 3917–3932.
Riley, E. M., and B. E. Mapes, 2009: Unexpected peak near −15°C in CloudSat echo top climatology. Geophys. Res. Lett., 36 , L09819. doi:10.1029/2009GL037558.
Schumacher, C., and R. A. Houze Jr., 2003: Stratiform rain in the tropics as seen by the TRMM Precipitation Radar. J. Climate, 16 , 1739–1756.
Schumacher, C., and R. A. Houze Jr., 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.
Schumacher, C., R. A. Houze Jr., and I. Kraucunas, 2004: The tropical dynamical response to latent heating estimates derived from the TRMM precipitation radar. J. Atmos. Sci., 61 , 1341–1358.
Sherwood, S. C., V. Ramanathan, T. P. Barnett, M. K. Tyree, and E. Roeckner, 1994: Response of an atmospheric general circulation model to radiative forcing of tropical clouds. J. Geophys. Res., 99 , 20829–20846.
Slingo, J. M., and A. Slingo, 1991: The response of a general circulation model to cloud longwave radiative forcing. II: Further studies. Quart. J. Roy. Meteor. Soc., 117 , 333–364.
Sobel, A. H., S. E. Yuter, C. S. Bretherton, and G. N. Kiladis, 2004: Large-scale meteorology and deep convection during TRMM KWAJEX. Mon. Wea. Rev., 132 , 422–444.
Stephens, G. L., 2005: Cloud feedbacks in the climate system: A critical review. J. Climate, 18 , 237–273.
Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the A-Train: A new dimension of space-based observations of clouds and precipitation. Bull. Amer. Meteor. Soc., 83 , 1771–1790.
Theisen, C. J., P. A. Kucera, and M. R. Poellot, 2009: A study of relationships between Florida thunderstorm properties and corresponding anvil cloud characteristics. J. Appl. Meteor., 48 , 1882–1901.
TRMM Precipitation Radar Team, 2005: Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar Algorithm instruction manual for version 6. Japan Aerospace Exploration Agency and NASA Rep., 180 pp. [Available online at http://www.eorc.jaxa.jp/TRMM/document/pr_manual/pr_manual_v6.pdf].
Vera, C., and Coauthors, 2006: Toward a unified view of the American monsoon systems. J. Climate, 19 , 4977–5000.
Webster, P. J., 1983: Large-scale structure of the tropical atmosphere. Large-Scale Dynamical Processes in the Atmosphere, B. J. Hoskins and R. P. Pearce, Eds., Academic Press, 235–275.
Webster, P. J., and G. L. Stephens, 1980: Tropical upper-tropospheric extended clouds: Inferences from Winter MONEX. J. Atmos. Sci., 37 , 1521–1541.
Yao, M.-S., and A. D. Del Genio, 1999: Effects of cloud parameterization on the simulation of climate changes in the GISS GCM. J. Climate, 12 , 761–779.
Yuan, J., and R. A. Houze Jr., 2010: Global variability of mesoscale convective system anvil structure from A-Train satellite data. J. Climate, 23 , 5864–5888.
Zender, C. S., and J. T. Kiehl, 1997: Sensitivity of climate simulations to radiative effects of tropical anvil structure. J. Geophys. Res., 102 , 23793–23803.
Zipser, E. J., D. J. Cecil, C. Liu, S. W. Nesbitt, and D. P. Yorty, 2006: Where are the most intense thunderstorms on earth? Bull. Amer. Meteor. Soc., 87 , 1057–1071.
An example image of an anvil associated with deep convection over West Africa observed by the TRMM PR (courtesy of Schumacher and Houze 2006). Definition of ice anvils and mixed anvils are based on Frederick and Schumacher (2008).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
An example image of an anvil associated with deep convection over West Africa observed by the TRMM PR (courtesy of Schumacher and Houze 2006). Definition of ice anvils and mixed anvils are based on Frederick and Schumacher (2008).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
An example image of an anvil associated with deep convection over West Africa observed by the TRMM PR (courtesy of Schumacher and Houze 2006). Definition of ice anvils and mixed anvils are based on Frederick and Schumacher (2008).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Images of a coincident CloudSat and TRMM overpass. (a) TRMM PR horizontal cross sections at (left) 2.0 and (right) 7.5 km for orbit 55 469 with CloudSat track in magenta. Vertical cross sections of (b) CloudSat CPR and (c) TRMM PR, with PR-defined anvils marked as in red along the abscissa, which represents along-track bins. The scan time of the images is around 1923 local time on 10 Aug 2007. CloudSat is about 5 minutes in front of TRMM with the track centered at 19.85°N, 87.93°W. The color bars are reflectivities in dBZ.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Images of a coincident CloudSat and TRMM overpass. (a) TRMM PR horizontal cross sections at (left) 2.0 and (right) 7.5 km for orbit 55 469 with CloudSat track in magenta. Vertical cross sections of (b) CloudSat CPR and (c) TRMM PR, with PR-defined anvils marked as in red along the abscissa, which represents along-track bins. The scan time of the images is around 1923 local time on 10 Aug 2007. CloudSat is about 5 minutes in front of TRMM with the track centered at 19.85°N, 87.93°W. The color bars are reflectivities in dBZ.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Images of a coincident CloudSat and TRMM overpass. (a) TRMM PR horizontal cross sections at (left) 2.0 and (right) 7.5 km for orbit 55 469 with CloudSat track in magenta. Vertical cross sections of (b) CloudSat CPR and (c) TRMM PR, with PR-defined anvils marked as in red along the abscissa, which represents along-track bins. The scan time of the images is around 1923 local time on 10 Aug 2007. CloudSat is about 5 minutes in front of TRMM with the track centered at 19.85°N, 87.93°W. The color bars are reflectivities in dBZ.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
(a) Frequency distribution of how much echo the PR is missing at cloud top and (b) the area factor the PR is missing in the horizontal compared to CloudSat by TRMM PR echo-base height. Echo-top statistics are based on 3345 samples in along-track bins. Horizontal size statistics are based on 229 pairs of coincident PR–CPR images.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
(a) Frequency distribution of how much echo the PR is missing at cloud top and (b) the area factor the PR is missing in the horizontal compared to CloudSat by TRMM PR echo-base height. Echo-top statistics are based on 3345 samples in along-track bins. Horizontal size statistics are based on 229 pairs of coincident PR–CPR images.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
(a) Frequency distribution of how much echo the PR is missing at cloud top and (b) the area factor the PR is missing in the horizontal compared to CloudSat by TRMM PR echo-base height. Echo-top statistics are based on 3345 samples in along-track bins. Horizontal size statistics are based on 229 pairs of coincident PR–CPR images.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR anvil occurrence (i.e., how often an anvil was observed in a 2.5° daily grid box) for 1998–2007. Areas with deep convective rain (i.e., echo tops ≥9 km) occurrences of less than 5% were not included. The same threshold is applied to most contour plots in this paper.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR anvil occurrence (i.e., how often an anvil was observed in a 2.5° daily grid box) for 1998–2007. Areas with deep convective rain (i.e., echo tops ≥9 km) occurrences of less than 5% were not included. The same threshold is applied to most contour plots in this paper.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR anvil occurrence (i.e., how often an anvil was observed in a 2.5° daily grid box) for 1998–2007. Areas with deep convective rain (i.e., echo tops ≥9 km) occurrences of less than 5% were not included. The same threshold is applied to most contour plots in this paper.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Conditional TRMM PR areal coverage (i.e., how much of the 2.5° grid box was covered by anvils when anvils were observed) for (a) anvils, (b) ice anvils, and (c) mixed anvils for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Conditional TRMM PR areal coverage (i.e., how much of the 2.5° grid box was covered by anvils when anvils were observed) for (a) anvils, (b) ice anvils, and (c) mixed anvils for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Conditional TRMM PR areal coverage (i.e., how much of the 2.5° grid box was covered by anvils when anvils were observed) for (a) anvils, (b) ice anvils, and (c) mixed anvils for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Unconditional TRMM PR areal coverage (i.e., how much anvils covered the 2.5° grid box regardless if anvils were present or not) for (a) anvils, (b) ice anvils, and (c) mixed anvils for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Unconditional TRMM PR areal coverage (i.e., how much anvils covered the 2.5° grid box regardless if anvils were present or not) for (a) anvils, (b) ice anvils, and (c) mixed anvils for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Unconditional TRMM PR areal coverage (i.e., how much anvils covered the 2.5° grid box regardless if anvils were present or not) for (a) anvils, (b) ice anvils, and (c) mixed anvils for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR mean 17-dBZ echo tops for (a) anvils, (b) ice anvils, and (c) mixed anvils based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR mean 17-dBZ echo tops for (a) anvils, (b) ice anvils, and (c) mixed anvils based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR mean 17-dBZ echo tops for (a) anvils, (b) ice anvils, and (c) mixed anvils based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR mean anvil thickness based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR mean anvil thickness based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR mean anvil thickness based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Standard deviation of seasonal variability for TRMM PR anvil echo tops based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Standard deviation of seasonal variability for TRMM PR anvil echo tops based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Standard deviation of seasonal variability for TRMM PR anvil echo tops based on 2.5° grid averages for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR (a) anvil and (b) deep convective rain (echo tops > 7 km) occurrence anomalies for El Niño–La Niña events during 1998–2007 (events are listed in Table 2) based on 2.5° grid averages.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR (a) anvil and (b) deep convective rain (echo tops > 7 km) occurrence anomalies for El Niño–La Niña events during 1998–2007 (events are listed in Table 2) based on 2.5° grid averages.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
The TRMM PR (a) anvil and (b) deep convective rain (echo tops > 7 km) occurrence anomalies for El Niño–La Niña events during 1998–2007 (events are listed in Table 2) based on 2.5° grid averages.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Fraction of convective rain occurrence associated with anvil occurrence (i.e., the number of days with both convective rain and anvils divided by the total number of days with convective rain) based on 2.5° grid averages from 1998 to 2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Fraction of convective rain occurrence associated with anvil occurrence (i.e., the number of days with both convective rain and anvils divided by the total number of days with convective rain) based on 2.5° grid averages from 1998 to 2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Fraction of convective rain occurrence associated with anvil occurrence (i.e., the number of days with both convective rain and anvils divided by the total number of days with convective rain) based on 2.5° grid averages from 1998 to 2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Average 17-dBZ echo tops of convective rain based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Average 17-dBZ echo tops of convective rain based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Average 17-dBZ echo tops of convective rain based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distributions of daily grid average convective 17-dBZ echo tops with and without anvils.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distributions of daily grid average convective 17-dBZ echo tops with and without anvils.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distributions of daily grid average convective 17-dBZ echo tops with and without anvils.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Average reflectivity of convective rain at 2 km based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Average reflectivity of convective rain at 2 km based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Average reflectivity of convective rain at 2 km based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distributions of daily grid average convective reflectivity at 2 km with and without anvils.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distributions of daily grid average convective reflectivity at 2 km with and without anvils.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distributions of daily grid average convective reflectivity at 2 km with and without anvils.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Conditional areal coverage of convective rain based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Conditional areal coverage of convective rain based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Conditional areal coverage of convective rain based on 2.5° grid averages for 1998–2007 for (a) convective rain with anvils, (b) convective rain without anvils, and (c) the difference between (a) and (b). Only regions with 95% significance are plotted in (c).
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Probability distribution of multiple regression coefficients for (a),(b) anvil areal coverage and (c),(d) 17-dBZ echo top using convective area, echo top, and reflectivity at (a),(c) 2 km and (b),(d) using factors excluding convective area.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Probability distribution of multiple regression coefficients for (a),(b) anvil areal coverage and (c),(d) 17-dBZ echo top using convective area, echo top, and reflectivity at (a),(c) 2 km and (b),(d) using factors excluding convective area.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Probability distribution of multiple regression coefficients for (a),(b) anvil areal coverage and (c),(d) 17-dBZ echo top using convective area, echo top, and reflectivity at (a),(c) 2 km and (b),(d) using factors excluding convective area.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distribution of the anvil areal coverage as a function of convective rain area (i.e., a 2D histogram of convective rain and anvil area) for (a) Africa (land area in 5°S–15°N, 30°W–30°E), (b) South America (land area of 15°S–10°N, 40°–90°W), and (c) the west Pacific (ocean area within 15°S–10°N, 90°E–180°). Bin size is 0.5% for both convective rain and anvil areas. The contour interval is 0.5%.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distribution of the anvil areal coverage as a function of convective rain area (i.e., a 2D histogram of convective rain and anvil area) for (a) Africa (land area in 5°S–15°N, 30°W–30°E), (b) South America (land area of 15°S–10°N, 40°–90°W), and (c) the west Pacific (ocean area within 15°S–10°N, 90°E–180°). Bin size is 0.5% for both convective rain and anvil areas. The contour interval is 0.5%.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Frequency distribution of the anvil areal coverage as a function of convective rain area (i.e., a 2D histogram of convective rain and anvil area) for (a) Africa (land area in 5°S–15°N, 30°W–30°E), (b) South America (land area of 15°S–10°N, 40°–90°W), and (c) the west Pacific (ocean area within 15°S–10°N, 90°E–180°). Bin size is 0.5% for both convective rain and anvil areas. The contour interval is 0.5%.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
As in Fig. 18, but for stratiform rain area.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
As in Fig. 18, but for stratiform rain area.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
As in Fig. 18, but for stratiform rain area.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
(a) Zonal wind and (b) relative humidity profiles for situations when no stratiform rain (no SF) is present, stratiform rain without anvils (SF no AV) and stratiform rain with anvils (SF+AV) over (left to right) Africa, the west Pacific, and South America based on daily NCEP reanalysis fields for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
(a) Zonal wind and (b) relative humidity profiles for situations when no stratiform rain (no SF) is present, stratiform rain without anvils (SF no AV) and stratiform rain with anvils (SF+AV) over (left to right) Africa, the west Pacific, and South America based on daily NCEP reanalysis fields for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
(a) Zonal wind and (b) relative humidity profiles for situations when no stratiform rain (no SF) is present, stratiform rain without anvils (SF no AV) and stratiform rain with anvils (SF+AV) over (left to right) Africa, the west Pacific, and South America based on daily NCEP reanalysis fields for 1998–2007.
Citation: Journal of Climate 24, 6; 10.1175/2010JCLI3793.1
Anvil vertical extent across the tropics (20°S–20°N, 180°W–180°E). Unit: km.
El Niño–La Niña events selected during 1998–2007.
