1. Introduction
Predicting tropical cyclogenesis remains one of the great forecasting challenges to today’s meteorological community (Emanuel 2005). Much of our limited understanding can likely be attributed to our inability to differentiate the often subtle physical differences between developing and nondeveloping tropical cyclones (TCs), and any such differences, when observed, have been insufficiently documented (Dunkerton et al. 2009). Among the well-known necessary dynamic conditions for tropical cyclogenesis are background cyclonic vorticity, 850–200-hPa tropospheric wind shear of less than 15 m s−1 and preferably below 10 m s−1, and a sufficiently high Coriolis parameter (Gray 1968). Thermodynamic prerequisites exist as well, including sea surface temperatures greater than or equal to 26°C (Palmen 1948), an unstable or conditionally unstable environment, and relatively high moisture content from the surface through 5 km (Gray 1979). However, despite these well-known criteria, the exact sequence of events culminating in tropical cyclogenesis remains unknown.
Two differing views of tropical cyclone formation are the top-down and the bottom-up hypotheses. Ritchie and Holland (1997) and Simpson et al. (1997) describe a top-down mechanism for genesis by which successive mergers of mesoscale convective systems (MCSs) increase the size and/or strength of the midlevel vortex, which induces a surface circulation through vertical penetration and vortex stretching. Similarly, Bister and Emanuel (1997) propose that a stratiform rain region associated with an existing MCS acts to moisten and cool the mid- to lower levels. The level of peak cooling descends within the stratiform rain region, thereby lowering the level of maximum potential vorticity (PV) production, while moistening acts to limit the occurrence of dry downdrafts. Along with the necessity of a strengthening midlevel circulation, Nolan (2007) also found humidification of the inner core due to moist detrainment and precipitation from deep convective towers preceding genesis. However, Nolan (2007) does not necessitate a top-down genesis process. Lastly, a recent study by Raymond et al. (2011) of five tropical cyclogenesis events in the northwestern Pacific suggests that tropical cyclogenesis is facilitated by a preexisting midlevel vortex. This midlevel vortex creates a cold core at low levels, which alters deep convection as to facilitate spinup.
A slightly differing sequence, known as bottom-up genesis, is proposed by Hendricks et al. (2004) and Montgomery et al. (2006), in which individual deep moist convective updrafts or vortical hot towers (VHTs) develop within the tropical wave, amplify preexisting cyclonic vorticity, and gradually consolidate to form a low-level center of circulation. Latent heat released within these VHTs aids in the development of the midlevel warm core, and surface convergence and upper-level divergence commence. Observational evidence supporting a top-down mechanism for genesis is presented by Ritchie and Holland (1997) and Mapes and Houze (1995), while Houze et al. (2009) find evidence that support the VHT argument, all for individual case studies.
Regardless of the exact order of processes by which genesis occurs, the dependence upon some initial MCS or VHTs assumes sufficient tropospheric instability to allow deep convection. Using in situ data, Molinari and Vollaro (2010) find that highly sheared, generally weaker tropical cyclones tend to be associated with higher convective available potential energy (CAPE) than their nonsheared, generally stronger counterparts. Similarly, Braun (2010) found higher CAPE in environments for weakening TCs compared to strengthening TCs in the days following genesis. In idealized numerical simulations, Nolan et al. (2007) found that greater maximum potential intensity (MPI) resulted in greater likelihood of genesis, while greater CAPE did not. Nonetheless, the question of whether genesis becomes increasingly favored with increasing instability, or whether there is some threshold beyond which decreasing stability is detrimental to genesis, has not been conclusively answered via observational evidence.
A recent endeavor to better understand tropical cyclogenesis from a wave-relative framework is undertaken by Dunkerton et al. (2009). Known as the marsupial paradigm, tropical depression formation from a predepression wave trough in the lower troposphere is greatly favored within the critical-layer “pouch”—a region of closed material contours wherein the parent wave’s phase speed equals the mean flow. A young disturbance within the pouch is repeatedly moistened by deep moist convection within the critical layer while remaining somewhat protected from lateral intrusion of dry air and deformation by horizontal and vertical shear. This protovortex, collocated with the critical latitude, is then able to keep pace with the parent wave until it has strengthened into a self-maintaining entity. Hypothetically, the marsupial paradigm could be used in conjunction with either the top-down or bottom-up genesis hypotheses. Dunkerton et al. (2009) assume a bottom-up progression of genesis.
As already alluded to, much of the difficulty in identifying the exact order of processes that occur during genesis, or whether top-down or bottom-up sequences both occur under different conditions, can be attributed to a lack of in situ data prior to genesis. In an attempt to expand upon the limited dataset, several field campaigns have sampled tropical cyclones during and shortly after the genesis stage, including the Tropical Experiment in Mexico (TEXMEX; Bister and Emanuel 1997; Raymond et al. 1998), the Tropical Cloud Systems and Processes (TCSP) experiment in 2005 (Halverson et al. 2007), the National Aeronautics and Space Administration (NASA) component of the African Monsoon Multidisciplinary Analyses (AMMA) project in 2006 (Zipser et al. 2009), the Tropical Cyclone Structure experiment in 2008 (TCS-08; Elsberry and Harr 2008), as well as a handful of observations from the Hurricane Rainband and Intensity Change Experiment (RAINEX) of 2005 (Houze et al. 2006). Case studies using data from these experiments, such as Zipser et al. (2009), emphasize the difficulty of achieving genesis in excessively dry air masses. Ritchie and Holland (1997), Davis et al. (2008), Houze et al. (2009), and Braun et al. (2010) have shown that the progressive strengthening of a midlevel vortex, a gradual moistening of the column in a region of deep convection, and the development of a warm core are all evident in observations of various tropical cyclones during and shortly following genesis. While these studies allude to the development of a warm core, the altitude of the warm-core maxima and the timing of the development of the warm core are generally neglected. Earlier observational studies such as La Seur and Hawkins (1963) and Hawkins and Rubsam (1968) have found maximum warm anomalies at around 250 hPa in mature TCs, while Hawkins and Imbembo (1976) and Stern and Nolan (2012) suggest that the primary warm core is located from 500 to as low as 650 hPa. The level of maximum warm anomalies for pregenesis disturbances remains to be determined.
The most recent of the field campaigns involving genesis, known as Pre-Depression Investigation of Cloud-Systems in the Tropics (PREDICT), was an expansive investigation of tropical cyclogenesis in the central Atlantic during the 2010 season (Montgomery et al. 2012). One of the overarching goals of PREDICT was to gather in situ observations necessary to examine the marsupial theory of genesis. Ultimately, PREDICT has provided the most expansive in situ dataset comprising both developing and nondeveloping tropical waves prior to genesis, complemented by simultaneous observations from the Intensity Forecasting Experiment (IFEX) and Genesis and Rapid Intensification Processes (GRIP) campaigns. During PREDICT, the National Science Foundation (NSF)/National Center for Atmospheric Research (NCAR) Gulfstream-V (G-V) aircraft provided once- and, occasionally, twice-daily sampling of tropical waves for up to five consecutive days. Unlike in previous field campaigns, by incorporating the definition of the critical layer, it was possible to target a potential genesis region much smaller than an entire tropical wave. In total, 25 flight missions were completed between 15 August and 30 September 2010 (plus 1 calibration flight). The strength of this dataset comes not only from the unprecedented quantity of developing and nondeveloping tropical waves sampled, but also from the temporally evolving nature of data associated with distinct pouches.
Two recent papers, Smith and Montgomery (2011) and Davis and Ahijevych (2012), examined PREDICT dropsonde data for Tropical Storm (TS) Matthew, Hurricane Karl, and ex-TS Gaston. Smith and Montgomery (2011) found lower values of equivalent potential temperature between the surface and 3 km of nondeveloping ex-Gaston than in developing pre-Karl and pre-Matthew. The authors found evidence that dry air for the nondeveloping case was not necessarily associated with stronger downdrafts but rather that the drier midlevel air weakened the convective updrafts and thereby prevented sufficient amplification of system relative vorticity necessary for development. Last, greater CAPE and convective inhibition (CIN) were associated with ex-Gaston than either genesis event. Davis and Ahijevych (2012) found that a misalignment of the mid- and low-level circulation centers, due to vertical shear, made TS Gaston more susceptible to intrusion of dry air. They found that Karl and Matthew developed in a moister environment, with mid- to upper-level moisture increasing with time. An initial vertical misalignment of the vortex delayed genesis of Karl until the vortex could subsequently realign.
This study differs from previous studies in that multicase composite vertical profiles of temperature, moisture, and wind for genesis and nongenesis cases are compared. Additionally, the time evolution of the mean genesis profile is examined. Parcel-based metrics such as lifted condensation levels (LCLs), CAPE, and CIN are also included. Identification of the vertical level, timing, and magnitude of warm core development, if at all discernible prior to genesis, will be sought. The data will be evaluated to determine whether a progressive increase in moisture to near saturation or a top-down versus bottom-up transition of the mean vortex are observed. An investigation of whether cases of genesis are associated with greater instability is presented.
An overview of the PREDICT dropsonde dataset and the methodologies and analysis techniques used in this study appear in section 2. In section 3, comparisons of anomalous values of a variety of metrics with respect to the PREDICT mean sounding for genesis and nongenesis profiles are made. Conclusions will appear in section 4.
2. Data and methods
During PREDICT, 547 dropsondes were deployed over the course of 25 aircraft missions investigating tropical waves in the Caribbean and western Atlantic (Fig. 1). Five cases of genesis, three cases of nongenesis, and four TCs named during or prior to investigation (TC stage) constitute the PREDICT dataset (Table 1). The TC stage category overlaps with three of the five genesis cases: TSs Fiona, Matthew, and Nicole, in addition to an investigation of TS Gaston prior to weakening to a remnant low. Cases are sorted by genesis or nongenesis based upon whether or not the tropical wave under investigation eventually yields a tropical storm/depression as declared by the National Hurricane Center (NHC). The genesis category is further separated temporally into missions that occur 0–24 h pregenesis, 24–48 h pregenesis, 48–72 h pregenesis, and >72 h pregenesis. Missions that begin during one such time period and end in another are assigned to the period during which the majority of dropsondes are deployed. As such, the 0–24-h pregenesis category includes data from the 30 August flight into pre-TS Fiona, the 14 September flight into pre-Hurricane Karl, and the 27 September flight into pre-TS Nicole. The >72-h pregenesis category includes data from the 10 and 11 September flights into pre-TS Karl, the 20 September flight into pre-TS Matthew, and the 30 September flight into pre-Hurricane Otto. Tropical Storm Gaston is a special case. The first mission into Gaston occurred at a time when the storm was already a named system, and therefore data from this mission is assigned to the TC category. However, subsequent PREDICT flights occurred after Gaston was downgraded by the NHC to a remnant cyclone. Since Gaston was no longer a named system during these missions, and was given a 70% chance to redevelop by NHC but failed to do so, missions into the remnants of Gaston are added to the nongenesis category.
Cases of study during PREDICT comprising the genesis, nongenesis, and TC-stage groups with corresponding dates of G-IV deployments. Timing of mission prior to genesis is included for cases in which genesis occurred. Latitude and longitude of the target center location of each drop pattern, zonal phase speed Up of the wave, and the number of dropsondes released are shown.
Vortex-relative tangential (Vtan) and radial (Vrad) components of wind are also calculated. Since the tropical waves are moving within some background flow, and winds from the dropsonde data are earth relative, it is necessary to remove the parent wave’s zonal phase speed (Up in vortex-relative wind calculations). Zonal phase speed is calculated from a consensus of National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) and European Centre for Medium-Range Weather Forecasts (ECMWF) analyses and forecasts of meridional wind υ and RH (Table 1). The values used in this study were determined operationally during PREDICT (available online at http://met.nps.edu/~mtmontgo/storms2010.html) using Hovmöller diagrams to track features in these fields. Since meridional phase speed is generally much weaker than Up for westward-moving tropical waves and was not calculated operationally during PREDICT, it is not included here. As such, the mean Vtan profile is then computed as the sum of the cyclonic (positive) and anticyclonic (negative) contribution of each dropsonde with Up removed, normalized by the total number of dropsondes. For Vrad, the component of wind from each dropsonde blowing away from (toward) the center of circulation contributes positively (negatively).
Computation of Vtan and Vrad requires selection of a center of circulation, which is chosen to be the point at which mean 850–700-hPa Vtan is maximized for each flight. The methodology follows Marks et al. (1992) with one exception: since the radius of maximum winds (RMW) is poorly defined for many cases in this study, mean Vtan will be computed with respect to all dropsonde locations, rather than only those within an annulus around the RMW. Computation of Vtan is performed in
Vertical wind shear is calculated as the vector wind difference between the 850–500- and 850–200-hPa pressure levels. Prior to computing the total wind shear vector, an average within each quadrant relative to the circulation center is performed in order to eliminate any false signal associated with asymmetries in the drop pattern. Quadrant averages are then averaged together. However, vertical asymmetries of the vortex itself are not removed from the shear calculation.
Vertical profiles will be depicted as an anomaly with respect to the PREDICT mean profile. Using this framework, it will be possible to investigate and quantify the differences between the vertical profiles of developing and nondeveloping tropical systems. While there are certainly large-scale synoptic, mesoscale, and convective-scale differences between the two scenarios that cannot be captured in a mean dropsonde profile, a number of significant results can nonetheless be drawn.
Midlevel moisture can often vary immensely over relatively small distances over the spatial extent of a tropical wave because of a number of factors, including advection of moist or dry air, drying associated with subsidence, convective moistening from detrainment and precipitation processes, or drying associated with dry downdrafts. Much of both the top-down and bottom-up literature note a general trend of increasing convection near the center of the cyclone. However, averaging over the full areal extent of any one case might lead to a net cancellation of numerous moistening and drying processes, masking mesoscale variability. Therefore, in addition to domainwide averaging, profiles of q and RH from dropsondes located within a 150-km radius of the approximate center of circulation will be composited in order to investigate localized moisture anomalies. This will only be performed for genesis cases, and of these, only for cases in which the center is well defined. This limited dataset includes 8–12 dropsondes for each genesis time block (0–24 h pregenesis, 24–48 h, etc.) consisting of data from the pre-Fiona, pre-Karl, pre-Matthew, pre-Nicole, and pre-Otto missions.
Last, Geostationary Operational Environmental Satellite (GOES) infrared data are investigated in order to relate any dynamic and thermodynamic phenomena observed in the dropsonde data to the convective structure of the tropical waves. GOES cloud-top imagery in full 30-min resolution is composited over 6-h time windows centered temporally on the mean time of each dropsonde mission to the nearest half hour, and then further composited over all cases in each genesis category. While accurate estimates of static center locations were determined from dropsonde data alone, model data was preferred when attempting to locate the time-evolving center location. As such, satellite composites are centered geographically on the center of the pouch, determined by the intersection of the disturbance critical line with the axis of the wave trough, from a consensus of GFS and ECMWF analyses interpolated linearly between analysis times. Similar to Davis and Ahijevych (2012), convective activity is depicted as a fraction of the total time comprising each category within each 6-h period, using half-hourly data in the comoving frame of reference, that a grid box 10 km on a side exhibits an IR temperature less than −50°C.
It should be emphasized that the results presented herein are only valid for tropical systems in the western Atlantic basin during the 2010 hurricane season. Whether or not these results can be generalized is yet to be seen.
3. Analysis of dropsonde data
a. Genesis versus nongenesis
We begin our investigation by comparing T profiles of genesis and nongenesis cases with the PREDICT mean. Nongenesis cases are associated with slight warm anomalies of 0.1° to 0.2°C below 600 hPa, while the genesis mean is associated with cold anomalies of −0.2° to −0.3°C (Fig. 3a). Above 600 hPa, T anomalies steadily decrease with height in the nongenesis profile, while the genesis profile is instead associated with warm anomalies. While of a much lower magnitude, 0.2° versus 0.8°C, the greatest positive T anomalies for genesis appear within the same layer as for the TC stage profile between 400 and 200 hPa. It is possible that at least part of the warming associated with the genesis profile is the beginnings of the development of the warm core. Radial composites reveal the greatest warm anomalies occurring within a 200-km radius of the center of circulation for genesis and TC samples, supporting this hypothesis (the development of which is shown in section 3b). Alternatively, the nongenesis mean profile is associated with cold anomalies of −0.1° to −0.9°C from 600–200 hPa. It should be noted that standard deviations of these data are large, and only cold anomalies for nongenesis above 500 hPa are greater than one standard deviation from the PREDICT mean.
Maximum anomalies of q, both positive and negative, exist between 700 and 500 hPa for genesis, nongenesis, and TC stage cases (Fig. 3b). TC and pregenesis profiles are associated with moist q anomalies of +0.1 to +0.5 g kg−1 while nondeveloping systems are associated with dry q anomalies of −0.1 to −1.0 g kg−1 in this layer. While these values appear to be small, the nongenesis profile is 25% drier than the PREDICT mean q of 3.24 g kg−1 with a −0.81 g kg−1 anomaly at 500 hPa. Radial composites reveal dry anomalies as large as −1.8 g kg−1 at greater than 400-km radius for nongenesis, with nongenesis drier than genesis through the center of circulation, suggesting an influence of the dry air on the core of the tropical wave. Interestingly, profiles of q indicate that the nongenesis mean value below 850 hPa is more than 0.5 g kg−1 moister than the genesis mean value, suggesting that perhaps dry air at the midlevels was more detrimental to genesis than drier air at the surface during PREDICT. The PREDICT mean is greater than one standard deviation moister than the nongenesis mean from 700 to 500 hPa, suggestive of the relative significance of this dry air (Table 2).
PREDICT, genesis, nongenesis, and TC-stage mean mixing ratio values, standard deviations σ, and anomaly vs the PREDICT mean for select levels from 1000 to 200 hPa.
While comparing q profiles allows for direct comparisons of the mass of water vapor in a column of atmosphere, examination of RH (Fig. 3c) is necessary to identify near saturation of the lower and midlevels—a significant criterion for genesis in Bister and Emanuel (1997) and Nolan (2007). The presence of low ambient RH also suggests that the developing tropical wave is vulnerable to the potential detrimental effects of entrainment on parcel buoyancy. Consistent with q results, RH is notably lower by 10%–20% from 700 to 300 hPa for nongenesis than for genesis, indicating greater potential for entrainment of drier air.
Values of CAPE reveal an interesting result, in that not only are the pregenesis and TC profiles no more unstable than the nongenesis sounding, but the nongenesis profile is in fact much more unstable than the genesis profile (Table 3). Mean CAPE anomalies are +336 J kg−1 for nongenesis, −171 J kg−1 for genesis, and −42 J kg−1 for TC cases. All departures are relative to the PREDICT mean CAPE of 2096 J kg−1 when integrated to the height of the EL, or the highest available pressure level. Large CAPE values of 1900–2500 J kg−1, in conjunction with low LFCs of 940–920 hPa and very low CIN of −2 to −10 J kg−1 are not unexpected given that calculations are with respect to particularly moist surface-based parcels. In fact, many genesis and nongenesis dropsondes exhibited no CIN at all. While it is true that the height at which the dropsonde is deployed may be below the EL in some cases,
Instability data for different categories. Included are the LFC, EL, CAPE, CIN, standard deviations of CAPE, and CAPE anomaly vs the PREDICT mean.
Average Vtan values are 2–3.5 m s−1 for nongenesis and 3–5 m s−1 for genesis profiles from the surface through 500 hPa (Fig. 3d). While significant variability exists, with standard deviations greater than 2 m s−1, the developing tropical waves are generally associated with stronger circulations. As expected, these circulations are still significantly weaker than those of TC stage systems. For genesis, nongenesis, and TC cases, Vtan is cyclonic from the surface through 300 hPa and anticyclonic above.
Radial wind Vrad is also calculated (not shown). Maximum negative values of up to −1 m s−1 for TC stage and −0.3 m s−1 for genesis and nongenesis occur between the surface and 900 hPa, suggesting a shallow layer of inflow and boundary layer convergence. While these values appear to be small, they are averages over the entire sampled region of the tropical wave, and do not correspond to maxima at a particular radius. All three profiles are associated with positive Vrad above 300 hPa, indicative of divergence. Time-averaged genesis and nongenesis Vrad profiles are generally indistinguishable, and Vrad does not appear to be of much value as a discriminating characteristic for genesis. However, as will be demonstrated in section 3b, the genesis Vrad profile evolves considerably with time.
Midlevel, 850–500-hPa vertical wind shear is slightly greater for genesis than nongenesis with 2.70 m s−1 shear plus or minus a standard deviation of 0.22 m s−1 as compared to 2.22 ± 0.64 m s−1 for nongenesis (Table 4). Alternatively, 850–200-hPa deep-layer shear is slightly more hostile for the nongenesis than the genesis cases, with 7.39 ± 0.71 m s−1 compared to 6.97 ± 0.40 m s−1. However, wind shear did not appear to be the primary factor in differentiating genesis from nongenesis cases in 2010 as values of wind shear are statistically identical between the two cases. Given the high percentage of dropsondes deployed within 400 km of the center of circulation and comparatively few outside of this region, it is also possible that these data are not fully representative of the true environmental wind shear.
Vertical wind shear values and standard deviations for 850–500 and 850–200 hPa for genesis, nongenesis, and TC stage categories. Wind shear data for >72 h pregenesis, 48–72 h pregenesis, 24–48 h pregenesis, and 0–24 h pregenesis subsets also included.
GOES composites clearly demonstrate persistently colder cloud tops over a much larger area for genesis than nongenesis (Fig. 4). In fact, the area of cloud tops colder than −50°C more than 60% of the time is larger for genesis than our sample of TCs. Genesis is clearly associated with more consistent deep convection over a larger spatial area than nongenesis, which, in turn, would promote the development of a sustained midlevel vortex. Greater spatial coverage of deep convection is also consistent with the building of upper-level warm anomalies as a result of additional latent heating, as well as humidification of the inner core due to moist detrainment and precipitation from deep convective towers preceding genesis.
b. Time progression leading up to genesis
A major benefit of the pouch-tracking framework developed for the PREDICT field campaign was an ability to routinely identify and sample regions of potential genesis daily beginning a few days prior to the development of a tropical depression. This tracking and sampling technique yielded an impressive temporal genesis dataset, in which it is possible to subcategorize genesis profiles by time leading up to genesis. In this section, we will continue to examine differences between various mean profiles and the PREDICT mean, but from the perspective of a temporal progression.
Examination of radial profiles of T reveals widespread cold anomalies at large lead times for all radii gradually transitioning to warm anomalies at small lead times through much of the troposphere (Figs. 5a–d). The >72-h pregenesis profile is associated with negative T anomalies ranging from −0.1° to −1.0°C, with local minima between 800–600 and 400–200 hPa. Conversely, the entire 0–24-h pregenesis profile is associated with deep warm anomalies ranging from +0.1°C at 1000 hPa to as large as +2.0°C at 300 hPa within 100 km of the center of circulation. The most obvious warming trend occurs from 400 to 200 hPa, evident during every successive 24-h time increment. However, cold anomalies persist from 900 to 700 hPa through 24–48 h pregenesis, until finally warming rapidly 0–24 h pregenesis. Warm anomalies are observed from 500 to 200 hPa above cold anomalies below 500 hPa from 24–48 and 48–72 h pregenesis, characteristic of the development of a low-level cold core prior to genesis as described by Bister and Emanuel (1997) and Nolan (2007). The observed upper-level warming can likely be attributed to a combination of latent heat released by convection and warm-core development via subsidence induced by this latent heating. The predominant warm anomalies 0–24 h pregenesis are maximized near 300 hPa, at a similar altitude to what was found in La Seur and Hawkins (1963) and Hawkins and Rubsam (1968). These results suggest that warm core development commences prior to genesis and at the same altitude one would expect to find the warm core in a mature TC. There even exists some hint of a secondary T maximum developing between 500 and 600 hPa within 200 km of the center of circulation, as suggested by Hawkins and Imbembo (1976) and Stern and Nolan (2012), although much weaker than the primary warm core.
In contrast with temperature, neither q nor RH increases on average with time. Mixing ratio differences between the >72-h pregenesis and 0–24-h pregenesis cases tend to be small, demonstrating low variability with no evident large-scale humidification. However, the full wave mean for cyclogenesis events during PREDICT is simply more moist than for nongenesis events to begin with, even four or more days in advance. In a separate calculation, q and RH are composited for only a small subset of the total sample, only including 8–12 dropsondes for each 24-h period within a 150-km radius of the center of circulation. Local averaging reveals a moistening trend (Fig. 6a), with q increasing from +0.5 to +0.7 g kg−1 at >72 h pregenesis to +1.2 to +1.6 g kg−1 0–24 h pregenesis between 800 and 500 hPa, with the greatest increase in moisture occurring 24–48 h pregenesis. Radial cross sections of q azimuthally averaged in annuli reveal similar trends (not shown). This observation is consistent with Nolan (2007) in that moist detrainment and precipitation from deep convective towers are acting locally to humidify the center of the wave, although the greater low-level moistening observed here suggests a greater fraction of moist detrainment associated with cumulus congestus type clouds. The fact that this trend does not appear in the full composite may simply be reflective of the larger area whose profile is difficult to modify given the small area occupied by convection. This result is also consistent with pregenesis moistening local to the inner 60 km above 2 km observed in the numerical experiments of Bister and Emanuel (1997). A threshold of 60-km radius is not used here, as decreasing the radius by 40 km reduces the number of dropsondes encompassed by about 75%, which leads to a very small sample size. While Bister and Emanuel (1997) and Nolan (2007) suggest that a similar local moistening trend should be evident in the RH field, one was not observed in the limited dataset of this study (Fig. 6b). Instead, a very small increase RH from 800 to 600 hPa is observed 24–48 h pregenesis, followed by a decrease in RH 0–24 h pregenesis. Overall, the added moisture content nearly exactly cancels with the increased saturation deficit associated with warming temperatures such that the profile of RH remains relatively steady.
A time progression of CAPE reveals maximum instability >72 h pregenesis (Table 3), at a time when upper-level temperatures are the coldest. A sudden decrease in CAPE by approximately 500 J kg−1 occurs 48–72 h pregenesis as upper levels warm. The mean sounding becomes slightly more unstable again 24–48 h pregenesis as the boundary layer warms and moistens rapidly enough to offset continued warming at the upper levels, followed by a small decrease in CAPE again 0–24 h pregenesis.
Midlevel, 850–500-hPa wind shear gradually subsides from 2.82 m s−1 >72 h pregenesis to 2.45 m s−1 0–24 h pregenesis (Table 4), although both values are associated with very low shear and are highly favorable for tropical cyclogenesis. Upper-level, 850–200-hPa wind shear does not depict a coherent signal, fluctuating between 6.8 and 7.2 m s−1 between pregenesis time bins. However, as was the case for the comparison between genesis and nongenesis shear values, these values are statistically indistinguishable.
Profiles of Vrad indicate that mean 300–200-hPa upper-level outflow fluctuates significantly with time, but with no obvious trend (Fig. 6c). It would appear that level of organization of divergent flow aloft is not critical to any particular stage of genesis. Radial winds of +1.5 m s−1 at 200 hPa are observed both >72 h pregenesis and at the time of genesis, indicating that upper-level conditions were favorable for genesis several days before it occurred, but the tropical waves required additional time to organize. At lower levels, positive values of Vrad of +0.5 to +1 m s−1 are observed below 850 hPa for both the >72-h pregenesis and the 48–72-h pregenesis composites, indicating surface divergence. This suggests that, while upper-level outflow was favorable for genesis, lack of surface convergence may have hindered the process. While not evident in these composites, Davis and Ahijevych (2012) suggested that an initial vertical misalignment of vortex centers may have also delayed genesis of Karl. By 24–48 h pregenesis, radial outflow reverses to −1 m s−1 inflow, increasing to −2 m s−1 0–24 h pregenesis. While it is likely that some of the large-scale surface divergence 48–72 h pregenesis is associated with convective or mesoscale downdrafts, there does not appear to be sufficient dry air near the core of the system at any time for dry air induced downdrafts to be a prohibitive factor for genesis (Fig. 6b). Thereafter, a secondary circulation begins to develop within 48 h of genesis, and surface convergence increases rapidly with time.
Tangential winds (Fig. 6d) progress from weaker to stronger and more cyclonic with time above 600 hPa. This result is consistent with the progressive building of a mid- to low-level vortex described by Nolan (2007) and Raymond et al. (2011). However, the region of maximum Vtan exists between 850–550 hPa, depending upon the timeframe examined, and there is no well-defined peak. The altitude of the tangential wind maximum is of lower altitude than the warm core, consistent with La Seur and Hawkins (1963), Hawkins and Rubsam (1968), Hawkins and Imbembo (1976), and Stern and Nolan (2012). While it is difficult to determine an exact threshold, genesis appears imminent when systemwide deep-layer positive tangential wind anomalies of 6–7 m s−1 develop between 850 and 700 hPa, with much weaker values observed 24 or more hours earlier. Below 700 hPa, the strength of the mean tangential wind actually fluctuates from 72–24 h pregenesis, followed thereafter by an abrupt increase 24 h pregenesis. This is consistent with Nolan (2007) in that the transition to the intensification stage can be sharp, having less to do with a continuous strengthening of the mean wind and perhaps more to do with moistening of the core. These results show a 24-h lag between the greatest increase in moisture 24–48 h pregenesis and a strengthening of the vortex 0–24 h pregenesis. It is particularly striking that this sudden increase in organization is evident in the mean profile of five sampled systems.
While average tangential wind profiles depict a temporal progression of the strengthening vortex, they do not clearly demonstrate a top-down or bottom-up genesis evolution. To further investigate this issue, the time progression of relative vorticity is computed in radial coordinates relative to the center of circulation (Figs. 7a–d) from tangential wind data binned in annuli of 100-km radius. As expected, the greatest values of ζ are observed within r < 100 km near the center of circulation at all times. The initial altitude of the ζ maximum is not well defined >72+ h pregenesis, with relatively weak local maxima between the surface and 700 hPa (Fig. 7a). By 48–72 h pregenesis, vorticity has amplified considerably through a deep layer from 925 to 500 hPa, with the greatest amplification between 900 and 600 hPa (Fig. 7b). Vorticity remains only marginally cyclonic at the surface, likely because of a combination of previously demonstrated low-level divergence occurring in a region of cold anomalies and surface friction. Relative vorticity continues to strengthen at all levels below 400 hPa within 24–48 and 0–24 h pregenesis, although what was previously a broad region of maximum ζ from 900–600 hPa has evolved to become a distinct maximum slightly near 800 hPa (Figs. 7c,d). While we find no evidence of vorticity descending, a clear intensification of midlevel vorticity prior to the development of a robust surface circulation is more consistent with a Bister and Emanuel (1997), Nolan (2007), and Raymond et al. (2011) genesis framework.
The satellite presentation of the temporal progression leading up to genesis is not as intuitive as the satellite comparison between genesis and nongenesis. From >72 h pregenesis to 48–72 h pregenesis, there is a significant expansion of the persistent cloud tops colder than −50°C, indicating greater coverage of deep convection and presumably a more mature system approaching genesis (Figs. 8a,b). However, during subsequent 24–48- and 0–24-h pregenesis time increments, the coverage of deep convection diminishes considerably, both spatially and temporally (Figs. 8c,d). Despite the fact that significant changes are ongoing within the pregenesis vortex, including the development of a warm core, moistening of the core, and amplification of system relative vorticity, the organization and persistence of temporally averaged −50°C cloud tops is no more indicative of genesis 0–24 h pregenesis than it is >72 h pregenesis. This result demonstrates the obvious advantage of having available in situ data when assessing how close a tropical disturbance is to developing into a tropical cyclone.
4. Conclusions
Observations from the 2010 PREDICT field campaign, when analyzed from a composite mean framework, offer discernible differences between developing and nondeveloping tropical waves that may be advantageous to the understanding and prediction of tropical cyclogenesis. Temperature, mixing ratio, relative humidity, radial and tangential components of wind, relative vorticity, and CAPE are examined.
Temperature profiles reveal a progressive building of warm anomalies from 500 to 200 hPa, relative to the PREDICT mean, of +0.5° to +1.0°C at 24–48 h pregenesis, increasing to +1.0 to +2.0°C 0–24 h pregenesis within 200 km of the center of circulation. While the existence of a warm core in mature TCs has been well established in previous literature, the magnitude and timing of the warm-core development with respect to time of genesis has not. The observation of maximum warm anomalies just below tropopause-level pregenesis suggests that warm-core development occurs at the same altitude as observed in mature TCs by La Seur and Hawkins (1963) and Hawkins and Rubsam (1968). A local maximum in warm anomalies below 500 hPa also suggests that formation of a secondary warm core is possible pregenesis, at a similar altitude as the Hawkins and Imbembo (1976) secondary warm core. This is also consistent with the level of the Stern and Nolan (2012) warm core, although it is not the primary warm core as they suggest. It should be noted that any presence of a secondary warm core is much weaker than the primary warm core found at higher altitude. In contrast with the genesis cases, negative T anomalies of −0.5° to −1.0°C exist from 500 to 200 hPa for nondeveloping systems.
In terms of moisture, positive q anomalies of +0.1 to +0.5 g kg−1 from 800 to 300 hPa are observed in developing systems, even 72 or more hours pregenesis. Moisture does not increase significantly with time on the spatial scale of the entire tropical wave. Meanwhile, nondeveloping systems are associated with significant dry anomalies from 800 to 300 hPa. When only examining dropsondes located within 150 km from the center of circulation, moist convective processes appear to increase moisture as the tropical wave approaches genesis, as suggested by Bister and Emanuel (1997), Nolan (2007), and others. The maximum increase in moisture of 1 g kg−1 from 800 to 600 hPa occurs 24–48 h pregenesis. This trend is likely washed out when all dropsondes are included because of the large spatial area of averaging in the full composite, possibly coupled with some large-scale entrainment of dry air into the wave circulation. Nonetheless, the full q composite still demonstrates that time-evolving genesis profiles are all significantly moister than nondeveloping systems, even more than 72 h prior to genesis. Nongenesis RH profiles are on the order of 10%–20% drier than the PREDICT mean from 700 to 500 hPa, suggesting a greater potential for dry air entrainment into convective towers. Conversely, the nongenesis mean is actually moister than the genesis mean from the surface through 850 hPa, possibly suggesting that dry air at the midlevels is more detrimental to genesis than dry air at the low levels.
Examination of the wind field reveals a progressive strengthening of the vortex above 600 hPa, with an initial delay in intensification from 850 to 700 hPa. Tangential wind at these levels fluctuates between 3 and 5 m s−1 from 72 through 24 h pregenesis, before jumping suddenly to 6–7 m s−1 less than 24 h pregenesis. This sudden intensification of the vortex appears to lag the greatest increase in moisture by 24 h. Radial wind profiles suggest that many cases of genesis may have been delayed by low-level outflow. Alternatively, an initial stage of low-level outflow may instead be a necessary first step prior to genesis, induced by cool sinking air in the stratiform precipitation region as in Bister and Emanuel (1997). During the final 48 h before genesis, low-level inflow of 1–2 m s−1 develops and strengthens with time.
Vorticity fields reveal a broad region of maximum ζ from 900 to 600 hPa at 48–72 and 24–48 h pregenesis. This feature appears simultaneously with low-level cold anomalies 24–72 h pregenesis, as well as divergence 48–72 h pregenesis. Thereafter, a distinct ζ maximum develops near 800 hPa 0–24 h pregenesis. This evolution of the vortex is consistent with a process described by Bister and Emanuel (1997), in which the level of maximum PV production descends as the level of peak cooling descends in the stratiform rain region. Ritchie and Holland (1997) and Simpson et al. (1997) propose a similar mechanism by which a midlevel vortex induces a surface circulation through vertical penetration and vortex stretching. While vertically descending vorticity was not identified, the development of a robust midlevel vortex prior to the intensification of a surface vortex was more consistent with these studies, along with Nolan (2007) and Raymond et al. (2011), than with studies that suggest a bottom-up genesis mechanism. Differences between genesis and nongenesis Vtan and ζ also reveal that developing waves are, on average, associated with a slightly stronger circulation than nondeveloping waves.
Genesis cases are associated with slightly greater 850–500-hPa midlevel wind shear than nongenesis cases: 2.70 versus 2.22 m s−1. On the other hand, genesis cases are associated with slightly lower 850–200-hPa deep-layer wind shear: 6.97 compared to 7.39 m s−1. Midlevel wind shear also gradually decreases with time from 2.82 to 2.45 m s−1 during the time progression toward genesis. However, in both cases, differences in wind shear are statistically indistinguishable.
Last, calculation of virtual CAPE indicates significantly greater instability associated with nongenesis profiles than with either pregenesis or TC-stage profiles. Results suggest that 2000 J kg−1 of CAPE may be sufficient for tropical cyclogenesis, and additional instability does not aid in the genesis process.
While other recent studies have examined and compared individual cases sampled during PREDICT, we have presented an alternative perspective in comparing genesis to nongenesis cases via creating composite vertical profiles for all sampled tropical waves, as well as examining the day-to-day evolution of multicase pregenesis composites. Further in-depth investigation of tropical waves with new aircraft data and corroboration with model analyses could potentially increase the robustness of these results.
Acknowledgments
The author acknowledges funding from the National Science Foundation, Grant ATM-0848753. The author is grateful to professors Sharan Majumdar, Brian Mapes, and David Nolan for their comments and advice throughout the study. The author would also like to thank Chris Davis for his insight and suggestions, as well as Dave Ahijevych for providing the GOES composites.
REFERENCES
Bister, M., and K. A. Emanuel, 1997: The genesis of Hurricane Guillermo: TEXMEX analyses and a modeling study. Mon. Wea. Rev., 125, 2662–2682.
Braun, S. A., 2010: Reevaluating the role of the Saharan air layer in Atlantic tropical cyclogenesis and evolution. Mon. Wea. Rev., 138, 2007–2037.
Braun, S. A., M. T. Montgomery, K. J. Mallen, and P. D. Reasor, 2010: Simulation and interpretation of the genesis of Tropical Storm Gert (2005) as part of the NASA Tropical Cloud Systems and Processes Experiment. J. Atmos. Sci., 67, 999–1025.
Davis, C. A., and D. A. Ahijevych, 2012: Mesoscale structural evolution of three tropical weather systems observed during PREDICT. J. Atmos. Sci., 69, 1284–1305.
Davis, C. A., C. Snyder, and A. Didlake, 2008: A vortex-based perspective of eastern Pacific tropical cyclone formation. Mon. Wea. Rev., 136, 2461–2477.
Doswell, C. A., and E. N. Rasmussen, 1994: The effect of neglecting the virtual temperature correction on CAPE calculations. Wea. Forecasting, 9, 625–629.
Dunkerton, T. J., M. T. Montgomery, and Z. Wang, 2009: Tropical cyclogenesis in a tropical wave critical layer: Easterly waves. Atmos. Chem. Phys., 9, 5587–5646.
Elsberry, R. L., and P. A. Harr, 2008: Tropical Cyclone Structure (TCS08) field experiment science basis, observational platforms, and strategy. Asia-Pac. J. Atmos. Sci., 44, 209–231.
Emanuel, K. A., 2005: Divine Wind: The History and Science of Hurricanes. Oxford University Press, 285 pp.
Gray, W. M., 1968: Global view of the origin of tropical disturbances and storms. Mon. Wea. Rev., 96, 669–700.
Gray, W. M., 1979: Hurricanes: Their formation, structure, and likely role in the tropical circulation. Meteorology over Tropical Oceans, D. B. Shaw, Ed., Royal Meteorological Society, 155–218.
Halverson, J., and Coauthors, 2007: NASA’s Tropical Cloud Systems and Processes Experiment: Investigating tropical cyclogenesis and hurricane intensity change. Bull. Amer. Meteor. Soc., 88, 867–882.
Hawkins, H. F., and D. T. Rubsam, 1968: Hurricane Hilda, 1964. II. Structure and budgets of the hurricane on October 1, 1964. Mon. Wea. Rev., 96, 617–636.
Hawkins, H. F., and S. M. Imbembo, 1976: The structure of a small, intense hurricane—Inez 1966. Mon. Wea. Rev., 104, 418–442.
Hendricks, E. A., M. T. Montgomery, and C. A. Davis, 2004: The role of “vortical” hot towers in the formation of Tropical Cyclone Diana (1984). J. Atmos. Sci., 61, 1209–1232.
Houze, R. A., Jr., and Coauthors, 2006: The Hurricane Rainband and Intensity Change Experiment: Observations and modeling of Hurricanes Katrina, Ophelia, and Rita. Bull. Amer. Meteor. Soc., 87, 1503–1521.
Houze, R. A., Jr., W.-C. Lee, and M. M. Bell, 2009: Convective contribution to the genesis of Hurricane Ophelia (2005). Mon. Wea. Rev., 137, 2778–2800.
La Seur, N. E., and H. F. Hawkins, 1963: An analysis of Hurricane Cleo (1958) based on data from research reconnaissance aircraft. Mon. Wea. Rev., 91, 694–709.
Mapes, B. E., and R. A. Houze Jr., 1995: Diabatic divergence profiles in western Pacific mesoscale convective systems. J. Atmos. Sci., 52, 1807–1828.
Marks, F. D., R. A. Houze, and J. F. Gamache, 1992: Dual-aircraft investigation of the inner core of Hurricane Norbert. Part I: Kinematic structure. J. Atmos. Sci., 49, 919–942.
Molinari, J., and D. Vollaro, 2010: Distribution of helicity, CAPE, and shear in tropical cyclones. J. Atmos. Sci., 67, 274–284.
Montgomery, M. T., M. E. Nicholls, T. A. Cram, and A. Saunders, 2006: A “vortical” hot tower route to tropical cyclogenesis. J. Atmos. Sci., 63, 355–386.
Montgomery, M. T., and Coauthors, 2012: The Pre-Depression Investigation of Cloud-Systems in the Tropics (PREDICT) experiment: Scientific basis, new analysis tools, and some first results. Bull. Amer. Meteor. Soc., 93, 153–172.
Nolan, D. S., 2007: What is the trigger for tropical cyclogenesis? Aust. Meteor. Mag., 56, 241–266.
Nolan, D. S., E. D. Rappin, and K. A. Emanuel, 2007: Tropical cyclogenesis sensitivity to environmental parameters in environments of radiative–convective equilibrium. Quart. J. Roy. Meteor. Soc., 133, 2085–2107.
Palmen, E. H., 1948: On the formation and structure of tropical cyclones. Geophysica, 3, 26–38.
Raymond, D. D. J., C. López-Carillo, and L. López Cavazos, 1998: Case-studies of developing east Pacific easterly waves. Quart. J. Roy. Meteor. Soc.,124, 2005–2034.
Raymond, D. D. J., S. L. Sessions, and C. López-Carillo, 2011: Thermodynamics of tropical cyclogenesis in the northwest Pacific. J. Geophys. Res., 116, D18101, doi:10.1029/2011JD015624.
Ritchie, E. A., and G. J. Holland, 1997: Scale interactions during the formation of Typhoon Irving. Mon. Wea. Rev., 125, 1377–1396.
Simpson, J., E. A. Ritchie, G. J. Holland, J. Halverson, and S. Stewart, 1997: Mesoscale interactions in tropical cyclone genesis. Mon. Wea. Rev., 125, 2643–2661.
Smith, R. K., and M. T. Montgomery, 2011: Observations of the convective environment in developing and non-developing tropical disturbances. Quart. J. Roy. Meteor. Soc., 137, 1–20.
Stern, D. P., and D. S. Nolan, 2012: On the height of the warm core in tropical cyclones. J. Atmos. Sci., 69, 1657–1680.
Zipser, E. J., and Coauthors, 2009: The Saharan air layer and the fate of African easterly waves—NASA’s AMMA field study of tropical cyclogenesis. Bull. Amer. Meteor. Soc., 90, 1137–1156.