1. Introduction
The intensification of tropical cyclones (TCs) is related to the individual and collective interactions between the large-scale environment, storm-scale processes, and ocean–atmosphere coupling (Bosart et al. 2000). Studies investigating environmental properties have found that while important, those conditions alone do not explain the intensity change problem. For example, Hendricks et al. (2010) found that the environments of rapid and slow intensification rates are similar, and thus concluded that on average the rate of intensification may be only weakly dependent on environmental conditions. Likewise, forecasts from the Statistical Hurricane Intensity Prediction Scheme (SHIPS), an intensity change guidance product that combines synoptic predictors with climatology and persistence, are found to only explain approximately 50% of the variability in observed TC changes (DeMaria and Kaplan 1994).
In response to the limited predictability of intensification rates using environmental properties, several studies have examined the role of storm-scale processes, which are intimately linked to the precipitation properties of TCs. Studies examining the azimuthal distribution of convection in TCs have almost unanimously agreed that vertical wind shear is the dominant factor in the placement of precipitation, which tends to maximize in the downshear-left quadrant (Marks et al. 1992; Franklin et al. 1993; Frank and Ritchie 1999; Rogers et al. 2003; Wingo and Cecil 2010; Hence and Houze 2011; DeHart et al. 2014; Tao and Jiang 2015). Hence and Houze (2011) and DeHart et al. (2014) explored vertical profiles of radar reflectivity [from the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) and airborne Doppler radar, respectively] and further found that deep convection tends to initiate downshear right and matures downshear left before dissipating upshear. The upshear-right quadrant contains a “dearth” of intense reflectivity as the largest hydrometeors from the downshear-left quadrant have already fallen out.
Studies examining inner-core precipitation during intensification have invariably investigated the role of intense convection within organized convective bursts (Cecil and Zipser 1999; Eastin et al. 2005; Montgomery et al. 2006; Nolan et al. 2007; Sitkowski and Barnes 2009; Guimond et al. 2010; Jiang 2012; Kieper and Jiang 2012; Chen and Zhang 2013; Jiang and Ramirez 2013; Rogers et al. 2013, 2015; Zagrodnik and Jiang 2014; Tao and Jiang 2015). Gray (1998) found that TCs will not intensify without outbreaks of organized deep convection, even when all environmental conditions are favorable for intensification. Typically, these studies have sought to answer a few important questions regarding deep convection in TC intensification: 1) Are convective bursts a cause of TC intensification or are they a reflection of vortex-scale processes that enable convective bursts? 2) Do intensifying TCs [particularly those that undergo rapid intensification (RI)] have a higher proportion of convective bursts within the inner core? 3) If important, is there a favored location for these bursts during intensification?
Rodgers et al. (1998, 2000) and Guimond et al. (2010) analyzed several TCs in which intense convective bursts precede or are coincident with the start of RI. One hypothesis for how convective bursts are favorable for intensification is that they moisten the middle troposphere so that deep convection, and thus latent heating, can occur symmetrically within the inner core, which is favorable for eyewall contraction (Nolan et al. 2007; Montgomery et al. 2006). Others have also identified the important contributions from convective bursts and their compensating subsidence (Holland et al. 1984; Heymsfield et al. 2001; Rogers et al. 2002; Guimond et al. 2010) to the development of the warm core. As for the location most favorable for TC intensification, Shea and Gray (1973) emphasized the importance of the “inner radar radius,” defined as the innermost location of reflectivity, and its close proximity to the radius of maximum winds (RMW) in the most intense TCs. More recently, Rogers et al. (2013, 2015) composited airborne Doppler observations and concluded that convective bursts in intensifying TCs are preferentially located inside the RMW.
As many of the previously mentioned studies investigated a limited sample of intensification periods, robust conclusions can only be achieved by examining a larger dataset. One such dataset, an 11-yr TRMM Tropical Cyclone Precipitation Feature (TCPF) database (Jiang et al. 2011), was utilized by Jiang (2012) to examine the importance of intense convection for TC intensification. The TRMM TCPF database contains quantified parameters from the suite of instruments on the TRMM satellite [PR, the TRMM Microwave Imager (TMI), the Visible and Infrared Scanner (VIRS), and the Lightning Imager Sensor (LIS)] for areas of contiguous raining pixels centered within 500 km of the TC center. Jiang (2012) concluded that extremely intense convection in the inner core increases the chance of RI, but the increase is not substantial. Jiang and Ramirez (2013) used the same dataset to conclude that RI storms have a larger raining area and volumetric rain in the inner core; convective intensity (at the highest end of the spectrum), though, is not significantly greater for RI storms than for slow or neutral intensification. Cecil and Zipser (1999) also examined passive microwave statistics from another sensor, the Special Sensor Microwave Imager (SSM/I), and found a lack of a relationship between indicators of intense convection and tropical cyclone intensification. Harnos and Nesbitt (2011), using a passive microwave dataset of both SSM/I and TMI, similarly could not differentiate the role of convective bursts as a precursor or by-product of RI; instead, they found RI is better discerned by the formation of an axisymmetric convective ring of moderate intensity near the RI onset. These studies suggest that axisymmetric latent heat release is more crucial for vortex intensification than asymmetric heating, in contrast with those that investigate individual cases.
This study utilizes a 15-yr (1998–2012) passive microwave dataset to quantitatively compare the precipitation and convective characteristics during TC intensification. The dataset developed for this study complements the TCPF database (Jiang et al. 2011), by not only including data from TMI, but also by extending the analysis to other high-resolution passive microwave sensors available between 1998 and 2012. The temporal and spatial evolution of the passive microwave snapshots are investigated before and during intensification periods in an attempt to further understand the complicated relationships between precipitation and TC intensity change. This study is unique from previous satellite analyses in that precipitation properties are not only described more frequently during the period of intensification, but also in the 18 h preceding intensification. And while many previous studies have typically categorized TCs as RI versus non-RI, this study analyzes the entire spectrum of intensification rates, including weakening, when possible. New metrics for quantifying the symmetry of precipitation are also explored (i.e., the symmetry index).
2. Data and methods
The multiplatform dataset, hereafter referred to as the tropical cyclone–passive microwave dataset (TC-PMW), features 21 822 passive microwave overpasses of 497 tropical cyclones from 1998 to 2012 in the eastern Pacific (EPAC) and Atlantic basins. Satellite sensors included in the dataset are TMI, the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSRE), as well as SSM/I and the Special Sensor Microwave Imager/Sounder (SSMIS). NHC best track centers are interpolated to each passive microwave overpass using the speed calculated between each 6-hourly time period. The best track maximum sustained wind speed and minimum sea level pressure, however, are not interpolated; instead, the value associated with the nearest 6-hourly best track time is used. Without interpolation of the intensity estimates, which are reported to the nearest 5 kt (1 kt = 0.5144 m s−1), the resulting intensity change computations always have values that are multiples of 5 kt. Only those snapshots containing 100% data coverage within 1° of the interpolated best track center, and without land interaction during a 24-h period, are considered; these requirements reduce the sample of overpasses to a still-substantial 9390. Figure 1 shows the contribution of overpasses by each sensor with respect to 24-h intensity changes beginning at the 6-hourly best track time closest to the overpass time. While a unique aspect of this study is an investigation of the entire spectrum of 24-h intensification and weakening rates possible, many analyses also compare composite statistics for slow intensification (SI; 24-h intensity changes of +10 to +30 kt) and RI (24-h intensity changes of ≥30 kt).
Overlapping bar plot of the sample size of passive microwave snapshots used from the dataset, and their associated 24-h intensity changes.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
For each overpass of the inner-core region, a variety of parameters are computed from the 85–91-GHz channels, such as the mean, minimum, and median polarization-corrected temperature (PCT), as well as the fractional area of various PCT thresholds. These parameters serve as proxies for convective intensity, raining area, as well as symmetry of rainfall around the center. Using the 85-GHz channel, Spencer et al. (1989) determined that PCT below 250–260 K signifies sufficient ice scattering to indicate surface rainfall from those formerly frozen hydrometeors. They proposed a threshold of 255 K to delineate rain areas with rates greater than or equal to 1–3 mm h−1. PCT is used rather than brightness temperature in order to help distinguish depressed brightness temperatures due to ice scattering from deep convection over the ocean, from those due to the low emissivity of the ocean surface (Spencer et al. 1989; Cecil et al. 2002). In one application, Mohr and Zipser (1996) use a 225-K threshold for deep convection, with lower PCTs (below 200 K) an indicator of “stronger” convection.
The inner-core region encompasses the center or “eye,” eyewall (full or partial), and any near-center convection. Ramirez (2011) determined a mean inner-core radius of 82 km from the center with a standard deviation of 18 km. Given this result, a radial distance of 1° from the interpolated best track center serves as a reasonable approximation for the inner core in this study.
Despite differing frequencies (85 GHz for TMI and SSMI, 89 GHz for AMSRE, and 91 GHz for SSMIS) and footprints among the sensors included, all of the overpasses are composited together for this study. Yang et al. (2014) examined cloud-resolving simulations of a hurricane and squall line and found differences of up to 13 K due to differing sensor frequencies (85, 89, and 91 GHz). They emphasized the importance of a unified PMW sensor calibration to remove ambiguities caused by sensor differences when examining an individual TC’s life cycle. While that methodology may be critical for any individual case study, for this study compositing numerous cases without a unified calibration was considered acceptable after figures were analyzed individually for each sensor, leading to substantially the same conclusions. Most parameters, including the mean PCT and fractional areas, are relatively unaffected by the differing frequencies and spatial resolutions among the sensors included. Therefore, from a composite standpoint the differences between brightness temperatures at these frequencies are small and do not alter the overall results. The minimum PCT is the only statistic that is significantly affected by the spatial resolution. Compared to SSM/I and SSMIS, which have the largest footprints (15 km × 13 km), the sensors with the smaller footprints, TMI (6 km × 4 km preboost and 8 km × 6 km postboost) and AMSRE (6 km × 4 km), are more likely to capture particularly depressed pixels associated with intense convective towers and thus have a higher frequency of the coldest PCTs (~10-K difference) at the extreme end of the spectrum (<150 K). Even the highest-resolution satellites, however, are typically too coarse to resolve individual convective towers and, instead, reveal information about an aggregation of convective towers, also commonly referred to as a convective burst.
All overpasses in this study are degraded to a common grid [14-km resolution, which approximates the lower spatial resolution of SSMI(S)] by averaging “nearest neighbor” pixels, and then composited to examine the temporal and spatial distributions of PCTs for the spectrum of 24-h intensity changes possible. The composites use corresponding shear direction information from SHIPS (DeMaria and Kaplan 1994) to rotate each centered overpass relative to the shear vector. An example of this methodology can be seen using a passive microwave overpass from Hurricane Wilma in Fig. 2. These methodologies follow closely with several previous studies (Rogers et al. 2003; Lonfat et al. 2004; Wingo and Cecil 2010); however, this study is distinguishably different, not only because of the larger dataset, but also because analyses before and during the intensification period are produced at higher temporal frequency. This unique “timeline” approach allows for a detailed investigation of convective evolution during the intensity change period. The onset of the 24-h intensification period is referred to as the “0 hour” with periods during intensification referred to as +6, +12, +18, and +24 h. Periods prior to onset of intensification span from the −18 to −6 h.
(a) Original resolution, (b) degraded resolution, and (c) degraded resolution rotated relative to the shear vector (black arrow) for an 89-GHz PCT AMSRE overpass of Hurricane Wilma on 25 Oct 2005.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
In previous studies (Jiang 2012; Jiang and Ramirez 2013) a minimum of 85–91 GHz within the inner core is used as a proxy for the strength of the most intense convection. This study uses the 95th percentile of the “degraded resolution” 85–91-GHz PCT (for all “precipitating pixels” less than 250 K), approximately 190 K, as the threshold for “strong” convection. All precipitation, including “moderate” rain and convection, is defined by thresholds of PCT less than or equal to 250 K.
3. Areal coverage and spatial symmetry of precipitation
a. Spatial distributions
Figure 3 shows the spatial frequency distribution with respect to the bearing relative to the shear vector (0°) of 85–91-GHz PCT less than 250 K at the onset, or 0 h, of a range of 24-h intensity changes for Atlantic and EPAC cases. TC quadrants are referenced with the following abbreviations: downshear left (DL), downshear right (DR), upshear right (UR), and upshear left (UL). Storms that weaken most rapidly (+24-h intensity change of <−20 kt) have the second-highest occurrence (behind only the most rapid intensification periods) of 85–91-GHz PCT < 250 K in the DL quadrant (Fig. 3), a symptom of shear disrupting the storm; these cases also have the highest median initial storm intensity: >70 kt. Composites of steady-state TCs (+24-h intensity changes of −10 to +10 kt) have the lowest areal coverage of PCT < 250 K and a minimum in frequency in all quadrants; these cases have the lowest median initial storm intensity. As intensification rates increase, the coverage of PCT < 250 K at the onset increases in all quadrants. The most significant increase in coverage occurs for RI periods (+24-h intensity change of ≥30 kt); the frequency increases by >15% in the DL quadrant and by >20% in the UR quadrant. These results indicate that the occurrence of inner-core precipitation increases in all quadrants with increasing intensification rate. The RI cases exhibit the greatest initial precipitation symmetry around the center, a result similar to the axisymmetric convective organization found by Harnos and Nesbitt (2011) in low-shear cases near RI onset.
Frequency distributions of 85–91-GHz PCT < 250 K for all intensity change categories at onset (0 h). All composites are constructed relative to the shear vector (pointing straight up). Boxplots of the initial intensities for each intensity change category are displayed.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
As Fig. 3 only shows the onset of the +24-h intensity change, the remainder of the figures referenced in this section offer the complete time evolution prior to, and during, the intensification period. Figure 4 shows the composite time evolution of the frequency of 85–91-GHz PCT < 250 K in 6-hourly increments for RI periods. Within the 18-h period leading up to RI, areal coverage of PCT < 250 K (i.e., the “precipitating area”) increases slightly, most notably in the upshear quadrants, which indicates that the precipitation symmetry is already increasing prior to the onset of RI. The median TC intensity also increases by 10 kt during the pre-RI period. The observation that rapidly intensifying storms are already intensifying at the onset of RI was also found by Zagrodnik and Jiang (2014). During the +6- and +12-h periods after RI onset [“RI initial” stage, defined in Zagrodnik and Jiang (2014)], the occurrence of PCT < 250 K increases more substantially in all quadrants. The period +12 to +24 h (“RI continuing” stage) after RI onset features the highest PCT frequencies and symmetry.
Timeline of the frequency distributions of 85–91-GHz PCT < 250 K for all RI cases. All composites are constructed relative to the shear vector (pointing straight up). Boxplots of the TC intensity and shear are displayed for each 6-h interval. The 0 h is the start of the +24-h intensity change.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
Figure 5 summarizes Fig. 4 for all intensity change categories by showing Hovmöller diagrams of frequency of PCT < 250 K with respect to the bearing relative to the shear vector (0°). The TCs that weaken most rapidly have the highest initial intensities and the highest frequency of PCT < 250 K during the −18- to 0-h period. As stated before, this category consists almost exclusively of mature TCs. As one would expect for weakening TCs, the percent coverage of PCT < 250 K (raining area) tends to decrease with time in all quadrants.
Hovmöller diagrams of 85–91-GHz PCT frequency distribution < 250 K, azimuthally averaged within 1° of the center for all intensity change categories. The methodology for computing the statistics for this figure uses the PCT distributions (as in Fig. 4) and averages them for 20° azimuthal increments within 1° of the center for all intensity change categories. Quadrants are designated (x axis) with respect to bearing relative to the shear vector.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
Figure 5 also shows an important relationship between the coverage of moderate rain and the intensification rate during the preintensification period (−18 to 0 h); the occurrence and coverage of moderate rain in the DL (0°–270°) quadrant is greater for higher intensification rates (>20–25 kt), even during the preintensification period. Another subtle, but perhaps important increase is observed in the UL (180°–270°) quadrant preceding only the strongest intensification rates (≥20–25 kt). As shown previously in Fig. 4, an increase of occurrence in the UL and UR (90°–180°) quadrants indicates that precipitation symmetry is already increasing prior to the onset of RI. An important caveat is whether this result can be partially attributed to storms that, by the onset, are already intensifying and have a higher initial intensity. Removing those storms already intensifying (>5 kt) in the 18 h prior, however, yields similar results. Therefore, this result reveals that not only is increasing precipitation symmetry a possible indicator that RI is occurring, but potentially could also be used as a predictor that RI will soon occur. Finally, the maximum in coverage of PCT < 250 K actually increases and rotates cyclonically from DL toward the upshear quadrants during RI. This result is in agreement with recent observational case studies of rapidly intensifying Hurricane Earl (2010) by Stevenson et al. (2014), Rogers et al. (2015), and Susca-Lopata et al. (2015) that determined a convective burst initially DL prior to onset rotated upshear and may have contributed to vortex alignment and RI.
b. Quantifying symmetry
Symmetry is further quantified with respect to intensity change through a symmetry index created by differencing the mean 85–91-GHz PCT (within 1° of the center) for all pixels <250 K in each shear-oriented quadrant. Of the six possible mean quadrant differences, only the absolute value of the four that showed the greatest distinctions between RI and SI are used (UR − DR, UL − DL, UL − DR, and DL − UR). Figure 6 shows the time evolution of this symmetry index for slowly intensifying (dashed line) and rapidly intensifying (solid line) storms. Overall, the symmetry, according to the mean PCT < 250 K in each quadrant, changes very little for SI cases for the entire period from 18 h prior to onset to 24 h after. Additionally, the index remain fairly large (>20 K), an indication that the SI periods have a higher asymmetry than the RI periods. The RI storms differ in that they initially have an increasingly symmetric PCT distribution (significantly smaller mean quadrant difference than SI storms at the 99.9% confidence level) during the 18-h preintensification period. During the +6- to +12-h period after onset, the symmetrization increases at a more rapid rate than any other time frame. Overall, these results reaffirm that the rate of symmetrization (increasing) is a clear distinguisher of intensity change. Additionally, given that the symmetry index shown here demonstrates success in both quantifying the precipitation symmetry and clearly distinguishing the symmetry of RI storms (greater) compared to SI storms (less), this index shows promise as a potentially useful tool for intensification prediction within an operational framework.
Symmetry index for all TCs with a +24-h intensity change of 10–25 kt (SI; dashed line) and +24-h intensity change ≥30 kt (RI; solid line), which is defined by the absolute value of the shear-oriented quadrant differences of mean 85–91-GHz PCT (for all pixels <250 K) within 1° of the center (UL − DL, UL − DR, DL − UR, and UR − DR combinations used in calculations).
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
4. Investigation of “strong” convection in TC inner cores
Observational studies (Sitkowski and Barnes 2009; Guimond et al. 2010) and, more frequently modeling experiments (Hendricks et al. 2004; Montgomery and Kallenbach 1997), have demonstrated the importance of “strong” to “intense” convection within the inner core in the form of convective bursts containing “vortical” hot towers. Although the definition varies, for this study an 85–91-GHz PCT of 190 K is used as a threshold to identify strong convection and determine the relationship, if any, to TC intensity change.
Figure 7 shows the spatial frequency of 85–91-GHz PCT < 190 K at the 0 h. The highest occurrences of PCT < 190 K in the TCs that weaken most rapidly are in the DL and UL quadrants, maximizing at approximately 3.5%, a particularly small number compared to the maximum PCT < 250 K occurrence (60%). The higher occurrence of strong convection UL is consistent with the fact that the initial intensity of TCs included in this sample are typically already at hurricane strength (median intensity of 80–85 kt). Among all intensification rates shown, TCs with a +24-h intensity change of 0 to +15 kt have the lowest occurrence of strong convection in all quadrants, similar to the moderate rain distributions (Fig. 3). As intensification rate increases, Fig. 7 reveals that the occurrence of strong convection noticeably increases as well. Not unlike the 250-K threshold, those storms that intensify most rapidly have the highest occurrence of PCT < 190 K in the DL quadrant. This appears to verify the radar results of Hence and Houze (2011) and DeHart et al. (2014) that convection matures DL. For unknown reasons, the distribution of the 30–35-kt intensity change category has a double maximum: one almost directly aligned with the shear vector and another 90° left of shear. Those TCs with a +24-h intensity change greater than 35 kt have a maximum DL with substantial increases also seen in the upshear quadrants, a result analogous to the moderate rain distribution.
As in Fig. 3, but for the frequency distribution of 85–91-GHz PCT < 190 K for all intensity change categories.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
Whereas Fig. 7 shows the spatial frequency distribution of strong convection with respect to intensity change at onset, Fig. 8 shows the time evolution of that frequency distribution, except for rapidly intensifying storms only. The 18-h period prior to the onset of RI features a relatively unchanging fractional occurrence of strong convection (PCT < 190 K) nearest to the center (within 50 km). This represents a key difference from the rain coverage (previously shown in Fig. 4), which featured an increasing occurrence of moderate rain and symmetry prior to RI. There is, however, a slight increase in occurrence of strong convection downshear at radii greater than 50 km from 18 h prior to onset. This secondary maximum rotates counterclockwise slightly left of shear with time, which further emphasizes the transient nature of convection during the early stages of intensification. As mentioned previously, Rogers et al. (2015) and Stevenson et al. (2014) found a similar convective evolution prior to and during Earl’s (2010) RI. Within the 12 h after the onset of RI, the overall areal coverage of strong convection (PCT < 190 K) increases and concentrates left of shear, much like the result for the moderate rain distribution previously shown in Fig. 4. A continued invigoration of the maximum in occurrence 90° to the left of the shear vector during the “continuing RI” period (12–24 h after onset) is clearly demonstrated; this maximum rotates farther upshear at the end of the +24-h intensification period. It is hypothesized that this “convective burst” may contribute toward the consolidation and inward development of an eyewall.
As in Fig. 4, but for 85–91-GHz PCT < 190 K.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
Figure 9 summarizes the time evolution of the radial distribution of strong convection (PCT < 190 K) within the inner core for all intensity change categories. As shown before, the rapidly weakening TCs have a maximum occurrence closer to the center initially (radial distance of 20–40 km from the center) during the period within 18 h of the onset of weakening; however, this maximum, and the overall frequency of strong convection, rapidly dissipates as the TC weakening ensues. Slow intensifiers feature an increase in the occurrence of strong convection at a radial distance of 40–80 km during the intensification period; however, it only subtly contracts inward toward the center as intensification proceeds. In contrast to the observed distribution of moderate rain, the differences in the distribution of strong convection during the 18-h preintensification period among the intensity change categories >+15 kt are not easily distinguishable. Whereas the occurrence of moderate rain increases prior to RI onset, the strong convection does not noticeably increase throughout the 18 h prior to intensification. These results, in contrast to the radial distributions of moderate rainfall (not shown), indicate that the occurrence of strong convection near the center within the 18 h before the onset of intensification does not appear to be strongly correlated with the +24-h intensification rate. The frequency of strong convection during the preintensification period is only slightly greater, but of possibly greater importance it is located preferentially nearer to the center for the cases that subsequently intensify most rapidly. In addition, not surprisingly, for the greater intensification rates, there is a definitive inward contraction of strong convection as the intensification proceeds.
Hovmöller diagrams of 85–91-GHz PCT frequency distribution <190 K, radially averaged within 1° of the center for all intensity change categories.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
5. Contributions of different convective intensities to storm intensification rate
To help compare the overall contributions of various convective intensities toward intensity change, the observed distribution of 85–91-GHz PCT is partitioned into three categories: strong convection (pixels of PCT < 190 K), moderate convection (190–220 K), and moderate rain (220–250 K). Figure 10 shows the different PCT contributions within 60 km of the center for all 24-h intensity change categories at the onset of the intensification period. This smaller radius is used to emphasize where the most significant differences occur, within the far inner core. Moderate rain and convection exhibit the most distinguishable, and significant (at the 99% confidence level), differences between intensification rates, with the most rapidly intensifying periods (>+40 kt) featuring the highest occurrence of moderate rain and convection at the onset of the intensification period. In contrast, the differences in strong convection between the intensification rates are marginal and are not statistically significant at the 95% level.
Stacked bar plot of partitioned PCT thresholds used as proxies for convective intensity: <190 K (strong convection), 190–220 K (moderate convection), and 220–250 K (moderate rain). The fractional area (%) for each threshold is calculated within 60 km of the center at the onset of all 24-h intensity changes.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
A similar analysis focusing on the evolution of convective contributions for SI and RI periods (timeline methodology similar to Figs. 4 and 8) is shown in Fig. 11. Whereas Fig. 10 suggests no difference in the occurrence of strong convection for a 10-kt incremental increase in intensity change, Fig. 11, which bins those changes into either RI or SI within a 60-km radius, indicates a different result. The occurrence of strong convection is now significantly higher for RI (at the 95% confidence level) than SI not only at the onset, but also for time periods preceding and during intensification. Likewise, the fractional areas of moderate rain and convection are significantly greater for RI (at the 99% confidence level) preceding and throughout intensification. These results indicate that while moderate rain and moderate convection overwhelmingly have the highest overall contributions, observing the entire spectrum of convection, even the most intense areas, during intensification is desirable.
As in Fig. 10, but the fractional area (%) for each threshold is calculated for RI and SI storms using the timeline methodology.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
RI and SI cases are further broken down into their shear-oriented quadrants in Fig. 12 (also using a 60-km radius). Figure 12 reveals statistically significant differences at the 99% confidence level between RI and SI at the onset of intensification in the DL and UL quadrants for all convective intensity proxies (satisfies the 95% confidence level for DR). In the upshear quadrants (UL and UR) the differences between SI and RI for moderate rain and convection are even greater than in the downshear quadrants. In contrast to DL, UL, and DR, the difference between strong convection in RI and SI is not significant and minimizes in occurrence in the UR quadrant. Overall, the left-of-shear quadrants feature the largest and most significant differences between SI and RI periods for strong convection not only at onset and during intensification, but also during the 18-h preintensification period (not shown). The greatest differences for moderate rain and moderate convection, however, are seen in the upshear quadrants.
As in Fig. 10, but the fractional area (%) for each threshold is calculated in quadrants 0–60 km from the center at the onset of RI and SI storms.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
Additionally, quadrant differences in precipitation (PCT < 250 K) at the onset help further validate the results in Fig. 6. In the 18 h prior to intensification, SI storms have a more asymmetric distribution of precipitation initially than do RI storms. Figure 13 demonstrates this with a timeline evolution of quadrant-based pixel counts <250 K; pre-RI storms (18 h before onset) have significantly more “precipitating” pixels downshear (black and purple lines) and UL (red lines) than do pre-SI storms, whereas the UR quadrants (green lines) of pre-RI and pre-SI storms have more similar pixel counts. Prior to and after RI onset, the mean pixel count increases rapidly in all quadrants, significantly more so than during the same period of SI cases.
Average number of pixels <250 K at 14-km resolution for shear-oriented quadrants with 60-km radii: DL, black; DR, purple; UL, red; and UR, green; for RI (solid) and SI (dashed) storms.
Citation: Monthly Weather Review 143, 11; 10.1175/MWR-D-15-0065.1
To eliminate the potential bias related to the RI periods exhibiting intensification during the period leading up to the onset, all 24-h periods with substantial intensification prior to onset are removed from Figs. 12 and 13 (not shown). This yields similar results with significant differences between nearly all convective intensity proxies prior to and during intensification, a strong indication the results are not being excessively skewed by 24-h periods that may be a continuation of RI.
6. Conclusions
Precipitation properties are investigated using a TC-PMW dataset comprising 15 yr of satellite overpasses from multiple passive microwave sensors [AMSRE, SSMI(S), and TMI] across the Atlantic and east Pacific basins. Building on previous satellite studies (Cecil and Zipser 1999; Jiang 2012; Jiang and Ramirez 2013; Zagrodnik and Jiang 2014), a spectrum of PCT thresholds, used as proxies for precipitation and convective intensity, are investigated in this study both spatially (in shear-oriented quadrants) and temporally. However, in contrast with previous studies, precipitation distributions are examined not only more frequently during the period of intensification, but also uniquely in the 18 h preceding intensification. Similarly, while many previous studies have typically categorized TCs as RI versus non-RI, this study analyzes the entire spectrum of intensification rates, including weakening, when possible. New metrics for quantifying the symmetry of precipitation are also explored (i.e., the symmetry index).
In agreement with Cecil and Zipser (1999), Jiang and Ramirez (2013), and Kieper and Jiang (2012), this study verifies that the occurrence of inner-core precipitation at the onset (“0 hour”) increases with intensification rate in all quadrants. It is also determined that symmetry is already increasing prior to the onset of RI, a result analogous to Harnos and Nesbitt (2011), and coincident with a median increase in intensity of 10 kt during this period. Removing those storms already intensifying, however, reveals a similar result. Symmetry also increases throughout the entire 42-h period (18 h before to 24 h after onset) with the highest occurrence of PCT < 250 K (used as a proxy for raining area) in the “RI continuing” stage, which agrees with the results in Zagrodnik and Jiang (2014).
The symmetry indices developed for this study all reveal that the rate of symmetrization increases with increasing intensity change. During the 18-h period preceding SI, storms have a significantly (at the 95% confidence level) greater precipitation asymmetry, according to the mean PCT in each quadrant, than those rapidly intensifying. Throughout RI, storms definitively have a more symmetric distribution of precipitation than those with slower intensification rates. These results indicate that not only is increasing precipitation symmetry a possible indicator that RI is occurring, but it potentially could also be used as a predictor that RI will soon occur. Because symmetry is such a distinguishing quantity, additional work should attempt to better quantify it for operational forecasting.
Differing from the results for “moderate” rain, the 18-h period prior to RI onset features only a slightly higher fractional occurrence of “strong” convection than all other intensity change periods prior to onset. Therefore, the occurrence of strong convection (greater) prior to intensification does not seem to be intimately linked to increased rates of intensity change. Of possibly greater importance, strong convection is located preferentially nearer to the center for the cases that subsequently intensify most rapidly. Furthermore, as intensification rates increase, the occurrence of strong convection (proxy using 85–91-GHz PCT < 190 K) during intensification noticeably increases. A continued invigoration of strong convection is noted during the continuing RI period (+12 to +24 h after onset), possibly contributing to a consolidating eyewall. As a result of the temporal and spatial limitations of passive microwave, however, it remains to be seen whether the link with intensification is due to the intensity of individual towers or the aggregate of many towers. The spatial frequency distributions of strong convection also reveal that the maximum coverage actually increases and rotates counterclockwise from DL toward the upshear quadrants during RI. This result is analogous to a similar rotation identified in radar case studies of rapidly intensifying Hurricane Earl (2010) (Rogers et al. 2015; Stevenson et al. 2014; Susca-Lopata et al. 2015) and, thus, verifies their result for a larger sample. While the results from this study clearly demonstrate important contributions from strong convection, similar to the conclusions of Jiang (2012), because of its common occurrence across the intensity change spectrum, this study concludes that strong convection, alone, is not a sufficient condition for RI and is limited in terms of its predictive value as a precursor for RI.
Finally, an examination of contributions from three different PCT thresholds, proxies for moderate rain (220–250 K), moderate convection (190–220 K), and strong (<190 K) convection, reveals that the occurrence of strong convection is significantly higher for RI (at the 95% confidence level) than SI not only at the onset, but also for time periods preceding and during intensification. Moderate rain and convection, however, have the highest overall fractional areas, and appear to better distinguish intensification rates (statistically significant differences at the 99% confidence interval). In addition, the greatest differences for moderate rain and convection between RI and SI periods are found in the upshear quadrants. The largest and most significant differences for strong convection between SI and RI periods, however, are found in the left-of-shear quadrants (closely followed by DR) not only at onset, but also in the preceding period.
Overall, while both moderate rainfall and strong convection are deemed important for TC intensification, the greater occurrence of moderate rainfall (compared to strong convection) and its more obvious relationship with intensity change leads the authors to recommend that operational forecasters focus on precipitation of this type, rather than strong or intense convection, particularly when using passive microwave imagery. This conclusion is not meant to devalue the importance of strong and intense convection in the intensification process; rather, instead it is an acknowledgment of the limitations of the larger footprint of passive microwave sensors for adequately identifying their presence and for robustly analyzing their importance to the intensification process.
What is certain, however, is that the increased precipitation occurrence in the upshear quadrants of RI storms leads to a more symmetric distribution of precipitation initially than in SI storms and reiterates the importance of shear-oriented quadrant-based analyses. In particular, more in situ and remote sensing observations (airborne radar) are needed in the early stages of intensification, when intense convection is often more transient, to help fill the gaps between passive microwave overpasses. Observations during these preintensification periods could help give a better understanding of convective bursts and their transition to symmetry often seen during rapid intensification.
Acknowledgments
The authors thank Dan Cecil and an anonymous reviewer, whose comments greatly improved the manuscript. This research was supported by NASA Grants NNX09AC44G and NNX11AB59G under the leadership of Dr. Ramesh Kakar.
REFERENCES
Bosart, L. F., W. E. Bracken, J. Molinari, C. S. Velden, and P. G. Black, 2000: Environmental influences on the rapid intensification of Hurricane Opal (1995) over the Gulf of Mexico. Mon. Wea. Rev., 128, 322–352, doi:10.1175/1520-0493(2000)128<0322:EIOTRI>2.0.CO;2.
Cecil, D. J., and E. J. Zipser, 1999: Relationships between tropical cyclone intensity and satellite-based indicators of inner core convection: 85-GHz ice-scattering signature and lightning. Mon. Wea. Rev., 127, 103–123, doi:10.1175/1520-0493(1999)127<0103:RBTCIA>2.0.CO;2.
Cecil, D. J., E. J. Zipser, and S. W. Nesbitt, 2002: Reflectivity, ice scattering, and lightning characteristics of hurricane eyewalls and rainbands. Part I: Quantitative description. Mon. Wea. Rev., 130, 769–784, doi:10.1175/1520-0493(2002)130<0769:RISALC>2.0.CO;2.
Chen, H., and D. Zhang, 2013: On the rapid intensification of Hurricane Wilma (2005). Part II: Convective bursts and the upper-level warm core. J. Atmos. Sci., 70, 146–162, doi:10.1175/JAS-D-12-062.1.
DeHart, J. C., R. A. Houze Jr., and Robert F. Rogers, 2014: Quadrant distribution of tropical cyclone inner-core kinematics in relation to environmental shear. J. Atmos. Sci., 71, 2713–2732, doi:10.1175/JAS-D-13-0298.1.
DeMaria, M., and J. Kaplan, 1994: A Statistical Hurricane Intensity Prediction Scheme (SHIPS) for the Atlantic basin. Wea. Forecasting, 9, 209–220, doi:10.1175/1520-0434(1994)009<0209:ASHIPS>2.0.CO;2.
Eastin, M. D., W. M. Gray, and P. G. Black, 2005: Buoyancy of convective vertical motions in the inner core of intense hurricanes. Part II: Case studies. Mon. Wea. Rev., 133, 209–227, doi:10.1175/MWR-2849.1.
Frank, W. M., and E. A. Ritchie, 1999: Effects of environmental flow upon tropical cyclone structure. Mon. Wea. Rev., 127, 2044–2061, doi:10.1175/1520-0493(1999)127<2044:EOEFUT>2.0.CO;2.
Franklin, J. L., S. J. Lord, S. E. Feuer, and F. D. Marks Jr., 1993: The kinematic structure of Hurricane Gloria (1985) determined from nested analyses of dropwinsonde and Doppler wind data. Mon. Wea. Rev., 121, 2433–2451, doi:10.1175/1520-0493(1993)121<2433:TKSOHG>2.0.CO;2.
Gray, W. M., 1998: The formation of tropical cyclones. Meteor. Atmos. Phys., 67, 37–69, doi:10.1007/BF01277501.
Guimond, S. R., G. M. Heymsfield, and F. J. Turk, 2010: Multiscale observations of Hurricane Dennis (2005): The effects of hot towers on rapid intensification. J. Atmos. Sci., 67, 633–654, doi:10.1175/2009JAS3119.1.
Harnos, D. S., and S. W. Nesbitt, 2011: Convective structure in rapidly intensifying tropical cyclones as depicted by passive microwave measurements. Geophys. Res. Lett., 38, L07805, doi:10.1029/2011GL047010.
Hence, D. A., and R. A. Houze Jr., 2011: Vertical structure of hurricane eyewalls as seen by the TRMM Precipitation Radar. J. Atmos. Sci., 68, 1637–1652, doi:10.1175/2011JAS3578.1.
Hendricks, E. A., M. T. Montgomery, and C. A. Davis, 2004: On the role of “vortical” hot towers in formation of Tropical Cyclone Diana (1984). J. Atmos. Sci., 61, 1209–1232, doi:10.1175/1520-0469(2004)061<1209:TROVHT>2.0.CO;2.
Hendricks, E. A., M. S. Peng, B. Fu, and T. Li, 2010: Quantifying environmental control on tropical cyclone intensity change. Mon. Wea. Rev., 138, 3243–3271, doi:10.1175/2010MWR3185.1.
Heymsfield, G. M., J. B. Halverson, J. Simpson, L. Tian, and T. P. Bui, 2001: ER-2 Doppler radar investigations of the eyewall of Hurricane Bonnie during the Convection and Moisture Experiment-3. J. Appl. Meteor., 40, 1310–1330, doi:10.1175/1520-0450(2001)040<1310:EDRIOT>2.0.CO;2.
Holland, G. J., T. D. Keenan, and G. D. Crane, 1984: Observations of a phenomenal temperature perturbation in Tropical Cyclone Kerry (1979). Mon. Wea. Rev., 112, 1074–1082, doi:10.1175/1520-0493(1984)112<1074:OOAPTP>2.0.CO;2.
Jiang, H., 2012: The relationship between tropical cyclone intensity change and the strength of inner-core convection. Mon. Wea. Rev., 140, 1164–1176, doi:10.1175/MWR-D-11-00134.1.
Jiang, H., and E. M. Ramirez, 2013: Necessary conditions for tropical cyclone rapid intensification as derived from 11 years of TRMM data. J. Climate, 26, 6459–6470, doi:10.1175/JCLI-D-12-00432.1.
Jiang, H., C. Liu, and E. J. Zipser, 2011: A TRMM-based tropical cyclone cloud and precipitation feature database. J. Appl. Meteor. Climatol., 50, 1255–1274, doi:10.1175/2011JAMC2662.1.
Kieper, M., and H. Jiang, 2012: Predicting tropical cyclone rapid intensification using the 37GHz ring pattern identified from passive microwave measurements. Geophys. Res. Lett., 39, L13804, doi:10.1029/2012GL052115.
Lonfat, M., F. D. Marks Jr., and S. S. Chen, 2004: Precipitation distribution in tropical cyclones using the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager: A global perspective. Mon. Wea. Rev., 132, 1645–1660, doi:10.1175/1520-0493(2004)132<1645:PDITCU>2.0.CO;2.
Marks, F. D., Jr., R. A. Houze Jr., 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, doi:10.1175/1520-0469(1992)049<0919:DAIOTI>2.0.CO;2.
Mohr, K. I., and E. J. Zipser, 1996: Mesoscale convective systems defined by their 85-GHz ice scattering signature: Size and intensity comparison over tropical oceans and continents. Mon. Wea. Rev., 124, 2417–2437, doi:10.1175/1520-0493(1996)124<2417:MCSDBT>2.0.CO;2.
Montgomery, M. T., and R. J. Kallenbach, 1997: A theory for vortex Rossby-waves and its application to spiral bands and intensity changes in hurricanes. Quart. J. Roy. Meteor. Soc., 123, 435–465, doi:10.1002/qj.49712353810.
Montgomery, M. T., M. E. Nicholls, T. A. Cram, and A. B. Saunders, 2006: A vortical hot tower route to tropical cyclogenesis. J. Atmos. Sci., 63, 355–386, doi:10.1175/JAS3604.1.
Nolan, D. S., Y. Moon, and D. P. Stern, 2007: Tropical cyclone intensification from asymmetric convection: Energetics and efficiency. J. Atmos. Sci., 64, 3377–3405, doi:10.1175/JAS3988.1.
Ramirez, E. M., 2011: Convective and rainfall properties of tropical cyclone inner cores and rainbands in relation to tropical cyclone intensity changes using eleven years of Tropical Rainfall Measuring Mission data. M.S. thesis, Dept. of Atmospheric Sciences, University of Utah, 134 pp.
Rodgers, E. B., W. S. Olson, V. M. Karyampudi, and H. F. Pierce, 1998: Satellite-derived latent heating distribution and environmental influences in Hurricane Opal (1995). Mon. Wea. Rev., 126, 1229–1247, doi:10.1175/1520-0493(1998)126<1229:SDLHDA>2.0.CO;2.
Rodgers, E. B., W. Olson, J. Halverson, J. Simpson, and H. Pierce, 2000: Environmental forcing of Supertyphoon Paka’s (1997) latent heat structure. J. Appl. Meteor., 39, 1983–2006, doi:10.1175/1520-0450(2001)040<1983:EFOSPS>2.0.CO;2.
Rogers, R., S. Aberson, J. Kaplan, and S. Goldenberg, 2002: A pronounced upper-tropospheric warm anomaly encountered by the NOAA G-IV aircraft in the vicinity of deep convection. Mon. Wea. Rev., 130, 180–187, doi:10.1175/1520-0493(2002)130<0180:APUTWA>2.0.CO;2.
Rogers, R., S. Chen, J. Tenerelli, and H. Willoughby, 2003: A numerical study of the impact of vertical shear on the distribution of rainfall in Hurricane Bonnie (1998). Mon. Wea. Rev., 131, 1577–1599, doi:10.1175//2546.1.
Rogers, R., P. Reasor, and S. Lorsolo, 2013: Airborne Doppler observations of the inner-core structural differences between intensifying and steady-state tropical cyclones. Mon. Wea. Rev., 141, 2970–2991, doi:10.1175/MWR-D-12-00357.1.
Rogers, R., P. Reasor, and J. Zhang, 2015: Multiscale structure and evolution of Hurricane Earl (2010) during rapid intensification. Mon. Wea. Rev., 143, 536–562, doi:10.1175/MWR-D-14-00175.1.
Shea, D. J., and W. M. Gray, 1973: The hurricane’s inner core region. I. Symmetric and asymmetric structure. J. Atmos. Sci., 30, 1544–1564, doi:10.1175/1520-0469(1973)030<1544:THICRI>2.0.CO;2.
Sitkowski, M., and G. M. Barnes, 2009: Low-level thermodynamic, kinematic, and reflectivity fields of Hurricane Guillermo (1997) during rapid intensification. Mon. Wea. Rev., 137, 645–663, doi:10.1175/2008MWR2531.1.
Spencer, R. W., H. M. Goodman, and R. E. Hood, 1989: Precipitation retrieval over land and ocean with the SSM/I: Identification and characteristics of the scattering signal. J. Atmos. Oceanic Technol., 6, 254–273, doi:10.1175/1520-0426(1989)006<0254:PROLAO>2.0.CO;2.
Stevenson, S. N., K. L. Corbosiero, and J. Molinari, 2014: The convective evolution and rapid intensification of Hurricane Earl (2010). Mon. Wea. Rev., 142, 4364–4380, doi:10.1175/MWR-D-14-00078.1.
Susca-Lopata, G., J. Zawislak, E. Zipser, and R. Rogers, 2015: The role of observed environmental conditions and precipitation evolution in the rapid intensification of Hurricane Earl (2010). Mon. Wea. Rev., 143, 2207–2223, doi:10.1175/MWR-D-14-00283.1.
Tao, C., and H. Jiang, 2015: Distributions of shallow to very deep precipitation/convection in rapidly intensifying tropical cyclones. J. Climate, doi:10.1175/JCLI-D-14-00448.1, in press.
Wingo, M. T., and D. J. Cecil, 2010: Effects of vertical wind shear on tropical cyclone precipitation. Mon. Wea. Rev., 138, 645–662, doi:10.1175/2009MWR2921.1.
Yang, S., J. Hawkins, and K. Richardson, 2014: The improved NRL tropical cyclone monitoring system with a unified microwave brightness temperature calibration scheme. Remote Sens., 6, 4563–4581, doi:10.3390/rs6054563.
Zagrodnik, J. P., and H. Jiang, 2014: Rainfall, convection, and latent heating distributions in rapidly intensifying tropical cyclones. J. Atmos. Sci., 71, 2789–2809, doi:10.1175/JAS-D-13-0314.1.
