5. Discussion 5.1 PRE Climatology 5.1.1 Statistical Climatology A nine-year study of 47 PREs occurring downstream of 21 TCs revealed that approximately one third of all TCs making landfall in the U.S. between 1998 and 2006 produced at least one PRE, with an average occurrence of 2.2 PREs per PPTC. PRE occurrence was generally directly proportional to TC occurrence, with the exception that a PRE minimum corresponded to a TC maximum during 1–10 September (Fig. 3.1). Although this PRE minimum corresponded to a minimum in U.S. landfalling TCs (Fig. 3.2), both minima likely are statistical artifacts resulting from the limited period of study. Approximately 64% of TCs following a SR track produced PREs, which is a higher percentage than TCs following any other track. Most PREs downstream of SR PPTCs formed in a band from the Florida panhandle northeastward to eastern Massachusetts (Fig. 3.3b), corresponding well with elevated 50-yr recurrence rainfall amounts (see Schumacher and Johnson 2006) along the East Coast (Fig. 1.1). This correspondence indicates that PREs forming downstream of SR PPTCs were located in climatologically favored locations for heavy rainfall. Maxima in 50-yr recurrence rainfall amounts from northeastern Georgia to southwestern North Carolina and in southeastern New York (Fig. 1.1) are located on the eastern side of elevated terrain (see Fig. 3.4), suggesting that orography likely played an important role in the formation of some PREs near those locations (see Fig. 3.3b). 156 Many PRE formation locations downstream of AR PPTCs were near the East Coast (Fig. 3.5b), while PRE formation locations downstream of CG PPTCs (Fig. 3.6b) were scattered across the U.S. Comparison of Figs. 3.5b and 3.6b with Fig. 1.1 shows that the majority of PREs forming downstream of AR and CG PPTCs occurred in climatologically favored areas for heavy rainfall, similar to the PREs forming downstream of SR PPTCs. There were a few PREs, however, that formed in western New York and western South Dakota, where the 50-yr recurrence rainfall amount is 5 in. (125 mm) or less (Fig. 1.1). Therefore, PREs occurring in these locations provided anomalously high rainfall amounts. The preference for PREs downstream of a TC to occur LOT (see Fig. 3.7) agrees with Atallah et al. (2007), who found that the rainfall distribution directly associated with a TC often shifts to LOT as the TC moves into the midlatitudes. Numerous authors (e.g., Harr and Ellsberry 2000; Harr et al. 2000; Atallah and Bosart 2003; Atallah et al. 2007) have shown that a LOT shift occurs when a TC interacts with a midlatitude cyclonic circulation located northwest of the TC, suggesting that LOT PRE formation also may be favored with this positioning of a TC in relation to a midlatitude cyclonic circulation. PREs during 1998–2006 had a median SD of 935 km from their parent TC, a time lag of 36 h before the center of the TC reached the latitude of the PRE centroids, and lasted for 12–15 h (see subsection 3.1.4). The all-PRE SD (Fig. 3.8) and time lag (Fig. 3.9) distributions both are positively skewed, indicating that outliers during the period of study were more likely to be significantly above the median than below it. SDs and time lags of AT PREs exhibited considerably 157 less variability than LOT or ROT PREs, suggesting that the locations of AT PREs may be more predictable than LOT or ROT PREs. This potentially enhanced predictability could be advantageous to forecasters, especially given the observation that AT PREs produced high rainfall rates (Fig. 3.11) that combined with high TC rainfall rates to produce excessive rainfall totals. ROT PREs produced 24-h rainfall rates equally as high as AT PREs (Fig. 3.11), mostly because of slow ROT PRE movement (Fig. 3.12). 5.1.2 Composite Climatology Composite plots provided insight into the synoptic-scale features that contributed to PRE formation downstream of SR TCs during 1998–2006. A maximum of 700-hPa upward vertical motion occurred 12 h prior to PRE initiation time within a 700-hPa confluence zone near the southeast U.S. coast (Fig. 3.13a), in a region similar to Region A of the BC78 conceptual model of antecedent rainfall (Fig. 1.9b). The overall synoptic-scale pattern was similar to that described by the M79 synoptic flash flood pattern (Fig. 1.2) in that both composite PREs occurred 1) ahead of a slow-moving midlevel trough (Fig. 3.15a), and 2) within a warm, moist air mass just ahead of a quasi-stationary lowlevel θe gradient (Fig. 3.15b). The θe gradient was in a favorable location for the slowing or stalling of low-level fronts because it was oriented along the spine of the Appalachians (see Fig. 3.4) (see O’Handley and Bosart 1996) and approximately parallel to southwesterly geostrophic flow at 700 hPa (M79). Both 158 PREs also were located in the equatorward entrance region of a 200-hPa jet (Fig. 3.15b), which is a favored location for MCS occurrence (Junker et al. 1999). 5.2 Case Studies 5.2.1 PREs Associated with Gaston (2004) The two LOT PREs presented in association with Gaston (2004) (see section 4.1) both occurred downstream of an AR TC. These PREs were studied so that the important synoptic-scale, mesoscale, and thermodynamic properties of “common” PREs in the climatology could be determined. The results showed that PRE2 and PRE3 downstream of Gaston both occurred within a 700-hPa confluence zone over New York (Fig. 4.5a), which bore a strong resemblance to the midlevel confluence zone (Region A) depicted in the BC78 conceptual model (Fig. 1.9b). The two PREs also occurred near the equatorward entrance region of a southwest–northeast-oriented 200-hPa jet streak (Fig. 4.5b), which is a signature seen in the SR composite plots (see subsection 5.1.2). This jet streak became anchored northwest of Gaston as the TC moved poleward, likely contributing to the LOT distribution of rainfall directly associated with Gaston (see Fig. 4.1a), in accordance with Atallah et al. (2007), and the LOT occurrence of PRE2 and PRE3. Precipitable water values of 47.8 mm at Buffalo, NY, near PRE2 (Fig. 4.13), and 45.5 mm at Gray, ME, near PRE3 (Fig. 4.15), both were unusually high (see subsection 4.1.3) and much greater than the M79 flash flood composite value of 37 mm. K-index values at the two locations of 37 and 35, respectively, 159 were near the M79 composite value of 35, and indicated a 70–90% chance of thunderstorms according to Henry (1988). PRE2 occurred near cyclonic geostrophic relative vorticity advection by the thermal wind at 850 hPa (see Figs. 4.3c and 4.4c), which is similar to what was observed by BC78 in the Wellsville, NY, case of heavy rain downstream of Agnes. Forcing for ascent due to geostrophic deformation also was present just southwest of PRE2 (Figs. 4.4c,d). Furthermore, PRE2 occurred near a midlevel short-wave trough and a midlevel ridge axis (Fig. 4.4a) on the “cool” side of a low-level θe boundary (see Fig. 4.12). The occurrence of the θe boundary and heavy rainfall near a midlevel ridge axis is representative of the M79 frontal flash flood pattern (Fig. 1.3), and supports the D96 hypothesis that a ridge axis aloft may favor excessive precipitation because of a tendency for sharp temperature and moisture boundaries to collect there. The low-level θe boundary near PRE2 was reflected in the surface MFC (Fig. 4.7b) and surface θ gradient (Fig. 4.7c) located in western New York. Surface MFC at 1200 UTC 30 August near and southwest of PRE2 (Fig. 4.7b) accurately predicted where heavy rainfall would be located 6 h later (see Fig. 4.8a), supporting the contention of Banacos and Schultz (2005) that MFC has some utility as a short-term forecast tool. The redevelopment of heavy rainfall upstream of PRE2 allowed for training echoes, which can prolong the duration of heavy rainfall (see D96). Furthermore, PRE2 became organized into a linear structure, which Dial and Racy (2004) suggested can happen when convection occurs along a frontal boundary in the presence of a unidirectional vertical wind 160 profile. The linear organization of PRE2 within this environment supports the suggestion of Dial and Racy (2004) that a uniform vertical wind profile may help to organize convection into linear structures. The surface θ gradient in western New York intensified between 1200 UTC (Fig. 4.7c) and 1800 UTC (Fig. 4.8c) 30 August, which is an indication of frontogenesis. Given the surface observations in Figs. 4.7c and 4.8c, differential diabatic heating along the θ gradient likely led to the frontogenesis (see Langmaid and Riordan 1998). A cross section through PRE2 (Fig. 4.12) shows that the frontogenesis extended up to 925 hPa, likely enhancing precipitation through a direct thermal circulation similar to that described in the Moore et al. (2003) schematic cross section of elevated convection (Fig. 1.6). The environment surrounding PRE2 further agreed with the Moore et al. (2003) cross section in that PRE2 was located within the equatorward entrance region of a 200-hPa jet streak (Fig. 4.3b) with no surface-based CAPE present (see Fig. 4.13). Tall and narrow elevated positive area was present, however, in the Buffalo, NY, sounding, which may have led to long residence times of air parcels in the warm cloud layer because of weak vertical velocities (see Blanchard 1998) and resulted in increased precipitation efficiency via the collision–coalescence process (Beard and Ochs 1993). PRE3 formed in an area of upslope flow (see Fig. 4.14) along the southeastern slopes of the Catskills (see Fig. 3.4) in an environment characterized by little or no synoptic-scale QG forcing for ascent (Figs. 4.5c,d). Southeasterly surface winds along the New Jersey coast (Fig. 4.9b) induced by 161 the TC circulation helped to transport the moisture necessary for excessive rainfall poleward of the TC, similar to the cases examined by Lin et al. (2001). The southeasterly surface winds also transported slightly cooler air inland from the Atlantic Ocean, inducing a weak thermal boundary with a temperature difference of 3°–4°C from central New Jersey to eastern Long Island (see Fig. 4.8c). Convection initiated just on the warm side of this boundary 3 h later (Fig. 4.9a) within a low-level θe ridge, which is a favorable region for warm-season convection (see Funk 1991) because CAPE is maximized in this region (Schwartz et al. 1990). In contrast to PRE2, the convection associated with PRE3 was surface-based, as indicated by the CAPE value at Gray, ME, of 646.6 J kg−1 (Fig. 4.15). The thermal boundary may have been akin to the subtle boundaries that numerous authors (e.g., Heideman and Fritsch 1988; Koch and Ray 1996; Fovell 2005) have suggested can help initiate convection in the warm season, and may have played a role in the development of PRE3 in addition to orographic lift. The presence of a “bentback ridge” (Maddox et al. 1978) in the midlevels of the atmosphere (Fig. 4.5a) makes the synoptic-scale pattern surrounding PRE3 resemble those described by Maddox et al. (1978) and Petersen et al. (1999) in their studies of orographically induced flash flooding events in the Rockies, and in the Barros and Kuligowski (1998) and Pontrelli et al. (1999) studies of flash flooding in upslope regions of the Appalachians. One key difference, however, is that deep warm-air advection acted as a forcing mechanism with PRE3 downstream of Gaston, but was absent in the other 162 cases. Additionally, the poleward advance of Gaston prevented the large-scale pattern from remaining favorable for orographic ascent over an extended period of time, which was an important aspect of the other cases. PRE3 subsequently moved northeastward into northern New England (Fig. 4.6a) under the support of surface MFC from eastern New York to northern Maine (Fig. 4.11b), limiting rainfall totals in the Catskills. 5.2.2 PREs Associated with Katrina (2005) Synoptic-scale studies of four PREs (one AT and three ROT) downstream of Katrina (2005) were presented in section 4.2. Mesoscale and cross-sectional analyses of the AT PRE and two of the ROT PREs also were presented so that the important properties of AT and ROT PREs occurring downstream of a CG TC could be compared and contrasted with the “common” LOT PREs that occurred downstream of Gaston, which followed a SR track. The results showed that PRE1 (the AT PRE) downstream of Katrina (2005) occurred in the equatorward entrance region of a 200-hPa jet streak (Fig. 4.19b). The strengthening of this jet streak from a wind speed maximum of greater than 40 m s −1 prior to PRE development (Fig. 4.17b) to a wind speed maximum of greater than 60 m s −1 during the PRE (Fig. 4.19b) may have been a reflection of downstream ridging induced by diabatic heating associated with Katrina. Numerous authors (e.g., Jones et al. 2003; Atallah and Bosart 2003; Atallah et al. 2007) have noted that the resulting enhancement of an upper-level jet streak often is associated with 163 increased upper-level divergence in the equatorward entrance region and intense precipitation, sometimes well downstream of a TC. PRE1 occurred just on the “warm” side of an east–west-oriented 925-hPa θe boundary (Fig. 4.19b). The east–west orientation of the θe boundary with Katrina moving northward toward it likely contributed to the AT occurrence of PRE1 downstream of Katrina. In contrast, the south-southwest–north-northeast orientation of a θe boundary downstream of Gaston with Gaston moving approximately parallel to it likely contributed to the LOT instead of AT occurrence of PRE2 downstream of Gaston. Surface analyses at 1200 UTC 29 August 2005 depict no thermal boundaries near PRE1 downstream of Katrina (see Fig. 4.23c). Furthermore, comparison of Fig. 4.23b with Fig. 4.24a shows that surface MFC did not accurately predict where heavy rain associated with PRE1 would fall. Banacos and Schultz (2005) suggest that MFC may be present but go unanalyzed either when the gridded surface analysis is too coarse to resolve any boundaries near the heavy rainfall, or when the lifting mechanisms are located above the surface. In the case of PRE1 downstream of Katrina, weak QG forcing due to cyclonic geostrophic relative vorticity advection by the thermal wind existed at 850 hPa in western Kentucky (Fig. 4.19c) near a low-level θe gradient (Fig. 4.19b), and may have provided background lift that aided in the production of PRE1. Differential diabatic heating caused by evaporational cooling associated with falling precipitation acted to enhance the thermal gradient northwest and southeast of PRE1 (Fig. 4.24c), which likely led to frontogenesis (see Langmaid and Riordan 164 1998). However, a cross section through PRE1 (Fig. 4.28) did not depict any regions of frontogenesis. The occurrence of PRE1 south of a nearly stationary midlevel cyclonic circulation (Fig. 4.19a) and on the “warm” side of a low-level θe gradient (Fig. 4.19b) rendered the synoptic-scale pattern surrounding PRE1 somewhat reminiscent of the M79 synoptic flash flood pattern (Fig. 1.2). PRE1 also formed southwest of a mesohigh located near PRE2 (Fig. 4.23a), placing PRE1 in a location that the M79 mesohigh flash flood composite suggests could be susceptible to heavy rainfall. It is difficult, however, to infer that an outflow boundary from the convection associated with PRE2 helped to trigger or enhance PRE1, since they developed at approximately the same time. The atmosphere near PRE1 showed evidence of moist absolute instability (see Fig. 4.29), which Bryan and Fritsch (2000) observed can be found near MCSs. The precipitable water value of 58 mm and the K-index value of 42.3 in the Nashville, TN, sounding at 1200 UTC 29 August both were far above the M79 flash flood composite values (see subsection 5.2.1). A uniform southwesterly wind profile helped to organize PRE1 into a linear structure, in accordance with Dial and Racy (2004), while light midlevel winds likely contributed to the slow movement of PRE1. PRE2 formed around 1200 UTC 29 August within a 700-hPa confluence zone (Fig. 4.18a) and a 925-hPa θe gradient (Fig. 4.18b), making the synopticscale pattern similar to that near PRE2 in advance of Gaston, as well as the M79 frontal flash flood (Fig. 1.3) and BC78 (Fig. 1.9b) conceptual models. While 165 PRE2 in advance of Gaston remained nearly stationary within the θe gradient, PRE2 in advance of Katrina moved steadily northeastward on the anticyclonic shear side of a 200-hPa jet streak and intensified when it reached a 925-hPa θe ridge in central New York by 0600 UTC 30 August (see Fig. 4.21b). Cyclonic geostrophic relative vorticity advection by the thermal wind at 850 hPa (see Fig. 4.21c) helped maintain PRE2 in much the same way that weak differential cyclonic vorticity advection did in the Wellsville, NY, rainstorm ahead of Agnes (BC78). PRE3 occurred in the poleward exit region of a 200-hPa jet streak (Fig. 4.21b) and in an area of cyclonic geostrophic relative vorticity advection by the thermal wind at 850 hPa (Fig. 4.21c). Weak forcing for ascent due to cyclonic geostrophic relative vorticity advection by the thermal wind also was present near PRE4 in Rhode Island and southeastern Massachusetts (Fig. 4.21c), but PRE4 was not located in an area of favorable upper-level jet dynamics (Fig. 4.21b). While forcing for ascent by QG processes clearly played a role in the occurrence of PRE3 and PRE4, it is difficult to place the synoptic-scale pattern in which they formed into either the BC78 or the M79 flash flood conceptual models. PRE2, PRE3, and PRE4 all were ROT PREs, likely because a meridionally oriented low- and midlevel ridge well east of the TC created deep southerly geostrophic flow between Katrina and the ridge (see, e.g., Figs. 4.21a,b). This synoptic-scale setup is in agreement with the findings of Atallah et al. (2007), who showed that ROT distributions of precipitation directly associated with TCs often interacted with downstream ridges. The ridge present east of 166 Katrina, however, differed in that it was oriented north–south, while the ridge in the Atallah et al. (2007) ROT composites was oriented east–west. In contrast to PRE1, surface MFC did accurately predict the locations of heavy rainfall associated with PRE3 and PRE4 in eastern New England (see Figs. 4.27a,b). PRE3 formed in Maine on the southeastern edge of surface MFC (Fig. 4.25b) and on the northwestern edge of a preexisting surface θ gradient along the coast (Fig. 4.25c). The areal extent of this θ gradient diminished by 0600 UTC, but surface observations suggest that the θ gradient extended southward to off the Massachusetts coast, despite the lack of an analyzed θ gradient there (see Fig. 4.26c). The rainfall associated with both PRE3 and PRE4 fell just on the warm side of the gradient (Fig. 4.26a). The precipitable water value at Chatham, MA, increased from 40.7 mm 6 h prior to the onset of PRE4 (Fig. 4.31) to 53.3 mm 6 h after its onset (Fig. 4.33), both of which exceeded the M79 flash flood composite value. The K-index value increased from 24.2 to 32.0 during the same period; both of these values are somewhat below the M79 flash flood composite value. Conditional instability and warm-air advection (see Figs. 4.31 and 4.33) present prior to the onset of PRE4 combined with near-surface frontogenesis (Fig. 4.32) and weak forcing for ascent at 850 hPa due to cyclonic geostrophic relative vorticity advection by the thermal wind (see Fig. 4.21c) to produce PRE4. 5.2.3 Null Case: Cindy (2005) The null case of Cindy (2005) (see section 4.3) was studied so that the synoptic-scale, mesoscale, and thermodynamic properties of a CG TC that did 167 not produce any PREs could be compared with Katrina, which was a CG TC that did produce PREs. The track of Cindy (Fig. 4.34) was similar to the track of Katrina (Fig. 4.16a), and a southwesterly 200-hPa jet streak downstream of Cindy (Fig. 4.37b) was oriented in a similar direction to the one downstream of Katrina. However, a trough in the low- and midlevels was located northeast of Cindy along the northeastern U.S. coast (Figs. 4.36b and 4.37b) instead of north (as with Katrina) or northwest (as with Gaston) of the TC. The positioning of the trough likely inhibited interaction between the TC and the midlatitude system, since the most favorable setup for tropical–midlatitude interaction is with the midlatitude system northwest of the TC (Harr and Elsberry 2000; Harr et al. 2000; Atallah et al. 2007). An isolated area of rain that developed in New York at 1200 UTC 7 July occurred near a 700-hPa ridge axis (Fig. 4.38a), which was a main feature of the M79 frontal flash flood composite (Fig. 1.3), and near cyclonic geostrophic relative vorticity advection by the thermal wind at 850 hPa (Fig. 4.38c). However, west-northwesterly 700-hPa geostrophic flow (Fig. 4.37a) and a significant 925hPa ridge (Fig. 4.37b) poleward of Cindy likely prevented the atmosphere from becoming set up in the manner of the BC78 conceptual model, or any of the M79 flash flood composites (Figs. 1.2–1.4). The 925-hPa θe value of 320 K near the area of rain in New York (Fig. 4.38b) was well below the values seen in association with the Gaston and Katrina PREs, reflecting the inability of high θ e air and deep moisture to be transported poleward. 168 Westerly surface winds in North and South Carolina at 1200 UTC 6 July (Fig. 4.39b) and 0000 UTC 7 July (Fig. 4.40b) exhibited a significant offshore component, in contrast to the onshore winds observed near PRE3 in advance of Gaston and PRE3 and PRE4 in advance of Katrina. Most observations indicated that surface winds in the vicinity of the area of rain in New York at 1200 UTC 7 July (Fig. 4.41b) were out of the east and northeast, and therefore exhibited no connection to Cindy. Surface observations in Fig. 4.41b suggested weak confluence near the New York rain area, but the absence of thermal gradients (and therefore, of frontogenesis) made the environment different from PRE2 downstream of Gaston and PRE3 and PRE4 downstream of Katrina. The New York rain area was located near elevated terrain west-southwest of Albany (see Fig. 3.4), in a region of weak cold-air damming shown by an inverted ridge axis in the mean sea level pressure field (Fig. 4.41a). The cold-air damming may have been induced partially by evaporation of the falling rain, as suggested by Fritsch et al. (1992). The Albany, NY, sounding at 1200 UTC 7 July (Fig. 4.43) shows that the low- and midlevels of the atmosphere near the New York rain area were completely saturated. The precipitable water value of 40 mm was slightly above the M79 flash flood composite value of 37 mm, but only 130% of average for Albany, in contrast to precipitable water values that were 200–245% of average for the PREs downstream of Gaston and Katrina. Backing of the winds up to 700 hPa in both the cross section (Fig. 4.42) and the Albany sounding (Fig. 4.43) implied cold-air advection, which acts to suppress upward vertical motion. This 169 vertical wind profile also contrasts with the Gaston and Katrina PREs, which featured a vertical wind profile that either veered with height or was unidirectional. Furthermore, little potential instability was evident in the cross section near the New York rain area (Fig. 4.42), and surface-based CAPE was negligible (Fig. 4.43). 5.2.4 PREs Associated with Fran (1996) The three PREs (two AT and one LOT) presented in association with Fran (1996) (see section 4.4) were studied to determine how known historical cases of heavy rainfall occurring downstream of a TC compared to cases of PREs that were included in the climatology. One difference between Fran and the other two case studies of PPTCs studied in chapter 4 was that it did not fit into one of the TC track categories outlined in subsection 3.1.2. Instead of recurving along the East Coast as AR TCs do (see Fig. 3.5a), Fran continued moving northwestward until it finally recurved near the St. Lawrence Valley (see Fig. 4.44a). PRE1 and PRE3 formed northwest of Fran in upslope regions (see Figs. 4.54, 4.55, and 4.57) along the eastern side of the Appalachians (see Fig. 3.4) and downstream of a nearly stationary midlevel trough. A “bentback ridge” (Maddox et al. 1978) in the midlevels of the troposphere (Figs. 4.45a and 4.48a) combined with Fran’s circulation to induce southeasterly flow oriented perpendicular to the mountains. This synoptic-scale pattern is similar to the one that spawned PRE3 downstream of Gaston, and is consistent with the studies of 170 Maddox et al. (1978) and Petersen et al. (1999) in the Rockies, and Barros and Kuligowski (1998) and Pontrelli et al. (1999) in the Appalachians. Figures 4.46a,b clearly show the upward vertical motion associated with PRE1 within a 925-hPa θe gradient, whereas PRE3 downstream of Gaston formed within a 925-hPa θe ridge. This 925-hPa θe gradient was reflected at the surface by a temperature gradient featuring a confluent surface wind field along it, which implies surface frontogenesis. The cross section in Fig. 4.54 reveals that both frontogenesis and warm-air advection aided orographic lift in the production of PRE1. Frontogenesis and warm-air advection also were important in the development of PRE2 downstream of Gaston in the absence of an orographic component to upward vertical motion. Unlike PRE2 downstream of Gaston, which showed evidence of elevated moist absolute instability, PRE1 downstream of Fran featured a moist–neutral stability profile, which Emanuel (1985) suggested was conducive to vigorous ascent in the presence of a deep saturated layer and low-level frontogenesis. No significant 925-hPa θe gradient (Fig. 4.48b) was evident in association with the upward vertical motion associated with PRE3 downstream of Fran (Fig. 4.48a), but surface observations in Fig. 4.52b suggest a temperature gradient was present in association with PRE3. East-southeasterly winds in the vicinity of this surface temperature gradient contributed to upward vertical motion through weak low-level warm-air advection, which also is suggested by the veering wind profile in a cross section through PRE3 (Fig. 4.57). Frontogenesis can be inferred from these surface observations, but is not resolved by the NARR cross section. 171 The Greensboro, NC, sounding (Fig. 4.58) shows that the low- and midlevels were conditionally unstable near PRE3, whereas the low- and midlevels near PRE1 featured a moist-neutral stability profile. The location of PRE3 in western North Carolina in the equatorward entrance region of a 200-hPa jet streak over the Northeast at 0000 UTC 5 September (see Fig. 4.48) agrees with the findings of Johnstone and Burrus (1998). PRE3 also appears to have been located in the poleward entrance region of a mesoscale jet streak in central North Carolina at 200 hPa, which dynamically favors descent. The presence of PRE3 in an area that was favorable for midlevel descent with respect to the mesoscale jet streak raises the issues of whether the NARR accurately depicts the mesoscale jet streak, and what, if any, dynamical relationship there was between the mesoscale jet streak and PRE3. NARR 200hPa analyses every 3 h (not shown) indicate that the mesoscale jet streak strengthened in place. The mesoscale jet streak may be a reflection of diabatically driven outflow associated with Fran and/or the PREs. The observed 200-hPa wind speed of 46.3 m s−1 in the Greensboro, NC, sounding at 0000 UTC 5 September (Fig. 4.58) is in excess of 6 m s−1 greater than the NARR 200-hPa wind speed seen in the same region (see Fig. 4.48a). Greensboro soundings from 1200 UTC and 1800 UTC 4 September, as well as 0600 UTC 5 September (not shown), confirm the continuous underestimation of 200-hPa wind speeds in the NARR, suggesting the possibility that the NARR may misrepresent the upper-level jet structure near PRE3. Further analysis would be required to determine if the placement of the mesoscale jet streak in 172 central North Carolina at this time is accurate and if PRE3 actually was near its equatorward, rather than its poleward, entrance region. The upward vertical motion associated with PRE2 downstream of Fran (Fig. 4.47a) was located near the equatorward entrance region of another 200hPa mesoscale jet streak along the North Carolina coast at 1200 UTC 4 September (Fig. 4.47b). This location placed it just on the “warm” side of a 925hPa θe gradient in southeasterly onshore flow. Despite the mesoscale nature of this jet streak and the 925-hPa θe gradient, a pronounced midlevel trough upstream of the upward vertical motion associated with PRE2 made the synopticscale pattern resemble the M79 synoptic flash flood pattern. The positioning of a 700-hPa ridge axis near PRE2, however, added an element from the M79 frontal and mesohigh flash flood composites. A coastal front first started to form at 0000 UTC 4 September in eastern North Carolina, which is a climatologically favored region for coastal frontogenesis (see Bosart et al. 1972; Bosart 1975). A surface temperature difference of about 2°C was present at this time between interior eastern North Carolina and the Outer Banks (Fig. 4.50b). The temperature gradient associated with the coastal front strengthened by 1200 UTC (Fig. 4.51b), coincident with the evolution of PRE2 along and on the cold side of the coastal front. Weak warm-air advection, shown by the veering wind profile throughout the troposphere on the right-hand side of Fig. 4.54, acted as a forcing mechanism for heavy precipitation. The 3°–4°C surface temperature difference across the coastal front was weaker than a cold season coastal front might be expected to have, but 173 agrees in magnitude with the TC-induced coastal fronts that occurred in the DeLuca (2004) and Srock (2005) case studies. 5.3 Application of PRE Research to Forecasting As noted in chapter 1, a prime motivation for this CSTAR research project has been the development of an operational framework for the accurate prediction of PREs that can be used by NWS meteorologists. The climatology and case-study results presented in chapters 3 and 4, respectively, have been intended to increase understanding and awareness of the potential for PREs to occur well in advance of TCs, because historically, PREs have been forecasted poorly. For example, archived area forecast discussions from the Paducah, KY, NWS forecast office indicate that the potential for PRE1 to occur downstream of Katrina was missed completely. Although a total of 50–100 mm of rain was forecast across the area, approximately twice that amount fell from the combination of PRE1, which was an AT PRE, and the rainfall directly associated with Katrina. Similarly, the potential for PRE4 to occur downstream of Katrina also was missed by the Taunton, MA, NWS forecast office. In both cases, flash flood watches and warnings were not issued until well after the heavy rainfall had begun, causing lead time for the events to be little to none. These examples illustrate the need for a forecast procedure to anticipate PRE formation downstream of TCs. 5.3.1 Synthesis and Introduction of Conceptual Models 174 Table IV provides a summary of the synoptic-scale, mesoscale, and thermodynamic characteristics of the SR composites and the four case studies presented in this research, with Table V serving as a reference for the acronyms used in Table IV. Not every spatial scale of each case study was examined in this thesis, making Table IV incomplete. For example, only the synoptic-scale aspects of PRE2 downstream of Katrina were studied for expediency, so any mesoscale forcing mechanisms or pertinent thermodynamic properties associated with it were omitted from Table IV. Table IV shows that several different forcing mechanisms on various spatial scales were responsible for the PREs studied. Despite the differences, a number of general statements can be made based on Table IV, including that: 1) upper-level jet dynamics played a critical role in the formation of many of the PREs, 2) many of the PREs formed either within a low-level θe gradient or near a low-level θe ridge, 3) precipitable water values far above average almost always were present when the PREs formed, 4) PREs often formed in the presence of some type of instability, and 5) the PREs often formed within a synoptic-scale or mesoscale pattern similar to conceptual models based on past research. The different characteristics of PREs represented in Table IV make an allinclusive description difficult. Therefore, the synoptic-scale and mesoscale conceptual models of PREs shown in Figs. 5.1 and 5.2, respectively, are only intended to be general forecasting aids. The conceptual models are patterned after the original BC78 conceptual model of antecedent rainfall, as well as the SR composites and AR case study of Gaston presented in this thesis. 175 Figure 5.1 shows the tendency for PREs occurring downstream of SR and AR TCs to form within a midlevel confluence zone as first described by BC78, and in the equatorward entrance region of an upper-level jet streak that often is enhanced by upper-level outflow streaming poleward from the TC (see, e.g., Atallah and Bosart 2003; Jones et al. 2003; DeLuca 2004, Atallah et al. 2007). Figure 5.1 also shows the locations of PREs near a midlevel ridge axis, which is a common feature of the M79 frontal (Fig. 1.3) and mesohigh (Fig. 1.4) flash flood composites and the BC78 conceptual model (Fig. 1.9b). The midlevel ridge axis is intersected by a low-level θe-ridge axis extending northeastward from the TC, which represents a poleward surge of moist, tropical air. Some PREs have been shown to form along such a θe-ridge axis, where low-level moisture and instability are maximized. This θe-ridge axis reflects a synoptic-scale pattern in which moisture from the TC is “open for business,” which is a phrase commonly used when the synoptic-scale pattern allows moisture to be transported poleward from the Gulf of Mexico. In contrast, the midlevel west-northwesterly flow in advance of Cindy reflected an environment in which moisture from the TC was “closed for business.” The precise location(s) of PRE formation generally is (are) governed by the positioning of the mesoscale and physiographic features shown in the mesoscale conceptual model of PREs (Fig. 5.2). This conceptual model highlights the potential for PREs to form on the windward sides of mountain ranges, where orographic lift can produce or enhance rainfall, and along lowlevel temperature/moisture boundaries. Examples of such boundaries include: 1) 176 a quasi-stationary, synoptic-scale front (as occurred near PRE2 downstream of Gaston), 2) a weak warm front (as occurred near PRE3 and PRE4 downstream of Katrina), and 3) a coastal front (as occurred near PRE2 downstream of Fran). The conceptual models do not portray vertical wind profiles in the vicinity of PREs, but a general description based on the four case studies still can be given. Vertical wind profiles near the PREs studied typically followed one of two patterns: 1) nearly constant wind direction with height approximately parallel to a frontal boundary, which likely helped to organize the PREs into linear structures (see Dial and Racy 2004), or 2) veering winds with height indicative of warm-air advection, which contributed a lifting mechanism. Those PREs featuring a veering wind profile were consistent with the Schumacher and Johnson (2005) schematic of backbuilding/quasi-stationary MCSs (their Fig. 13). Schumacher and Johnson (2005) note, however, that there appears to be no strong correlation between wind shear direction and convective organization with these types of MCSs. Meanwhile, weak midlevel wind speed shear caused some PREs to remain nearly stationary. 5.3.2 Suggested PRE Forecast Methodology Since it is apparent from subsection 5.3.1 that the proposed conceptual models of PREs are not sufficiently complete to accurately predict their formation, a flowchart illustrating a suggested PRE forecast methodology (Fig. 5.3) has been constructed. The methodology takes the form of a series of questions a forecaster should ask, with the tools he/she can use to answer them. 177 The first step is to compare the NHC forecast track of an approaching TC with the TC tracks favorable for PRE formation (see subsection 3.1.2). Placement of the NHC forecast track into one of the track categories presented can offer a forecaster a climatological perspective on potential PRE formation, and show him/her where in relation to the TC track PREs have formed historically. The next step is to compare the expected synoptic-scale pattern downstream of the TC, as depicted by numerical models and ensemble forecasts, to the proposed synoptic-scale conceptual model (Fig. 5.1), the BC78 conceptual model (Fig. 1.9b), and the M79 flash flood composites (Figs. 1.2–1.4). However, PREs result from the interaction of tropical and midlatitude airstreams, which is not adequately handled by numerical models (see Jones et al. 2003). The mesoscale conceptual model of PREs (Fig. 5.2) also can be used to hone in on specific locations where PREs may form, although numerical models may not resolve the pertinent boundaries if they are weak (see Heideman and Fritsch 1988; Koch and Ray 1996). The forecaster then can compare the anticipated locations of PREs with the NHC forecast of the TC track to determine if the PRE(s) will occur LOT, AT, or ROT. The PRE statistical distributions presented in subsection 3.1.4 should be used in conjunction with the conceptual models in Figs. 5.1 and 5.2 a as a climatological aid to forecasting possible PRE formation locations. Once the forecast PRE locations have been identified, examination of soundings and forecast cross sections is critical for determining if sufficient moisture and instability will be present in the area conducive to the formation of 178 PREs. Furthermore, the vertical wind profiles depicted in the soundings and cross sections can help the forecaster predict the future organization of any PREs that may form. If PREs do form, nowcasting using these same tools is essential for keeping up-to-date with possible flash flood watches and warnings. After the PRE is over and the TC no longer is affecting the U.S., the postevent rainfall analysis should include a depiction of the PRE(s). A concise way of accomplishing this task is to construct a total rainfall map encompassing the period of the TC’s passage across or near the U.S., without any attempt to separate rainfall directly associated with the TC from other rainfall. The total rainfall map should be accompanied by 24-h rainfall maps for each day depicted in the total rainfall map. These maps can be used as an aid to forecasting precipitation amounts when a TC approaches and/or makes landfall in the U.S., and also can be used to identify TCs that had significant PREs ahead of them for historical analysis. Reference to synoptic charts at multiple levels would be necessary, however, to determine if any particular area of rainfall ahead of a TC could be attributed to moisture streaming poleward ahead of the storm, and thus be classified as a PRE. 179 180 Table V. Acronyms used in Table IV. Acronym Definition SR Southeast recurvature AR Atlantic recurvature CG Central Gulf LOT Left of track AT Along track ROT Right of track EER Equatorward entrance region UL Upper-level LL Low-level AC Anticyclonic PER Poleward exit region MJS Mesoscale jet streak WAA Warm-air advection GRVATW Geostrophic relative vorticity advection by the thermal wind BC78 Bosart and Carr (1978) M78 Maddox et al. (1978) 181 UL Jet LL θe-Ridge Axis LOT PREs See Fig. 5.2 ML Ridge Axis ML Streamlines TC Rainfall Representative TC Tracks Fig. 5.1. Conceptual model of the synoptic-scale environment surrounding LOT PREs in advance of a SR or an AR TC, revised and updated from Bosart and Carr (1978). Position of TC is given by tropical storm symbol. Representative TC tracks are given by solid blue arrows. Low-level (LL) features are representative of the 925-hPa level, midlevel (ML) features are representative of the 700-hPa level, and upper-level (UL) features are representative of the 200-hPa level. Boxed region indicates the area of the mesoscale and physiographic conceptual model shown in Fig. 5.2. 182 UL Jet LOT PREs LL θe-Ridge Axis Mountain Axes LL Temperature/ Moisture Boundary Idealized LL Winds Fig. 5.2. Conceptual model of the mesoscale and physiographic features that can aid in producing LOT PREs in advance of SR and AR TCs. 183 Suggested PRE Forecast Methodology Is PRE formation climatologically favored? Tools: NHC forecast track PPTC climatological tracks and PRE formation locations Seasonal distribution of PRE formation Where are possible locations of PRE formation? Tools: Computer models and ensemble forecasts PRE synoptic-scale and mesoscale conceptual models M79 flash flood composites PRE SD and time lag statistics Does the thermodynamic environment favor PRE formation? Tools: Cross sections Soundings Is it possible to capture the PRE rainfall in the postevent TC rainfall analysis? Tools: Total QPE during TC passage across U.S. 24-h QPE for each day comprising the total QPE Fig. 5.3. Flowchart displaying a suggested PRE forecast methodology. 184