2546 MONTHLY WEATHER REVIEW VOLUME 138
The Antecedent Large-Scale Conditions of the ‘‘Perfect Storms’’ of Late October and Early November 1991
JASON M. CORDEIRA AND LANCE F. BOSART Department of Atmospheric and Environmental Sciences, University at Albany, State University of New York, Albany, New York
(Manuscript received 10 November 2009, in final form 26 January 2010)
ABSTRACT
The ‘‘Perfect Storms’’ (PSs) were a series of three high-impact extratropical cyclones (ECs) that impacted North America and the North Atlantic in late October and early November 1991. The PSs included the Perfect Storm in the northwest Atlantic, a second EC over the North Atlantic that developed from the interaction of the PS with Hurricane Grace, and a third EC over North America commonly known as the ‘‘1991 Halloween Blizzard.’’ The PSs greatly impacted the North Atlantic and North America with large waves, coastal flooding, heavy snow, and accumulating ice, and they also provided an opportunity to investigate the physical processes that contributed to a downstream baroclinic development (DBD) episode across North America that culmi- nated in the ECs. Downstream baroclinic development resulted from an amplification of the large-scale flow over the North Pacific that was influenced by anomalous tropical convection, the recurvature and extratropical transition of western North Pacific Tropical Cyclones Orchid, Pat, and Ruth, and the subsequent evolution of the extra- tropical flow. The progression of DBD occurred following the development of a negative PNA regime and the generation of baroclinic instability over North America associated with equatorward-displaced potential vorticity anomalies and poleward-displaced corridors of high moisture content. An analysis of the eddy ki- netic energy tendency equation demonstrated that the resulting baroclinic conversion of eddy available po- tential energy into eddy kinetic energy during the cyclogenesis process facilitated the progression of DBD across North America and the subsequent development of the ECs.
1. Introduction damage (Table 1). A majority of the meteorological impact from the PS occurred in association with high a. Background and motivation winds, large waves, and coastal flooding (Table 1). EC1, The ‘‘Perfect Storms’’ (PSs) were a series of three however, mainly impacted North Atlantic shipping routes extratropical cyclones (ECs) that occurred between with numerous high wind and wave reports. EC2 was as- 25 October and 4 November 1991 that included the sociated with an early-season cold air outbreak and record Perfect Storm (Cardone et al. 1996; Junger 1997) and snow and ice accumulations over the Midwest. The pur- two additional ECs over the North Atlantic and central pose of this paper is to investigate the antecedent large- United States (EC1 and EC2, respectively). EC1 was an scale flow evolution over the North Pacific (NP) and the intense cyclone that resulted from the interaction be- physical processes that contributed to downstream de- tween the PS and Hurricane Grace (HG; 1991), whereas velopment over North America and culminated in the EC2 is colloquially known as the ‘‘1991 Halloween three ECs. Blizzard’’ in the Midwest. The PS, EC1, and EC2 were Six-hourly locations of the PS, EC1, EC2, and HG, notorious cyclones that resulted in more than 34 deaths along with the mean sea level pressure (MSLP) and and over $300 million (1991 U.S. dollars) in property time-mean 500-hPa geopotential heights from 25 October through 4 November, show the positions of the storms relative to the time-mean amplified flow over North Corresponding author address: Jason M. Cordeira Department America and the North Atlantic (Fig. 1a). The de- of Atmospheric and Environmental Sciences, University at Al- bany, State University of New York, ES-351, 1400 Washington velopment of HG followed a baroclinic pathway before Ave., Albany, NY 12222. the system transformed into a tropical cyclone (TC) via the E-mail: [email protected] tropical transition (TT) process on 25 October (not shown;
DOI: 10.1175/2010MWR3280.1
Ó 2010 American Meteorological Society Unauthenticated | Downloaded 10/11/21 03:25 AM UTC JULY 2010 C O R D E I R A A N D B O S A R T 2547
TABLE 1. Summary of the meteorological and societal impacts of the PS, EC1, and EC2. Sources: The Environment Canada data are available online at http://www.meds-sdmm.dfo-mpo.gc.ca/ and the PS damage summary is available online at http://www.ncdc.noaa.gov/ oa/satellite/satelliteseye/cyclones/pfctstorm91/pfctstdam.html.
Impact Magnitude Location and time Source PS High winds 25.0 m s21 Environment Canada buoys 44141 Environment Canada; (sustained) and 44137 over NW Atlantic Eid et al. (1992) on 29–30 Oct High winds (gust) 28.0 (35.0) m s21 Milton, MA (Chatham, MA) Maglaras et al. (1995) on 31 Oct Maximum wave 28.5 (30.7) m Environment Canada buoy Environment Canada; heights 44141 (44137) on 29 Oct Eid et al. (1992) Heavy rain .100.0 mm Eastern Massachusetts on 30–31 Oct NCDC Property damage .$200 (.$100) East Coast (Massachusetts) Bigio (1992); NOAA (1992); million PS damage summary Coastal flood N/A New England Maglaras et al. (1995) forecasting EC1 Rapid deepening 50.0 hPa (24 h)21 North Atlantic between 0000 UTC ECMWF ERA-40 30 Oct and 0000 UTC 31 Oct High winds (gust) .30.0 m s21 North Atlantic on 31 Oct DeAngelis (1992) Large ocean swells ;10.0 m North Atlantic on 31 Oct EC2 Heavy snow .50.0 (93.7) cm East-central through northeast MN This day in weather history (Duluth, MN) on 31 Oct–2 Nov for southeast Minnesota, Ice accumulation 5.0 (7.5) cm Southwest through north-central northeast Iowa, and western IA (south-central and southeast Wisconsin (Rieck and IA) on 31 Oct–2 Nov Boyne 2006) Property damage $68.0 ($11.7) million IA (MN) Daily record min 219.08/223.38/211.68C Duluth, MN on 4 Nov/Bismarck, ND on NCDC summary of the day temperatures 31 Oct/Chicago, IL on 4 Nov
Davis and Bosart 2004). Hurricane Grace subsequently further described this amplified flow pattern on the dy- interacted with the PS, underwent extratropical transition namic tropopause (DT) as the result of possible down- (ET), and influenced the development of EC1 on stream development, diabatic heating, and the modulation 29 October. The PS later retrograded toward the northeast of the DT by TCs and TCs undergoing ET. Bosart (2003) United States, completed a counterclockwise loop over briefly reemphasized the amplified flow pattern from the Gulf Stream, underwent TT, and became a weak a potential vorticity (PV) perspective (Hoskins et al. 1985) unnamed hurricane on 1 November (Pasch 1991; Pasch and suggested EC2 was a ‘‘twin perfect storm.’’ Limited and Avila 1992). Concurrently, EC1 rapidly crossed the research, however, has focused on the evolution of the North Atlantic with characteristics of a diabatic Rossby antecedent large-scale flow. vortex (e.g., Moore and Montgomery 2004, 2005) and Studies have shown that noteworthy cyclogenesis explosively deepened (SLP decrease of 50 hPa in 24 h; events over North America such as the ‘‘Cleveland Su- Fig. 1b) approximately 2 bergerons (Sanders and Gyakum perbomb’’ of 25–26 January 1978 (e.g., Hakim et al. 1980) between 30 and 31 October. Upstream, EC2 formed 1995, hereafter HBK95), the ‘‘Superstorm’’ of 12– in the western Gulf of Mexico downstream of a deep time- 14 March 1993 (e.g., Bosart et al. 1996, hereafter B96), mean trough over the western United States and pro- and the ‘‘1998 Canadian Ice Storm’’ of 5–9 January 1998 gressed toward Hudson Bay between 31 October and (e.g., Gyakum and Roebber 2001) are often preceded by 4 November 1991. large-scale flow amplifications over the eastern NP and western North America. These studies illustrated how b. Revisiting the ‘‘Perfect Storms’’ upstream amplified flow influenced the juxtaposition of Previous research on the PS has primarily focused on equatorward-displaced regions of high PV air (e.g., upper- the extreme sea state in the North Atlantic (Table 1; level positive PV anomalies) and poleward-displaced Cameron and Parkes 1992; Eid et al. 1992; Maglaras regions of warm, moist air (e.g., lower-level positive PV et al. 1995; Cardone et al. 1996). Cameron and Parkes anomalies represented by positive potential temperature (1992), however, first depicted the development of the anomalies) during the cyclogenesis process. In regions PS during an amplification of the 500-hPa flow (e.g., of strong midlatitude baroclinity, conditions were favor- ‘‘strongly zonal to strongly meridional’’) as suggested by able for a mutual intensification of the upper- and lower- Fig. 1 (cf. their Figs. 1–4). Nielsen-Gammon (2001) level PV anomalies (Hoskins et al. 1985) and ‘‘type B’’
Unauthenticated | Downloaded 10/11/21 03:25 AM UTC 2548 MONTHLY WEATHER REVIEW VOLUME 138
FIG. 1. (a) The 6-hourly locations of HG, PS, EC1, and EC2 and the time-mean 500-hPa geopotential heights from 25 Oct through 4 Nov 1991 (dam, solid line). Tropical symbols de- note categorical intensity and characteristics of HG and the PS. (b) MSLP of HG, PS, EC1, and EC2 is given by a combination of the National Hurricane Center best-track data, ECMWF ERA-40 analyses, and surface data. The 1 and 2 bergeron (at 458N) deepening rates are given as reference.
cyclogenesis (Petterssen and Smebye 1971). Cyclogenesis of 41 strong cold-season cyclones between 1975 and 1985 of this nature is consistent with the Eady model for baro- in the eastern NP could be traced to the ascent of warm clinic growth (Eady 1949; Hoskins and Valdes 1990) air within the baroclinic environment of upstream cy- whereby baroclinic instability is favored in the presence clones in the western NP. Similarly, Chang et al. (2002) of vertical shear (given an upshear tilt between regions of suggested that the subsequent downstream energy prop- upper- and lower-level PV) and reduced static stability agation could maintain synoptic-scale eddies over less [associated with the vertical structure of an upper-level baroclinic regions through an extension of the ‘‘storm PV anomaly (Thorpe 1985)]. track.’’ Because large-scale flow amplifications and DBD Baroclinic instability/growth can also impact the down- can alter the energetic characteristics of the atmosphere stream propagation and intensity of eddy kinetic energy in response to a conversion between the zonal and eddy
Ke in association with downstream baroclinic development components of the available potential energy (APE) (DBD; Chang and Orlanski 1993; Orlanski and Sheldon and kinetic energy (KE), the Ke tendency can be used to 1993, 1995; Chang et al. 2002; Danielson et al. 2004). For identify the physical processes that contribute to DBD example, Danielson et al. (2004) found that the domi- (e.g., Orlanski and Sheldon 1993; Danielson et al. 2006; nant source for downstream energy propagation in 20 Harr and Dea 2009).
Unauthenticated | Downloaded 10/11/21 03:25 AM UTC JULY 2010 C O R D E I R A A N D B O S A R T 2549 c. Paper goals and organization TABLE 2. Summary of kinematic and thermodynamic variables partitioned for the Ke budget. Terms subscripted with an ‘‘m’’ The goals of this paper are to 1) elucidate the key represent the 30-day average, while lowercase and primed terms players in the NP that contributed to the large-scale flow represent the eddy component calculated by subtracting the total amplification and the initiation of DBD and 2) inves- from the 30-day average. tigate the physical processes that contributed to DBD Partition Parameter across North America and the development of the PS, V 5 Vm 1 v Horizontal (vector) wind EC1, and EC2 using an analysis of the Ke tendency. This v 5 vm 1 v9 Vertical (pressure) velocity paper will be organized as follows. Section 2 will outline F5Fm 1 f Geopotential height the data and methods used for the study. Sections 3 and 4 a 5 am 1 a9 Specific volume will present an overview of the Northern Hemisphere (NH) and NP flow evolutions, whereas section 5 will in- vestigate the Ke tendency during the period of DBD. as gridded brightness temperature (Tb)datafromthe Section 6 will discuss the results in the context of the Cloud Archive User Service (CLAUS; Hodges et al. 2000) development of the PS, EC1, and EC2, and section 7 will at 0.5830.58 horizontal and 3-h temporal resolutions. offer conclusions. The Ke tendency was derived by partitioning the ki- nematic and thermodynamic parameters into mean (30-day average) and eddy (derived by subtracting 6-h 2. Data and methods parameter values from the mean) components and fol- The structure of the NH, NP, and North American lowed the methodologies of Orlanski and Sheldon (1993), large-scale flow patterns was examined by using the Decker and Martin (2005), Danielson et al. (2006), and Pacific–North American (PNA) teleconnection index Harr and Dea (2009). The Ke tendency equation in (Wallace and Gutzler 1981; Barnston and Livezey 1987; pressure coordinates is given by Feldstein 2002) and the Arctic Oscillation (AO) as ›K › v9f a measure of the ‘‘zonal index cycle’’ (Rossby et al. 1939; e ( ) 5 $ (vf)a v9a9 [$ (VKe)] Rossby and Willett 1948; Thompson and Wallace 2000) ›t ›p between October and November 1991. Daily values of ›(v9K ) e v (v $V ) 1 residual, (1) the PNA index were calculated from the 2.5832.58 ›p m gridded 0000 and 1200 UTC National Centers for Envi- 2 ronmental Prediction–National Center for Atmospheric where Ke 5 ½jvj . The variables in (1) and partitioning Research (NCEP–NCAR) reanalysis (Kalnay et al. 1996; procedure is further presented in Table 2. The first three Kistler et al. 2001) of 500-hPa geopotential height fields terms on the right-hand side (rhs) of (1) represent the over fixed domains following Archambault et al. (2008). advection of eddy geopotential heights by the eddy ve- Daily values of the AO were obtained from the Climate locity [i.e., 2(v $f)] and were expanded in flux form Prediction Center (CPC) and were constructed by pro- following Orlanski and Katzfey (1991). The first term in jecting the daily 1000-hPa height anomalies onto the (1) represents the ageostrophic geopotential flux (AGF) leading empirical orthogonal function of the monthly divergence. The nondivergent component of the AGF mean 1000-hPa height anomaly poleward of 208N(more was removed using Eq. (4.5) from Orlanski and Sheldon information is available online at http://www.cpc.noaa. (1993). The AGF divergence is dynamically consistent gov/products/precip/CWlink/daily_ao_index/history/ with the advection of absolute vorticity by the irrota- history.shtml). tional component of the horizontal wind. The second Gridded datasets of the 40-yr European Centre for term represents (in an average sense) the baroclinic
Medium-Range Weather Forecasts (ECMWF) Re-Analysis conversion (BC) of eddy APE into Ke and is observed (ERA-40; Uppala et al. 2005), with 1.125831.1258 hor- during vertical displacements of warm and cold air izontal grid spacing, 13 vertical levels (1000–50 hPa), and (Lorenz 1955). For example, the BC term is positive 6-h temporal resolution constituted the primary data associated with processes involving the ascent of warm source for the large-scale overview and Ke tendency air (e.g., frontogenesis, warm air advection, and deep analysis. Additionally, the North Pacific overview is moist convection). The third term represents the vertical aided by infrared satellite (IR) imagery from the National flux divergence and is typically very small when in- Climatic Data Center (NCDC) through their Global In- tegrated over the depth of the atmosphere (Orlanski and ternational Satellite Cloud Climatology Project (ISCCP) Sheldon 1993), while the fourth and fifth terms represent
B1 Browse System (GIBBS) archive (more information horizontal and vertical Ke flux divergence, respectively. available online at http://www.ncdc.noaa.gov/gibbs/) and The sixth term represents barotropic Ke generation, or
Unauthenticated | Downloaded 10/11/21 03:25 AM UTC 2550 MONTHLY WEATHER REVIEW VOLUME 138
6 2 21 FIG. 2. Time-mean 300-hPa streamfunction (dark contours every 10 3 10 m s ) and streamfunction anomaly (shaded according to scale every 10 3 106 m2 s21 with negative values dashed) from an 11-day weighted climatology for (a) 10–20 Oct, (b) 15–25 Oct, (c) 20– 30 Oct, and (d) 25 Oct–4 Nov 1991. Figures were generated using the 2.58 NCEP–NCAR reanalysis.
the Ke tendency due to a conversion of KE between the across North America. The Ke analysis will be com- zonal mean and eddy flow. All other processes, which plemented by quasigeostrophic (QG) and PV diag- include frictional dissipation, are contained in the sev- nostics in order to identify important features that enth term (residual). contribute to the Ke tendency that are commonly rep- The vertically integrated Ke tendency was used to resented by traditional methods. investigate the downstream propagation of Ke, inter- actions between Ke maxima for the case of DBD, and 3. Time-mean hemispheric overview sources of Ke generation within the PSs. Upon vertical integration between 1000 and 50 hPa, (1) can be sim- Eleven-day overlapping periods of the time-mean 1 plified to 300-hPa streamfunction and streamfunction anomaly over the NP illustrate a flow pattern characterized by flat ›K e ridges over the western NP (R1) and western North ’ $ (vf)a v9a9 [$ (VKe)] ›t America (R2) and by troughs over the central NP (T1)