Abstract

Heavy precipitation in the northeastern United States is examined through observational and numerical modeling analyses for a weather system that produced extreme rainfall rates and urban flash flooding over the New York–New Jersey region on 4–5 October 2006. Hydrometeorological analyses combine observations from Weather Surveillance Radar-1988 Doppler (WSR-88D) weather radars, the National Lightning Detection Network, surface observing stations in the northeastern United States, a vertically pointing lidar system, and a Joss–Waldvogel disdrometer with simulations from the Weather Research and Forecasting Model (WRF). Rainfall analyses from the Hydro-Next Generation Weather Radar (NEXRAD) system, based on observations from WSR-88D radars in State College, Pennsylvania, and Fort Dix, New Jersey, and WRF model simulations show that heavy rainfall was organized into long-lived lines of convective precipitation, with associated regions of stratiform precipitation, that develop along a frontal zone.

Structure and evolution of convective storm elements that produced extreme rainfall rates over the New York–New Jersey urban corridor were influenced by the complex terrain of the central Appalachians, the diurnal cycle of convection, and the history of convective evolution in the frontal zone. Extreme rainfall rates and flash flooding were produced by a “leading line–trailing stratiform” system that was rapidly dissipating as it passed over the New York–New Jersey region. Radar, disdrometer, and lidar observations are used in combination with model analyses to examine the dynamical and cloud microphysical processes that control the spatial and temporal structure of heavy rainfall. The study illustrates key elements of the spatial and temporal distribution of rainfall that can be used to characterize flash flood hazards in the urban corridor of the northeastern United States.

1. Introduction

The East Coast of the United States is a densely urbanized region with a complex precipitation climatology. Extreme flood-producing rainfall in this region is strongly influenced by orographic precipitation mechanisms associated with the Appalachian Mountains to the west and differential surface heating and moisture availability associated with the land–ocean boundary to the east (Barros and Kuligowski 1998; Ntelekos et al. 2007, 2008; Holt and Pullen 2007; Pullen et al. 2007). In addition to these large-scale influences, the urban environment may affect rainfall climatology in and near heavily developed urban centers as a result of differential heating of the atmosphere from the “urban heat island,” contrasts in surface roughness associated with the “urban canopy” and microphysical controls of precipitation mechanisms associated with urban aerosols (Shepherd 2005; Ntelekos et al. 2007, 2008).

Extreme rainfall rates along the urban corridor of the northeastern United States, and the associated flash flood hazards, are closely linked to organized thunderstorm systems embedded in extratropical cyclone systems (Ntelekos et al. 2007, 2008). A cold front passed through the eastern United States on 4–5 October 2006, producing extreme short-term rainfall rates in the New York –New Jersey (NY–NJ) region. Two convective systems with linear organization were responsible for rainfall over the New York–New Jersey region. The first formed in western Pennsylvania around 1700 UTC 4 October and passed over the NY–NJ region from 0000 to 0200 UTC 5 October. The second formed around 2300 UTC 4 October in the Appalachian Mountains to the rear of the first system and passed over the NY–NJ region from 0300 to 0500 UTC 5 October.

Observations of raindrop size distributions from a Joss–Waldvogel disdrometer in Princeton, New Jersey, indicated peak rainfall rates of 125 mm h−1 over the Harry’s Brook watershed (Fig. 1). Peak discharge from Harry’s Brook, which drains the urban core of Princeton, New Jersey, exceeded 11.2 m3 s−1 from a drainage area of approximately 1.1 km2. The “unit discharge” (discharge divided by drainage area) peak of 10.2 m3 s−1 km−2 approached the envelope curve of flood peaks for the eastern United States (Smith et al. 2005). Variability of rainfall rates at 1–30 min time scales plays a critical role in flood response for small urban watersheds. In section 3, we examine short-term variability of rainfall rate for the 4–5 October 2006 storm.

Fig. 1.

Time series of 1-min rainfall rate (gray line with closed circles in mm h−1; from Joss–Waldvogel disdrometer, as described in section 2) and discharge (black line with open circles in m3 s−1) for Harry’s Brook watershed (at a drainage area of 1.1 square miles) in Princeton, NJ (see Fig. 2 for location).

Fig. 1.

Time series of 1-min rainfall rate (gray line with closed circles in mm h−1; from Joss–Waldvogel disdrometer, as described in section 2) and discharge (black line with open circles in m3 s−1) for Harry’s Brook watershed (at a drainage area of 1.1 square miles) in Princeton, NJ (see Fig. 2 for location).

Analyses of the 4–5 October 2006 storm combine numerical model simulations using the Weather Research and Forecasting Model (WRF; Skamarock et al. 2007) and observations from in situ and remote sensing systems to examine the space-time distribution of rainfall associated with passage of a frontal zone through the northeastern US. Analyses are used to examine structure and evolution of rainfall in the frontal zone. Regional rainfall analyses covering the area affected by the frontal zone are based on high-resolution rainfall fields derived using the Hydro-Next Generation Weather Radar (NEXRAD) system (Krajewski et al. 2007). Volume scan radar reflectivity observations and cloud-to-ground (CG) lightning observations from the National Lightning Detection Network (NLDN) are used to examine convective evolution of organized thunderstorm systems embedded in the frontal zone. Disdrometer and lidar observations are used to examine microphysical processes associated with extreme short-term rainfall rates in this region. Regional analyses of storm evolution are used to enhance the understanding of physical mechanisms associated with extreme short-term rainfall rates along the urban corridor of the NY–NJ region.

Characterization of rainfall distribution centers on Lagrangian analyses of storm structure, motion, and magnitudes of rainfall rate. Models of rainfall distribution in which the space–time rainfall distribution is tied to the organization of convective rainfall elements (Austin and Houze 1972; Waymire et al. 1984; Smith and Karr 1990) motivate the analyses of space–time rainfall variability. We focus on structure and motion of convective–stratiform rain elements relative to 1) the frontal boundary, 2) complex terrain of the central Appalachians, and 3) the diurnal cycle of convection (see Ntelekos et al. 2007). Radar, disdrometer, and lidar observations provide the observational resources for linking rainfall rate distribution over the urban corridor of the NY–NJ region to the evolving storm structure.

Contents of the sections are as follows. In section 2, we introduce data used for analyses and the details of the WRF implementation. Results are presented in section 3. The meteorological environment of the storm systems is introduced in section 3a. The mesoscale structure and evolution of the two convective systems that affect the New York–New Jersey region are discussed in section 3b. In section 3c, the coupled microphysical and dynamical processes associated with extreme rainfall rates over the urban corridor of the New York–New Jersey region are presented. A summary and conclusions are given in the final section.

2. Data and methods

Analyses of heavy rainfall for the 4–5 October 2006 storm system focus on the New York–New Jersey metropolitan area, but the study domain extends into western Pennsylvania (Fig. 2). The domain for model simulations is illustrated in Fig. 2, along with locations of principal observing systems.

Fig. 2.

Study domain for the 4–5 Oct 2006 storm with locations of observing systems, including WSR-88D weather radars KCCX and KDIX, a Joss–Waldvogel disdrometer at Princeton, NJ (PRZ), a lidar system at CCNY, and a radiosonde station in Upton, NY. The shaded area indicates the area of coverage for the KCCX and KDIX radars. Most of the WRF model simulations were performed for the region D1, shown in the figure; WRF-Chem simulations were performed with nested domains D1 and D2.

Fig. 2.

Study domain for the 4–5 Oct 2006 storm with locations of observing systems, including WSR-88D weather radars KCCX and KDIX, a Joss–Waldvogel disdrometer at Princeton, NJ (PRZ), a lidar system at CCNY, and a radiosonde station in Upton, NY. The shaded area indicates the area of coverage for the KCCX and KDIX radars. Most of the WRF model simulations were performed for the region D1, shown in the figure; WRF-Chem simulations were performed with nested domains D1 and D2.

The Hydro-NEXRAD system is used to derive radar rainfall fields, based on volume scan reflectivity observations from WSR-88D weather radars in State College, Pennsylvania (KCCX), and Fort Dix, New Jersey (KDIX; Fig. 2). Rainfall analyses are derived at 15-min time interval on a “super Hydrologic Rainfall Analysis Project (HRAP)” grid (roughly 1-km horizontal resolution; see Reed and Maidment 1999). Analyses use a “convective” ZR relationship (R = aZb; a = 0.017, b = 0.71, where R is rainfall rate in mm h−1 and Z is radar reflectivity factor in mm6 m−3) and an extensive set of quality-control algorithms (see Krajewski et al. 2007 for algorithm details).

Additional observing systems used in this study include the NLDN for CG lightning data (Orville and Huffines 2001; see also Ntelekos et al. 2007); a Joss–Waldvogel disdrometer at the Princeton study site for raindrop size distributions (Smith et al. 2008; Uijlenhoet et al. 2003); a vertically pointing lidar system at the City College of New York (CCNY) in Manhattan for aerosol observations; surface observations for the northeastern United States (Fig. 4); upper-air observations from Upton, New York; and discharge observations from Harry’s Brook stream gauging network (Fig. 1). The disdrometer provides raindrop size distribution observations at 1-min time interval and is used for analyses of microphysical properties of extreme rainfall rates. The lidar system provides vertical profiles of backscatter and extinction at 1064 (near IR), 532 (visible), and 355 nm (UV). Data were collected at 1-min time interval and are used to examine aerosol properties in the prestorm environment.

Fig. 4.

Surface wind fields (m s−1) and dewpoint temperature (°C) at surface meteorological stations for 0200 UTC 5 Oct 2006. Frontal location is illustrated.

Fig. 4.

Surface wind fields (m s−1) and dewpoint temperature (°C) at surface meteorological stations for 0200 UTC 5 Oct 2006. Frontal location is illustrated.

The WRF, developed by the National Center for Atmospheric Research (NCAR), is a nonhydrostatic mesoscale model. The Advanced Research WRF (ARW-WRF; Skamarock et al. 2007), version 2.2, was used in this study. Initial and boundary conditions for most of the ARW-WRF model simulations presented in this study are obtained from Eta Model output fields (Rogers et al. 1996) with 40-km resolution. For sensitivity studies, we also carry out simulations using Global Forecast System (GFS) model fields (Shea et al. 1995, 2003) with 1° resolution (GFS-BC), and North American Regional Reanalysis (NARR) model fields (Ebisuzaki et al. 2004) with 32-km resolution (NARR-BC) as initial and boundary conditions.

Most WRF simulations were performed on a grid with a resolution of 3 km. The study domain for WRF analyses covers the area shown in Fig. 2, which is approximately 900 km (37°–45°N) by 900 km (81°–71°W). The time step of model simulations is 18 s. The simulations were initiated at 0000 UTC 4 October 2006 and ended at 1200 UTC 5 October 2006. The physics options (Skamarock et al. 2007) for the “base” model simulation with Eta Model fields as initial and boundary conditions were the 1) Thompson microphysics scheme, 2) Rapid Radiative Transfer Model (RRTM) scheme for longwave radiation, 3) fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) Dudhia scheme for shortwave radiation, 4) Monin–Obukhov (Janjić) surface layer scheme, 5) Noah land surface model, and 6) Mellor–Yamada–Janjić turbulent kinetic energy (TKE) boundary layer scheme. Sensitivity analyses also use the Lin et al. (1983) microphysics scheme (Lin-MP) and WRF coupled with chemistry (WRF-Chem) with the modified Lin et al. microphysics scheme and the Carbon-Bond Mechanism version Z (CBMZ) chemical mechanism incorporating Model for Simulating Aerosol Interactions and Chemistry (MOSAIC) aerosol interactions, which also include some aqueous reactions (based on the results of Ntelekos et al. 2008). Unlike the other simulations, the WRF-Chem simulation was performed on a nested domain as shown in Fig. 2 with 9- and 3-km resolution, respectively.

3. Results

a. Structure and evolution of the frontal zone

A cold front moved across the northeastern United States from 1800 UTC 4 October 2006 until 0400 UTC 5 October 2006 (Figs. 3 and 4). The base simulation captures broad features of the structure and evolution of the frontal zone (Fig. 3). In the model simulation, the cold front moves too rapidly through the region compared with surface observations (Fig. 4). This feature is unchanged in the GFS-BC and NARR-BC simulations. Model fields and surface observations indicate dewpoint temperatures greater than 16°C in the warm sector ahead of the cold front. The model simulations provide good representations of steering-level winds. For the Upton, New York, radiosonde station, “storm motion,” derived as the vertically integrated mass-weighted wind field, is characterized by the direction of 294° and speed of 12 m s−1 at 0000 UTC 5 October. The model simulated values of steering wind 288° and 13 m s−1.

Fig. 3.

Model-simulated sea level pressure (hPa) and wind field at 10-m height from 1800 UTC 4 Oct through 0400 UTC 5 Oct 2006. Frontal location is illustrated for 0200 UTC 5 Oct 2006.

Fig. 3.

Model-simulated sea level pressure (hPa) and wind field at 10-m height from 1800 UTC 4 Oct through 0400 UTC 5 Oct 2006. Frontal location is illustrated for 0200 UTC 5 Oct 2006.

Heavy rainfall over the study region on 4–5 October 2006 was dominated by convective rainfall, and the life cycle of convection exhibited a pronounced temporal variation, as represented in Fig. 5 through the time series of CG lightning strike density. CG lightning strikes peaked between 2200 and 2300 UTC October 4 (Fig. 5) and a sharp increase in CG strikes initiated around 1900 UTC with a local maximum at 2000 UTC. CG strikes decreased steadily and rapidly after 0200 UTC October 5.

Fig. 5.

Time series of NLDN CG lightning strikes over the study domain shown in Fig. 2.

Fig. 5.

Time series of NLDN CG lightning strikes over the study domain shown in Fig. 2.

Model simulations were used to reconstruct the evolving fields of convective available potential energy (CAPE) during the storm life cycle (Fig. 6). Simulations show an axis of elevated CAPE in the warm sector that progresses eastward and assumes maximum values between 2100 and 2300 UTC (Fig. 6). The spatial positions of the CAPE maximum should be interpreted in light of the model simulation bias of rapid movement of the frontal boundary. As will be shown in the following section, organization of convective systems parallels the evolution of the CAPE field. The elevated values of CAPE are closely linked to strong low-level moisture transport, as reflected in the 850-hPa dewpoint temperature and wind fields (Fig. 7).

Fig. 6.

Model simulated maximum CAPE (J kg−1) from 1800 UTC 4 Oct through 0400 UTC 5 Oct 2006 with base model configuration described in section 2.

Fig. 6.

Model simulated maximum CAPE (J kg−1) from 1800 UTC 4 Oct through 0400 UTC 5 Oct 2006 with base model configuration described in section 2.

Fig. 7.

Model simulated wind field (m s−1) and dewpoint temperature (°C) at 850 hPa from 1800 UTC 4 Oct through 0400 UTC on 5 Oct 2006 with base model configuration.

Fig. 7.

Model simulated wind field (m s−1) and dewpoint temperature (°C) at 850 hPa from 1800 UTC 4 Oct through 0400 UTC on 5 Oct 2006 with base model configuration.

Model simulations (Fig. 8) capture broad features of the heavy rainfall distribution over the NY–NJ region but not the details of storm structure and rainfall rate variability. In addition to the base simulation, we examined rainfall distribution for GFS-BC, Lin-MP, and WRF-Chem model simulations. Although the rainfall distribution differs for all the simulations, rainfall is concentrated along the frontal boundary for all the simulations.

Fig. 8.

Model-simulated accumulated precipitation (mm) during the period 1800 UTC 4 Oct through 0500 UTC Oct 5 2006 over domain 2 (see Fig. 2). Simulations are as follows: (top left) base model configuration, (top right) with GFS model fields as initial and boundary conditions, (bottom left) with Lin et al. (1983) microphysical scheme replacing Thompson scheme, and (bottom right) WRF-Chem simulation with CBMZ-MOSAIC configuration.

Fig. 8.

Model-simulated accumulated precipitation (mm) during the period 1800 UTC 4 Oct through 0500 UTC Oct 5 2006 over domain 2 (see Fig. 2). Simulations are as follows: (top left) base model configuration, (top right) with GFS model fields as initial and boundary conditions, (bottom left) with Lin et al. (1983) microphysical scheme replacing Thompson scheme, and (bottom right) WRF-Chem simulation with CBMZ-MOSAIC configuration.

At 0200 UTC, composite radar imagery shows that rainfall was organized into convective elements in the warm sector to the southeast of the cold front (Fig. 9a; see also Fig. 4). A decaying convective system had passed over Manhattan and was located over Long Island at 0200 UTC. Convective rain areas extended along the frontal zone from New England through Pennsylvania, along the southern boundary of Pennsylvania, and into Ohio (Fig. 9a).

Fig. 9.

(top) Composite radar reflectivity (dBZ) at 0200 UTC 5 Oct 2006. (bottom) Conceptual model of spatial rainfall organization for the 4–5 Oct storm (see also top). Within the frontal zone, there are multiple mesoscale rain areas. The arrow indicates the direction of motion of the front. Systems C and D are the mesoscale rain areas analyzed in the text. Dark colors (shaded) represent convective (stratiform) rain areas.

Fig. 9.

(top) Composite radar reflectivity (dBZ) at 0200 UTC 5 Oct 2006. (bottom) Conceptual model of spatial rainfall organization for the 4–5 Oct storm (see also top). Within the frontal zone, there are multiple mesoscale rain areas. The arrow indicates the direction of motion of the front. Systems C and D are the mesoscale rain areas analyzed in the text. Dark colors (shaded) represent convective (stratiform) rain areas.

Rainfall organization in the frontal zone (Fig. 9a) can be related to the structure and motion of the frontal zone. Rainfall organization for the 4–5 October storm system was characterized by mesoscale rain areas with regions of convective (darker shading in Fig. 9b) and stratiform precipitation [lighter shading in Fig. 9b; see Steiner et al. (1995); Steiner and Smith (1998) for discussions of convective–stratiform separation]. Linear organization of convective rain areas (systems C, D, and E in Fig. 9b) was a persistent feature of rainfall distribution, but the relationship of linear organization of rainfall to organization and the position of the frontal zone varied over the life cycle of the system. The evolution of mesoscale rain areas [relative to terrain features (Fig. 2)], the position during the cycle of convective evolution over the course of the day (Fig. 4), and the position within the frontal zone were all important elements of the space–time distribution of rainfall. These elements are examined below in the context of their role in determining the heavy rainfall distribution along the urban corridor of the northeastern United States.

b. Mesoscale rainfall analyses

Rainfall over the metropolitan NY–NJ region on 4–5 October 2006 was associated with two convective systems that developed in association with the frontal zone. The first was a long-lived linear convective system that developed in western Pennsylvania around 1700 UTC 4 October and decayed over the NY–NJ region from 0000 to 0200 UTC October 5 (illustrated schematically as system D in Fig. 9b). The second system developed as isolated convective elements in the central Appalachians to the rear of the first system around 2300 UTC and organized into a leading line–trailing stratiform system (illustrated schematically as system C in Fig. 9b) that produced extreme short-term rainfall rates over the NY–NJ metropolitan region.

During the 9-h period from 1700 UTC 4 October until 0200 UTC 5 October, system D evolved from an extended line of convection along the western margin of the Appalachians to an intense convective system along the eastern portion of the Appalachians to a decaying convective system over the NY–NJ coastal region (Fig. 10). The position and orientation of the line of convection were analyzed from the 15-min Hydro-NEXRAD rainfall products. The “position” of the line is defined as the centroid of the 25 mm h−1 rain area, and the orientation refers to the angle of the line relative to due north (0°–90°, with 0° denoting the north–south orientation and 90° denoting the west–east orientation).

Fig. 10.

The 15-min Hydro-NEXRAD rainfall accumulation fields (mm) for the period from 1800 to 2300 UTC. Each of the rainfall fields represents accumulation during the 15-min period ending at the indicated time (i.e., 1800 UTC fields represents the period from 1745 to 1800 UTC).

Fig. 10.

The 15-min Hydro-NEXRAD rainfall accumulation fields (mm) for the period from 1800 to 2300 UTC. Each of the rainfall fields represents accumulation during the 15-min period ending at the indicated time (i.e., 1800 UTC fields represents the period from 1745 to 1800 UTC).

At 1800 UTC 4 October, the line of convection was located along the western margin of the central Appalachians. The orientation of the line was 70° from due north (see Fig. 10). From 1800 to 0000 UTC, the mean speed of the convective line was 6.7 m s−1. As the system evolved, the orientation of the convective line shifted toward an east–west orientation (Fig. 10); at 2000 UTC, the convective line was oriented 60° from north. The rainfall structure for system D was characterized by a line of convective rainfall and an evolving region of stratiform rainfall (Fig. 10). Stratiform rainfall was concentrated on the northeastern end of the line of convection over the life cycle of the system.

The evolving distribution of convective and stratiform rainfall was examined through time series analyses of the rainfall rate distribution within the line (Fig. 11). The rain area with a 15-min rainfall rate exceeding 25 mm h−1 is used as a surrogate for convective rain area (Fig. 11, top). The additional area associated with stratiform rainfall is characterized by the area with 15-min rainfall rates exceeding 2 mm h−1 (Fig. 11, bottom).

Fig. 11.

Time series of area (km2) with 15-min rain rates greater than (top) 25 mm h−1 and (bottom) 2 mm h−1.

Fig. 11.

Time series of area (km2) with 15-min rain rates greater than (top) 25 mm h−1 and (bottom) 2 mm h−1.

Convective rain area (Fig. 11, top) increased from 400 km2 at 1800 UTC to a local maximum of more than 900 km2 at 2000 UTC when the system reached the western margin of the Appalachian Mountains. Convective rain area increased rapidly to its maximum of 1600 km2 at 2200 UTC as it approached the eastern margin of the central Appalachians (Fig. 11, top). Convective rain area was less than the 2000 UTC local maximum and the 2200 UTC maximum during much of its passage through the interior mountain region of the central Appalachians (Fig. 11, top). The maximum value of rain area exceeding 2 mm h−1 (Fig. 11, bottom) was 16 000 km2 and coincided with the 2200 UTC convective rain area maximum.

The system decayed rapidly following the 2200 UTC maximum in convective and stratiform rain area (Fig. 11). From 0000 to 0200 UTC 5 October, surface precipitation from system D was produced mainly by a region of stratiform rainfall. The system produced light rainfall as it passed over the New York City (NYC) metropolitan region between 0100 and 0200 UTC.

The period with largest convective rain area for the first system (2200 UTC, see Figs. 10 and 11) corresponded to the regional peak in CG lightning (see Fig. 5). It was also the period of the most intensive convection for the system, based on vertical profiles of reflectivity (figures not shown). Values of reflectivity exceeding 35 dBZ extended above 10 km at 2200 UTC; 50-dBZ reflectivity values reached 7 km. The vertical extent of the large reflectivity values at 2200 UTC was much greater than earlier in the system life cycle and rapid decay in convective intensity occurred as the system moved east of the complex terrain of the central Appalachians after 2200 UTC.

Local maxima in the contour map of storm total CG lightning flash density (Fig. 12) in the western and eastern portions of the Appalachian region are separated by a local minimum in the interior of the central Appalachian region. The spatial distribution of CG lightning for the 4–5 October storm reflects the climatological distribution of lightning (Orville and Huffines 2001) and highlights the role of complex terrain for the evolution of thunderstorm systems in the region (see also Ntelekos et al. 2007, 2008). The spatial distribution of storm total lightning density also highlights the pronounced decay in convective intensity of storm elements east of the central Appalachians during the period after 2300 UTC 4 October.

Fig. 12.

Contour map of storm total lightning flash density strikes (km−2) over the study region.

Fig. 12.

Contour map of storm total lightning flash density strikes (km−2) over the study region.

System C formed to the rear of the long-lived line (system D) but was still in the warm sector ahead of the cold front. The system initiated before 2300 UTC as a region of isolated convective elements (Fig. 13); from 0000 to 0200 UTC, it organized into multioriented lines of convection (Fig. 14). By 0300 UTC, the system had assumed leading line–trailing stratiform structure (Fig. 14).

Fig. 13.

The 15-min Hydro-NEXRAD fields’ rainfall (mm) for 2300 UTC (2245–2300 UTC) 4 Oct.

Fig. 13.

The 15-min Hydro-NEXRAD fields’ rainfall (mm) for 2300 UTC (2245–2300 UTC) 4 Oct.

Fig. 14.

The 15-min Hydro-NEXRAD fields’ rainfall (mm) for (top left) 0100 UTC (i.e., 0045–0100 UTC), (top right) 0200 UTC, (bottom left) 0300 UTC, and (bottom right) 0400 UTC 5 Oct observed at KCCX and KDIX (see Fig. 2).

Fig. 14.

The 15-min Hydro-NEXRAD fields’ rainfall (mm) for (top left) 0100 UTC (i.e., 0045–0100 UTC), (top right) 0200 UTC, (bottom left) 0300 UTC, and (bottom right) 0400 UTC 5 Oct observed at KCCX and KDIX (see Fig. 2).

System C was responsible for heavy rainfall over the urban corridor of the New York–New Jersey metropolitan area. The leading line–trailing stratiform system developed a large area with rain rates exceeding 25 mm h−1. By 0200 UTC (Fig. 14), this area exceeded 2500 km2 and the rain area exceeding 2 mm h−1 was 13 000 km2. The corresponding maxima for the first system were 1600 and 16 000 km2. The rain area exceeding 25 mm h−1 at 0300 UTC had changed little from the preceding hour, but the total rain area grew to approximately 20 000 km2. Heavy rainfall over the region extending from Princeton to New York City (Fig. 1) during the period from 0300 to 0430 UTC was associated with a sharp drop in convective rain area. From 0300 to 0400 UTC, the rain area exceeding 25 mm h−1 decreased from more than 2500 km2 at 0300 UTC to less than 1000 km2 at 0400 UTC.

c. Microphysics of extreme rainfall rates

The links between storm structure and extreme rainfall rates are examined in this section using radar, disdrometer, and lidar observations from the New York–New Jersey region. The temporal variability in extreme rainfall rates at time scales ranging from 1–60 min plays a central role in determining flood hazards in many urban environments (Ogden et al. 2000; Smith et al. 2002, 2005; Zhang and Smith 2003).

The leading line–trailing stratiform system (system C), which produced extreme rainfall rates over the urban corridor of the NY–NJ region, exhibited pronounced contrasts in structure and convective intensity to the earlier system. Convective intensity was weak relative to the maximum intensity of system D (Fig. 15; see also the CG lightning time series of Fig. 5). Reflectivity values greater than 35 dBZ barely exceeded 5 km (Fig. 15) for system C. In contrast to system D, which produced large CG lightning flash rates, there was virtually no lightning in system C, even in the leading line of convective precipitation (Fig. 14).

Fig. 15.

Radar reflectivity (dBZ) field at 3-km AGL for the mesoscale rain area D (see Fig. 9) depicted by (top) radar observations, and (bottom) vertical cross section of radar reflectivity (dBZ) along a line perpendicular to the convective line and passing through the element D.

Fig. 15.

Radar reflectivity (dBZ) field at 3-km AGL for the mesoscale rain area D (see Fig. 9) depicted by (top) radar observations, and (bottom) vertical cross section of radar reflectivity (dBZ) along a line perpendicular to the convective line and passing through the element D.

System C produced extreme short-term rainfall rates in Princeton during the period from 0330 through 0430 UTC (Figs. 1 and 16). Peak 1-min rainfall rates in Princeton computed from disdrometer observations reached 125 mm h−1 (0401 UTC), with three rain rate spikes above 75 mm h−1 between 0340 and 0402 UTC. Extreme rainfall rates over Princeton were characterized by large drop arrival rates (Fig. 16) for each of the three rainfall rate peaks. The peak drop arrival rate of 5200 m−2 s−1 at 0343 UTC is one of the largest observed at the Princeton disdrometer site during a 2-yr period of observation (Smith et al. 2008). There was pronounced variation over the three peaks in the size and number of raindrops. The largest drop arrival rate and smallest mean diameter was for the first spike, with decreasing drop arrival rates and increasing values of mean diameter for the following two peaks (in contrast with many leading line–trailing stratiform systems in which a “large drop” zone at the leading line of convection is a common feature; see Uijlenhoet et al. 2003).

Fig. 16.

Time series of (a) rainfall rate (mm h−1), (b) mean drop diameter (mm), and (c) drop arrival rate (m−2 s−1) computed from raindrop size distribution measurements from a Joss–Waldvogel disdrometer in Princeton, NJ (see Fig. 2).

Fig. 16.

Time series of (a) rainfall rate (mm h−1), (b) mean drop diameter (mm), and (c) drop arrival rate (m−2 s−1) computed from raindrop size distribution measurements from a Joss–Waldvogel disdrometer in Princeton, NJ (see Fig. 2).

The temporal distribution of rainfall rate over Princeton from 0330 to 0430 UTC (Fig. 16) can be related to the vertical structure of the storm at 0400 UTC (Fig. 15) along a line perpendicular to the line of convection and reflecting the direction of motion over the Princeton disdrometer. The initial spike of the rainfall rate (0343 UTC in Fig. 16) was associated with weak convection ahead of the main convective line (represented by the storm structure at 0400 UTC around location 75 km in Fig. 15). The broader spike of extreme rainfall rates (centered around 0353 UTC) was produced by the most intense convection (locations 64–70 km in Fig. 15). The peak rainfall rate at 0401 UTC was associated with trailing region of enhanced low-level reflectivity (locations 58–61 km). A region of elevated stratiform rain rates (0408–0420 UTC) is linked to the region with strong brightband amplification of reflectivity (36–48-km location).

Lidar observations from the CCNY system in Manhattan were used, along with model analyses of evolution of vertical profiles of relative humidity, to examine aerosol properties in the NY–NJ corridor for the 4–5 October storm. In the prestorm environment, the mean aerosol backscatter coefficient (Kovalev and Eichinger 2004) between 2 and 4 km increased by a factor of more than 5 during a 3-h period from 1800 to 2100 UTC on 4 October 2006 (Figs. 17 and 18). In the corresponding time domain, WRF simulations of relative humidity profiles over CCNY indicate a rapid increase in relative humidity over the lowest 4 km of the atmosphere (figure not shown).

Fig. 17.

Vertical profiles of aerosol backscatter coefficient (km−1 sr−1) derived from lidar-received backscatter power over CCNY at a wavelength of 1064 nm at 1630, 1800, and 1930 UTC on 4 Oct 2006.

Fig. 17.

Vertical profiles of aerosol backscatter coefficient (km−1 sr−1) derived from lidar-received backscatter power over CCNY at a wavelength of 1064 nm at 1630, 1800, and 1930 UTC on 4 Oct 2006.

Fig. 18.

Time series of mean aerosol backscatter coefficient (km−1 sr−1) in the layer between 2 and 4 km over Manhattan on 4 Oct 2006.

Fig. 18.

Time series of mean aerosol backscatter coefficient (km−1 sr−1) in the layer between 2 and 4 km over Manhattan on 4 Oct 2006.

Lidar observations and WRF model simulations suggest that hygroscopic growth (Feingold and Morley 2003) played an important role in shaping the population of cloud condensation nuclei (CCN) for the 4–5 October storm systems over the urban corridor of the NY–NJ region. An important related issue concerns the role of CCN in promoting efficient precipitation mechanisms that maximize rainfall rates given the inflow of water vapor to the storm system. Efficient precipitation mechanisms can be especially important for extreme rainfall rates in settings with relatively weak convection (see, e.g., Petersen et al. 1999).

The 4–5 October 2006 storm was typical of flash flood–producing storms exhibiting pronounced temporal variability of rainfall rates at 1–60-min time scales (see Petersen et al. 1999; Ogden et al. 2000; Smith et al. 2005). This variability of point rainfall rates over time is closely linked to the structure and evolution of storm elements, such as system C for the 4–5 October storm. Microphysical processes, especially as they relate to anthropogenically altered aerosol populations, play a critical, but poorly understood, role in determining the climatology of extreme rainfall rates in urban environments.

4. Summary and conclusions

Observational and numerical model analyses of precipitation associated with a frontal zone that passed through the New York–New Jersey region on 4–5 October 2006 are used to examine the temporal and spatial variability of flash flood–producing rainfall in the urban corridor of the northeastern United States. The principal conclusions of the study are as follows:

  1. Regional radar analyses show that rainfall over the NY–NJ urban corridor on 4–5 October 2006 was produced by two mesoscale rain areas. The first (system D) was a long-lived line of convective rainfall with associated stratiform rain area on the northeastern end of the convective line. It produced light rainfall over the NY–NJ region from 0000 to 0200 UTC 5 October after a history of producing locally heavy rainfall rates during its path through much of Pennsylvania. The second mesoscale rain area (system C) evolved into a leading line–trailing stratiform structure and produced extreme short-term rain rates over the NY–NJ region from 0300 to 0500 UTC 5 October.

  2. Simulations with ARW-WRF capture the broad features of the storm environment. They do not, however, capture the structure, motion, and evolution of the two mesoscale rain areas. The evolution of the convective line of rainfall and associated stratiform rain area for system D was influenced by orographic mechanisms in the Appalachian region. The mesoscale rain area weakened east of the Appalachian Mountains as it propagated across the NY–NJ region and approached the Atlantic Coast. System C developed to the rear of system D from embedded convective elements that formed in the central Appalachians. Heterogeneities in rainfall distribution associated with the complex terrain of the central Appalachians and the land–water boundary at the Atlantic coast play an important role in the extreme rainfall climatology of the NY–NJ urban corridor.

  3. There are pronounced temporal heterogeneities in extreme rainfall over the NY–NJ urban corridor that are tied to the diurnal cycle of convection (see Ntelekos et al. 2007 for related analyses). The evolution of convection for the 4–5 October storm system was characterized by a sharp peak in CG lightning flash density between 2200 and 2300 UTC 4 October. Equally striking is the 2200 UTC peak in convective rain area for system D based on Hydro-NEXRAD rainfall analyses. Peak rainfall rates over the NY–NJ urban corridor occurred well after the peak in convective intensity of the larger storm system and were associated with a rapidly decaying phase of system C.

  4. The temporal variability of rainfall rate during the period of heavy rainfall over the NY–NJ urban corridor was closely linked to the leading line–trailing stratiform structure of system C. Disdrometer observations show that there are pronounced contrasts in drop arrival rates and characteristic drop sizes that are tied to the structural features of the mesoscale rain area.

  5. Lidar observations suggest that hygroscopic growth plays an important role in determining the CCN population associated with extreme rainfall rates over the NY–NJ region.

The long-term objective of this research is to develop an enhanced understanding of flash-flood hazards in the urban corridor of the eastern United States. Analyses of the 4–5 October 2006 storm provide insights into the processes that control the distribution of heavy rainfall from an important class of flash flood–producing storms in the northeastern United States. Future research will place the case-study analyses presented in this paper in the context of a flash-flood climatology for the urban corridor of the New York–New Jersey region. We will also examine the effects of urbanization on flash flood–producing storm systems in the region, with a particular focus on aerosol effects on warm-season thunderstorm systems.

Acknowledgments

The authors thank the WRF-help team for providing valuable feedback on various issues relating to ARW-WRF. NLDN data were provided by the NASA Lightning Imaging Sensor (LIS) instrument team and the LIS data center via the Global Hydrology Resource Center (GHRC) located at the Global Hydrology and Climate Center (GHCC), Huntsville, Alabama, through a license agreement with Global Atmospherics, Inc (GAI). The data available from the GHRC are restricted to LIS science team collaborators and to NASA EOS and TRMM investigators. This research was supported in part by the National Science Foundation (Grants EEC-0540832, ITR-0427325, and BES-0607036), NASA, the NOAA Cooperative Institute for Climate Studies, and NOAA Grant NA17AE1625. The views and findings contained in this report are those of the authors and should not be construed as an official NOAA position, policy, or decision.

REFERENCES

REFERENCES
Austin
,
P. M.
, and
R. A.
Houze
,
1972
:
Analysis of the structure of precipitation patterns in New England.
J. Appl. Meteor.
,
11
,
926
935
.
Barros
,
A. P.
, and
R. J.
Kuligowski
,
1998
:
Orographic effects during a severe wintertime rainstorm in the Appalachian Mountains.
Mon. Wea. Rev.
,
126
,
2648
2672
.
Ebisuzaki
,
W.
,
J.
Alpert
,
J.
Wang
,
D.
Jovic
, and
P.
Shafran
,
2004
:
North American Regional Reanalysis: End user access to large data sets.
Preprints, 20th Int. Conf. on Interactive Information and Processing Systems (IIPS) for Meteorology, Oceanography, and Hydrology, Seattle, WA, Amer. Meteor. Soc., 8.2. [Available online at http://ams.confex.com/ams/84Annual/techprogram/paper_73708.htm]
.
Feingold
,
G.
, and
B.
Morley
,
2003
:
Aerosol hygroscopic properties as measured by lidar and comparison with in situ measurements.
J. Geophys. Res.
,
108
,
4327
.
doi:10.1029/2002JD002842
.
Holt
,
T.
, and
J.
Pullen
,
2007
:
Urban canopy modeling of the New York City metropolitan area: A comparison and validation of single- and multilayer parameterizations.
Mon. Wea. Rev.
,
135
,
1931
1944
.
Kovalev
,
V. A.
, and
W. E.
Eichinger
,
2004
:
Elastic Lidar: Theory, Practice, and Analysis Methods.
John Wiley and Sons, 615 pp
.
Krajewski
,
W. F.
, and
Coauthors
,
2007
:
Towards better utilization of NEXRAD data in hydrology: An overview of Hydro-NEXRAD.
World Environmental and Water Resources Congress: Restoring Our Natural Habitat, Vol. 243, American Society of Civil Engineers, 1–8
.
Lin
,
Y-L.
,
R. D.
Farley
, and
H. D.
Orville
,
1983
:
Bulk parameterization of the snow field in a cloud model.
J. Climate Appl. Meteor.
,
22
,
1065
1092
.
Ntelekos
,
A. A.
,
J. A.
Smith
, and
W. F.
Krajewski
,
2007
:
Climatological analyses of thunderstorms and flash floods in the Baltimore metropolitan region.
J. Hydrometeor.
,
8
,
88
101
.
Ntelekos
,
A. A.
,
J. A.
Smith
,
M. L.
Baeck
,
W. F.
Krajewski
,
A. J.
Miller
, and
R.
Goska
,
2008
:
Extreme hydrometeorological events and the urban environment: Dissecting the 7 July 2004 thunderstorm over the Baltimore MD metropolitan region.
Water Resour. Res.
,
44
,
W08446
.
doi:10.1029/2007WR006346
.
Ogden
,
F. L.
,
H. O.
Sharif
,
U. S.
Senarath
,
J. A.
Smith
,
M. L.
Baeck
, and
J. R.
Richardson
,
2000
:
Hydrologic analysis of the Fort Collins, Colorado, flash flood of 1997.
J. Hydrol.
,
228
,
82
100
.
Orville
,
R. E.
, and
G. R.
Huffines
,
2001
:
Cloud-to-ground lightning in the United States: NLDN results in the first decade, 1989–98.
Mon. Wea. Rev.
,
129
,
1179
1193
.
Petersen
,
W. A.
, and
Coauthors
,
1999
:
Mesoscale and radar observations of the Fort Collins flash flood of 28 July 1997.
Bull. Amer. Meteor. Soc.
,
80
,
191
216
.
Pullen
,
J.
,
T.
Holt
,
A. F.
Blumberg
, and
R. D.
Bornstein
,
2007
:
Atmospheric response to local upwelling in the vicinity of the New York–New Jersey harbor.
J. Appl. Meteor. Climatol.
,
46
,
1031
1052
.
Reed
,
S. M.
, and
D. R.
Maidment
,
1999
:
Coordinate transformations for using NEXRAD data in GIS-based hydrologic modeling.
J. Hydrol. Eng.
,
4
,
174
182
.
Rogers
,
E.
,
T. L.
Black
,
D. G.
Deaven
,
G. J.
DiMego
,
Q.
Zhao
,
M.
Baldwin
,
N. W.
Junker
, and
Y.
Lin
,
1996
:
Changes to the operational “early” Eta analysis/forecast system at the National Centers for Environmental Prediction.
Wea. Forecasting
,
11
,
391
413
.
Shea
,
D. J.
,
S. J.
Worley
,
I. R.
Stern
, and
T. J.
Hoar
,
1995
:
An introduction to atmospheric and oceanographic datasets.
NCAR Tech. Note NCAR/TN-404+IA, 132 pp
.
Shepherd
,
J. M.
,
2005
:
A review of the current investigations of urban-induced rainfall and recommendations for the future.
Earth Interactions
,
9
.
[Available online at http://EarthInteractions.org]
.
Skamarock
,
W. C.
,
J. B.
Klemp
,
J.
Dudhia
,
D. O.
Gill
,
D. M.
Barker
,
W.
Wang
, and
J. G.
Powers
,
2007
:
A description of the Advanced Research WRF Version 2.
NCAR Tech. Note NCAR/TN-468+STR, 100 pp
.
Smith
,
J. A.
, and
A. F.
Karr
,
1990
:
A statistical model of extreme storm rainfall.
J. Geophys. Res.
,
95
, (
D3
).
2083
2092
.
Smith
,
J. A.
,
M. L.
Baeck
,
J. E.
Morrison
,
P. L.
Sturdevant-Rees
,
D. F.
Turner-Gillespie
, and
P. D.
Bates
,
2002
:
The regional hydrology of extreme floods in an urbanizing drainage basin.
J. Hydrometeor.
,
3
,
267
282
.
Smith
,
J. A.
,
A. J.
Miller
,
M. L.
Baeck
,
P. A.
Nelson
,
G. T.
Fisher
, and
K. L.
Meierdiercks
,
2005
:
Extraordinary flood response of a small urban watershed to short duration convective rainfall.
J. Hydrometeor.
,
6
,
599
617
.
Smith
,
J. A.
,
E.
Hui
,
M.
Steiner
,
M. L.
Baeck
,
W. F.
Krajewski
, and
A. A.
Ntelekos
,
2008
:
Variability of rainfall rate and raindrop size distributions in heavy rain.
Water Resour. Res.
,
45
,
W04430
.
doi:10.1029/2008WR006840
.
Steiner
,
M.
, and
J. A.
Smith
,
1998
:
Convective versus stratiform rainfall: An ice microphysical and kinematic conceptual model.
Atmos. Res.
,
48
,
317
326
.
Steiner
,
M.
,
R. A.
Houze
Jr.
, and
S. E.
Yuter
,
1995
:
Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data.
J. Appl. Meteor.
,
34
,
1978
2007
.
Uijlenhoet
,
R.
,
M.
Steiner
, and
J. A.
Smith
,
2003
:
Variability of raindrop size distributions in a squall line and implications for radar rainfall estimation.
J. Hydrometeor.
,
4
,
43
61
.
Waymire
,
E.
,
V. K.
Gupta
, and
I.
Rodriquez-Iturbe
,
1984
:
A spectral theory of rainfall intensity at the meso-β scale.
Water Resour. Res.
,
20
,
1453
1465
.
Zhang
,
Y.
, and
J. A.
Smith
,
2003
:
Space-time variability of rainfall and extreme flood response in the Menomonee River basin, Wisconsin.
J. Hydrometeor.
,
4
,
506
517
.

Footnotes

Corresponding author address: James A. Smith, Dept. of Civil and Environmental Engineering, Princeton University, C319 E-Quad, Princeton, NJ 08544. Email: jsmith@princeton.edu