Forecasting Precipitation Distributions. Associated with Cool-Season 500-hPa Cutoff Cyclones. in the Northeastern United States.

Transcription

1 Forecasting Precipitation Distributions Associated with Cool-Season 500-hPa Cutoff Cyclones in the Northeastern United States Abstract of a thesis presented to the Faculty of the University at Albany, State University of New York in partial fulfillment of the requirements for the degree of Master of Science College of Arts & Sciences Department of Atmospheric and Environmental Sciences Melissa D. Payer 2010

2 ABSTRACT Forecasting precipitation distributions associated with cool-season 500-hPa cutoff cyclones can be a challenge in the Northeast United States (US). Although the structure and evolution of cutoff cyclones have been extensively documented, forecasting precipitation associated with cutoff cyclones remains difficult, given that cutoff cyclones are generally slow moving and can have varying precipitation distributions throughout their lifetimes. The purpose of this thesis is to identify key synoptic-scale and mesoscale features that differentiate between various precipitation distributions associated with cool-season 500-hPa cutoff cyclones in the Northeast US. The results of this thesis provide tools to increase situational awareness in an attempt to improve future precipitation forecasts associated with cutoff cyclones in the Northeast US. Eight years ( ) of 500-hPa cutoff cyclones in the Northeast US are examined to determine the influence of ENSO and the MJO on cutoff cyclone frequency. Cutoff cyclones are defined as cyclones that maintain a 30-m geopotential height rise in all directions at 500 hpa for at least three consecutive analysis times (i.e., a 12-h period). Statistically significant results were obtained only when considering the combined influence of ENSO and the MJO on cutoff cyclone frequency, with cutoff cyclone frequency maximized when enhanced convection associated with the MJO was over the Maritime Continent during ENSO cooling or over the Western Hemisphere during ENSO warming. This thesis presents cyclone-relative composites of 384 cutoff cyclone days that occurred in the Northeast US during the 2004/ /09 cool seasons. Cutoff cyclone ii

3 days were placed into 15 composite categories according to precipitation amount and the tilt and structure of the cutoff cyclone at 500 hpa. The average location of cutoff cyclones within each of the composite categories indicates that there was a distinct difference in location of cutoff cyclones between precipitation amount categories, with cutoff cyclones associated with heavy precipitation (>25 mm) typically located west of the Northeast US. Schematic diagrams depicting the key synoptic-scale features that affect precipitation distributions, including upper-level and low-level jet streaks, midlevel absolute vorticity maxima, surface fronts, and regions of preexisting moisture, are presented for each of the composite categories. Case study analyses of three cutoff cyclone events that were associated with precipitation forecasting challenges and varying precipitation distributions are conducted. The events occurred on: 2 3 February 2009, 1 4 January 2010, and March The 2 3 February 2009 cutoff cyclone event was associated with light precipitation (<5 mm) throughout the Northeast US, while the 1 4 January 2010 and the March 2010 cutoff cyclone events were both long-duration cutoff cyclones associated with heavy precipitation (>25 mm) and varying daily precipitation distributions. Synopticscale features, including upper-level and low-level jet streaks, midlevel absolute vorticity maxima, regions of temperature advection, regions of low-level frontogenesis, and regions of moisture advection, for each day of the cutoff cyclone events are examined to determine their respective roles in contributing to the observed precipitation distributions. In addition, modification of the low-level flow by the topography of the Northeast US and lake-effect precipitation played a role in altering the mesoscale distributions of precipitation associated with the latter two cutoff cyclone events and is also considered. iii

4 Forecasting Precipitation Distributions Associated with Cool-Season 500-hPa Cutoff Cyclones in the Northeastern United States A thesis presented to the Faculty of the University at Albany, State University of New York in partial fulfillment of the requirements for the degree of Master of Science College of Arts & Sciences Department of Atmospheric and Environmental Sciences Melissa D. Payer 2010

5 ACKNOWLEDGEMENTS I would first like to thank my co-advisors, Lance Bosart and Dan Keyser, for providing me with the opportunity to pursue this research. Their invaluable guidance throughout the research process helped me develop the necessary knowledge and skills that will undoubtedly prove to be invaluable in my professional career. I would also like to thank National Weather Service (NWS) focal points, Neil Stuart and Tom Wasula from the Albany, NY, NWS forecast office. Their enthusiasm and feedback helped focus this research toward achieving forecasting applications. Other NWS personnel that contributed to this research include Dan St. Jean and Paul Sisson from the Gray, ME, and Burlington, VT, NWS forecast offices, respectively. This research was supported by the National Oceanic and Atmospheric (NOAA) Grant NA07NWS , awarded to the University at Albany, SUNY, as part of the CSTAR program. I would like to thank my fellow graduate students, especially Matt Scalora, who was especially instrumental in the beginning stages of this research by proving me with an established methodology as well as several scripts. I am also grateful to Jonas Asuma, Jay Cordeira, Heather Archambault, Nick Metz, Alan Srock, Tom Galarneau, Ben Moore, Tim Melino, and Scott Sukup for their research suggestions and their assistance with technical issues. I would like to acknowledge the faculty and staff of the Department of Atmospheric and Environmental Sciences, including: Patricia Seguin, Maria Moon, and Barbara Zampella for their administrative assistance; Kevin Tyle for helping me with GEMPAK and other technical issues; and my professors who provided me with the background knowledge necessary to conduct this research. Finally, I would like to thank my family and my boyfriend, Steve, for their continual love and support throughout my time at the University at Albany. Their words of encouragement have helped me stay motivated in order to complete this research. v

6 TABLE OF CONTENTS ABSTRACT... ii ACKNOWLEDGEMENTS... v TABLE OF CONTENTS vi LIST OF TABLES.. x LIST OF FIGURES... xi 1. Introduction Overview Literature Review Overview of Cutoff Cyclones Cutoff Cyclones and Precipitation Forecast Issues Associated with Cutoff Cyclones The Influence of ENSO and the MJO Study Goals Data and Methodology Data Sources The Influence of ENSO and the MJO Cutoff Cyclone Composites Case Study Analyses Methodology The Influence of ENSO and the MJO Standardized Anomalies.. 24 vi

7 2.2.3 Cutoff Cyclone Composites Case Study Analyses Results: The Influence of ENSO and the MJO on Cutoff Cyclone Frequency The Influence of ENSO Regional Influence of ENSO The Influence of ENSO on Cutoff Cyclone Frequency The Influence of the MJO Regional Influence of the MJO The Influence of the MJO on Cutoff Cyclone Frequency The Combined Influence of ENSO and the MJO Regional Influence of ENSO and the MJO Influence of ENSO and the MJO on Cutoff Cyclone Frequency Results: Cutoff Cyclone Composites Average Location of Cutoff Cyclones Cyclone-relative composites HP Cutoff Cyclones a In-depth Examination of HP Neutral Cutoff Category b Schematic Diagrams of HP Cutoff Cyclones LP Cutoff Cyclones NP Cutoff Cyclones Results: Case Study Analyses of Three Cutoff Cyclone Events The 2 3 February 2009 Cutoff Cyclone Event Event Overview vii

8 5.1.2 Meteorological Conditions Conceptual Summary The 1 4 January 2010 Cutoff Cyclone Event Event Overview Meteorological Conditions: 2 January Meteorological Conditions: 3 January Conceptual Summary The March 2010 Cutoff Cyclone Event Event Overview Meteorological Conditions: 13 March Meteorological Conditions: 14 March Conceptual Summary Summary and Discussion The Influence of ENSO and the MJO on Cutoff Cyclone Frequency Summary Comparison to Previous Work Cutoff Cyclone Composites Summary Comparison to Previous Work Case Study Analyses of Three Cutoff Cyclone Events Summary Comparison to Previous Work Forecasting Applications viii

9 7. Conclusions and Future Work Conclusions Suggestions for Future Work REFERENCES ix

10 LIST OF TABLES Table I. Significance levels based on the standard deviations from normal for a normal distribution. (Table and caption from Grumm and Hart 2001, Table 1). Table II. Total number and percentage of weeks characterized by each ENSO phase for Table III. Total number and percentage of weeks characterized by each phase of the MJO for Table IV. Total number of weeks characterized by each phase of the MJO for ENSO cooling, steady, and warming for A total of 159 weeks that occurred when the MJO was weak were not included. x

11 LIST OF FIGURES Fig Schematic representation of the development of a cold vortex as shown by the height contours (solid lines) and isotherms (dashed lines) on an isobaric surface in the middle troposphere. (Figure and caption from Hsieh 1949, Figs. 11a c.) Fig Sample 500-hPa geopotential height analyses illustrating the objective method used to identify closed circulation centers. (a) Three sample radial arms used to identify a 30 m closed contour around the cyclone center point A. Geopotential heights rise to at least 30 m larger than that of the point A before decreasing along each radial arm. Point A is therefore identified as a closed cyclonic circulation center. (b) As in (a) except that geopotential heights along the radial arm do not exceed 30 m higher than the point A before decreasing. Point A is therefore not identified as a closed circulation center. (Figure and caption from Bell and Bosart 1989, Fig. 1a,b.) Fig Schematic of a PV-θ contour in an Atlantic storm track sharing its main characteristics with (a) an LC1-type life cycle and (b) an LC2-type life cycle. The dashed line marks the approximate position of the mean jet at each stage. (Figure and caption from Thorncroft et al. 1993, Fig. 12a,b.) Fig Schematic representation of the distribution of precipitation relative to the isobaric contours (solid lines) of the surface of the cold dome. Relatively light precipitation occurs in the stippled area, with heavier precipitation occurring in the hatched area. (Figure and caption from Hsieh 1949, Fig. 13.) Fig Areas of maximum frequency of occurrence of measurable precipitation with the most intense lows (Class III) centered at the origin for 850-, 700-, 500-, and 300-hPa levels. The symmetrical circles represent idealized contours about the low center at any level. (Figure and caption from Klein et al. 1968, Fig. 8.) Fig Threat scores for the forecasters 0.50-, 1.00-, and 2.00-in. forecasts for day 1 from 1961 through (Figure and caption from Fritsch et al. 1998, Fig. 1.) Fig Annual threat scores of 24-h 1-in. Day 1 QPF since 1993 for the NAM (green), GFS (blue), and HPC forecasters (red). (Figure from hpcverif.shtml.) Fig. 1.8: Key mountain ranges (blue) and valleys (yellow) of the Northeast US. Fig Longitude latitude schematic of the day 15 σ 0.24 meridional wind perturbation for the heating on a December February zonal flow. The contour interval is 0.5 m s 1. The zero contour is not shown, and negative contours are dashed. (Figure and caption from Jin and Hoskins 1995, Fig. 8.) xi

12 Fig MJO phase diagram depicting the approximate locations of the enhanced convective signal of the MJO for each phase. Weak MJO activity is represented by the inner circle. (Figure from Wheeler and Hendon 2004, Fig. 7.) Fig Northeast cutoff cyclone domain (red outline) and precipitation domain (green outline). Fig Annual frequency of cool-season 500-hPa cutoff cyclones in the Northeast US. Fig Monthly frequency of 500-hPa cutoff cyclones in the Northeast US by cool season. Fig Histogram of the duration of cool-season 500-hPa cutoff cyclone events occurring within the Northeast US during 2004/ /09. Fig Schematic used to assign a tilt classification to each cutoff cyclone day. (Figure from Scalora 2009, Fig. 2.3.) Fig Number of cutoff cyclone days included in each composite category. Colors indicate the daily precipitation amount: heavy precipitation (blue), light precipitation (green), or no precipitation (red). Fig 3.1. Composites of hPa geopotential height anomaly (m, shaded) during weeks when ENSO was (a) cool and cooling, (b) cool and steady, (c) cool and warming, (d) neutral and cooling, (e) neutral and steady, (f) neutral and warming, (g) warm and cooling, (h) warm and steady, and (i) warm and warming. Fig Number of 500-hPa cutoff cyclones by ENSO phase for Error bars were determined using the bootstrap method and enclose the 95% confidence interval. Fig As in Fig. 3.2 except normalized to account for the total number of weeks characterized by each ENSO phase for Fig Composites of interpolated OLR anomaly (W m 2, shaded) during weeks when the MJO was in (a) phase 1, (b) phase 2, (c) phase 3, (d) phase 4, (e) phase 5, (f) phase 6, (g) phase 7, and (h) phase 8. The red X represents the estimated location of enhanced convection associated with each phase of the MJO as determined by Wheeler and Hendon (2004). A total of 159 weeks were removed during which the MJO was weak (i.e., amplitude < 1). Fig Composites of hPa geopotential height anomaly (shaded, m) during weeks when the MJO was in (a) phase 1, (b) phase 2, (c) phase 3, (d) phase 4, (e) phase 5, (f) phase 6, (g) phase 7, and (h) phase 8. The red X represents the estimated location of enhanced convection associated with each phase of the MJO as determined by Wheeler and Hendon (2004). A total of 159 weeks were removed during which the MJO was weak (i.e., amplitude < 1). xii

13 Fig Number of 500-hPa cutoff cyclones by phase of the MJO for Cutoff cyclones that occurred when the MJO was weak (i.e., amplitude < 1) were removed. Error bars were determined using the bootstrap method and enclose the 95% confidence interval. Fig As in Fig. 3.5 except for weeks when the ENSO trend was cooling. Fig As in Fig. 3.5 except for weeks when the ENSO trend was warming. Fig Number of 500-hPa cutoff cyclones for each phase of the MJO for ENSO warming and cooling for Cutoff cyclones that occurred when the MJO was weak (i.e., amplitude < 1) were removed. Error bars were determined using the bootstrap method and enclose the 95% confidence interval. Fig Average location of 500-hPa cutoff cyclones for each composite category. Color shading represents precipitation amount categories with blue, green, and red shading representing heavy precipitation (HP), light precipitation (LP), and no precipitation (NP) composite categories, respectively. Closed and open circles represent cutoff and trough composite categories, respectively. The brown bold lines depict the Northeast cutoff cyclone domain and the black bold line depicts the Northeast precipitation domain. Fig Cyclone-relative composite for the HP neutral cutoff composite category showing 500-hPa geopotential height (dam, solid black contours) and absolute vorticity (10 5 s 1, shaded) and 850-hPa potential temperature (K, dashed red contours) and wind (>25 kt, barbs). The number of cutoff cyclone days (n) included within the composite is indicated in the top right corner. Fig Cyclone-relative composite for the HP neutral cutoff composite category showing MSLP (hpa, solid black contours), PW (mm, dashed red contours), standardized anomaly of PW (σ, shaded), and 250-hPa wind (kt, barbs). Green solid lines outline regions of 250-hPa wind greater than 70 and 90 kt. The number of cutoff cyclone days (n) included within the composite is indicated in the top right corner. Fig Schematic diagrams depicting the cyclone-relative 500-hPa geopotential height (dam, solid black contours) and key synoptic-scale features that contribute to precipitation for the (a) HP negative cutoff, (b) HP negative trough, (c) HP neutral cutoff, (d) HP neutral trough, (e) HP positive cutoff, and (f) HP positive trough composite categories. The red X represents the location of the 500-hPa absolute vorticity maximum. The brown bold line depicts the Northeast precipitation domain. The number of cutoff cyclone days (n) included within each composite is indicated in the top right corner of each panel. Fig As in Fig. 4.4 except for the (a) LP negative cutoff, (b) LP negative trough, (c) LP neutral cutoff, (d) LP neutral trough, (e) LP positive cutoff, and (f) LP positive trough composite categories. xiii

14 Fig As in Fig. 4.2 except for the LP negative cutoff composite category. Fig As in Fig. 4.4 except for the (a) NP negative, (b) NP neutral, and (c) NP positive composite categories. Fig NCEP Global Ensemble Forecast System valid at 1200 UTC 3 February 2009 and initialized at (a,b) 0000 UTC 30 January, (c,d) 1200 UTC 30 January, and (e,f) 1800 UTC 30 January. Left panels show 1008 and 1020 hpa MSLP isobars of each member (hpa, colored contours), the mean of all members (hpa, black contour), and the spread about the mean (hpa, shaded). Right panels show the average MSLP isobars of all members (hpa, green contours) and the standardized anomalies computed from the mean (σ, shaded). (Figure modified from Grumm et al. 2009, Fig. 13.) Fig Mean 500-hPa geopotential height (dam, black contours) for 0000 UTC 3 February 0000 UTC 4 February 2009 and the track of the 500-hPa cutoff cyclone center every 6 h (red contours). The brown bold lines depict the Northeast cutoff cyclone domain. Fig Two-day NPVU QPE (mm, shaded) ending 1200 UTC 4 February The black bold line depicts the Northeast precipitation domain. Fig hPa geopotential height (dam, black solid contours), wind speed (m s 1, shaded), and divergence (10 5 s 1, red dashed contours) at (a) 1800 UTC 3 February 2009 and (b) 0000 UTC 4 February Fig hPa geopotential height (dam, black solid contours), absolute vorticity (10 5 s 1, shaded), cyclonic absolute vorticity advection [10 5 s 1 (3 h) 1, blue dashed contours] and wind (kt, barbs) at (a) 1800 UTC 3 February 2009 and (b) 0000 UTC 4 February Fig hPa geopotential height (dam, black solid contours), temperature ( C, shaded), and wind (>30 kt, barbs) at 1800 UTC 3 February Fig hPa geopotential height (dam, black solid contours), temperature ( C, green dashed contours), Q vectors (>5 x 10 7 Pa m 1 s 1, arrows), and Q-vector convergence (10 12 Pa m 2 s 1, shaded) at (a) 1800 UTC 3 February 2009 and (b) 0000 UTC 4 February Fig MSLP (hpa, black solid contours), hpa thickness (m, red dashed contours), and PW (mm, shaded) at 1800 UTC 3 February Fig Schematic depicting the 500-hPa geopotential height (dam, black contours) at 1800 UTC 3 February 2009 and key synoptic-scale features that contribute to precipitation for the 2 3 February 2009 cutoff cyclone event. The brown bold line depicts the Northeast precipitation domain. xiv

15 Fig Mean 500-hPa geopotential height (dam, black contours) for 0000 UTC 2 January 1200 UTC 4 January 2010 and the track of the 500-hPa cutoff cyclone center every 6 h (red contours). The brown bold lines depict the Northeast cutoff cyclone domain. Fig Four-day NPVU QPE (mm, shaded) ending 1200 UTC 5 January The black bold line depicts the Northeast precipitation domain. Fig h NPVU QPE (mm, shaded) ending (a) 1200 UTC 2 January 2010, (b) 1200 UTC 3 January 2010, (c) 1200 UTC 4 January 2010, and (d) 1200 UTC 5 January The black bold line depicts the Northeast precipitation domain. Fig As in Fig. 5.4 except at (a) 0000 UTC 3 January 2010 and (b) 0600 UTC 3 January Fig hPa geopotential height (dam, black contours) and standardized anomalies of 250-hPa zonal wind (σ, shaded) at 0000 UTC 3 January Fig As in Fig. 5.5 except at (a) 0000 UTC 3 January 2010 and (b) 0600 UTC 3 January Fig As in Fig. 5.6 except at 0000 UTC 3 January Fig hPa geopotential height (dam, black solid contours), temperature ( C, green dashed contours), Q vectors (>5 x 10 7 Pa m 1 s 1, arrows), and Q-vector convergence (10 12 Pa m 2 s 1, shaded) at (a) 0000 UTC 3 January 2010 and (b) 0600 UTC 3 January Fig hPa frontogenesis [K (100 km) 1 (3 h) 1, shaded], potential temperature (K, black solid contours), and wind (kt, barbs) at (a) 0000 UTC 3 January 2010 and (b) 0600 UTC 3 January 2010; (c) cross section of 925-hPa frontogenesis [K (100 km) 1 (3 h) 1, shaded], potential temperature (K, black solid contours), and omega (μb s 1, dashed contours; upward is indicated in red, downward is indicated in blue) at 0600 UTC 3 January The blue dashed line in (b) indicates the approximate location of the cross section in (c). Fig hPa geopotential height (dam, black solid contours), wind (kt, barbs), PW (mm, red dashed contours), and standardized anomalies of PW (σ, shaded) at 0000 UTC 3 January Fig Base reflectivity (dbz) and surface observations at 1000 UTC 3 January Fig As in Fig. 5.5 except at (a) 1200 UTC 3 January 2010 and (b) 1800 UTC 3 January Fig As in Fig. 5.6 except at 1200 UTC 3 January xv

16 Fig As in Fig. 5.7 except at 1200 UTC 3 January Fig As in Fig except at 1300 UTC 3 January Fig As in Fig. 5.8 except for (a) 2 January 2010 and (b) 3 January Fig Mean 500-hPa geopotential height (dam, black contours) for 0600 UTC 13 March 1800 UTC 16 March 2010 and the track of the 500-hPa cutoff cyclone center every 6 h (red contours). The brown bold lines depict the Northeast cutoff cyclone domain. Fig Four-day NPVU QPE (mm, shaded) ending 1200 UTC 16 March The black bold line depicts the Northeast precipitation domain. Fig h NPVU QPE (mm, shaded) ending (a) 1200 UTC 13 March 2010, (b) 1200 UTC 14 March 2010, (c) 1200 UTC 15 March 2010, and (d) 1200 UTC 16 March The black bold line depicts the Northeast precipitation domain. Fig As in Fig. 5.4 except at (a) 1800 UTC 13 March 2010 and (b) 0000 UTC 14 March Fig At in Fig except at 0000 UTC 14 March Fig As in Fig except at 0000 UTC 14 March Fig hPa geopotential height (dam, black contours), wind speed (m s 1, shaded), and wind (kt, barbs) at 0000 UTC 14 March Fig As in Fig except at 0000 UTC 14 March Fig As in Fig except at 1800 UTC 13 March Fig As in Fig. 5.5 except at (a) 1200 UTC 14 March 2010 and (b) 1800 UTC 14 March Fig As in Fig. 5.6 except at 1200 UTC 14 March Fig hPa equivalent potential temperature (K, black contours), equivalent potential temperature advection [K (3 h) 1, shaded], and wind (m s 1, barbs) at 1200 UTC 14 March Fig As in Fig. 5.17a except at 1200 UTC 14 March Fig As in Fig. 5.8 except for (a) 13 March 2010 and (b) 14 March xvi

17 1. Introduction 1.1 Overview Forecasting precipitation distributions associated with cool-season 500-hPa cutoff cyclones can be a challenge in the Northeast United States (US). Cutoff cyclones are cold-core cyclonic vortices that are displaced from the mean westerly steering flow and as a result are often associated with slow eastward movement (e.g., Palmén 1949; Palmén and Nagler 1949; Bell and Bosart 1989). Although the structure and evolution of cutoff cyclones have been extensively documented, current numerical weather prediction (NWP) models still have difficulty forecasting the track and strength of cutoff cyclones (e.g., Hawes and Colucci 1986; Ceppa and Colucci 1989). Furthermore, while model forecasts of wind and geopotential height fields have continually improved over the years, improvements in quantitative precipitation forecasts (QPFs) have lagged (e.g., Sanders 1979; Bosart 1980; Charba and Klein 1980; Gyakum and Samuels 1987; Jensenius 1990; Olson et al. 1995; Fritsch et al. 1998), warranting further study of the dynamic and thermodynamic processes that contribute to heavy precipitation associated with various weather systems, including cutoff cyclones. A recent study found that approximately 30% of the annual precipitation in the Northeast US can be attributed to 500-hPa cutoff cyclones (Aiyyer and Atallah 2002). The physiography of the Northeast US frequently complicates observed precipitation patterns due to terrain-induced precipitation enhancement and suppression, in addition to mesoscale circulations arising from differential roughness and heating across land water 1

18 boundaries (e.g., Opitz et al. 1995; Smith et al. 2002; St. Jean et al. 2004). Due to their slow-moving nature and interaction with the complex topography of the Northeast US, cutoff cyclones are often associated with varying precipitation distributions, thus posing additional forecasting challenges. Therefore, there is a need to further investigate synoptic-scale and mesoscale processes associated with cutoff cyclones that lead to varying precipitation distributions in the Northeast US. It is well known that tropical phenomena, including El Niño Southern Oscillation (ENSO) and the Madden Julian oscillation (MJO), often influence weather in the midlatitudes through atmospheric teleconnections. Past studies have shown that the linkage between the tropics and extratropics occurs as a result of the development of atmospheric Rossby wave trains (e.g., Sardeshmukh and Hoskins 1988; Kim et al. 2006). In general, both ENSO and the MJO are stronger and more active during the Northern Hemisphere winter; hence it follows that these phenomena may affect the occurrence of cool-season cutoff cyclones in the Northeast US. No previously published studies have examined the combined relationship between ENSO and the MJO and the frequency of 500-hPa cutoff cyclones in the Northeast US; however, knowing if such a relationship exists would be beneficial for forecasters by providing an indication of the likelihood of cutoff cyclone occurrence with a lead time of one to two weeks. The goal of the current study is to identify key synoptic-scale and mesoscale features that differentiate between various precipitation distributions associated with 500- hpa cutoff cyclones in the Northeast US. This goal will be achieved by compositing cutoff cyclones of similar precipitation amount, tilt, and structure and by conducting case study analyses of three cutoff cyclone events associated with varying precipitation 2

19 distributions that proved to be a challenge to forecast. In addition, the influence of ENSO and the MJO on cutoff cyclone frequency will be examined to determine the likelihood of cutoff cyclone occurrence during the active phases of these tropical modes. Ideally, this study aims to provide tools to increase forecaster situational awareness in an attempt improve future precipitation forecasts associated with 500-hPa cutoff cyclones in the Northeast US. 1.2 Literature Review Overview of Cutoff Cyclones Observations of midtropospheric cutoff cyclones have been documented extensively since the late 1940s (e.g., Hsieh 1949; Palmén 1949; Palmén and Nagler 1949). These early studies described observations of cold vortices within the background westerly flow cutting off from the source of polar air to the north (Fig. 1.1). Since cutoff cyclones are separated from the main flow, they are often observed to have slower forward speeds than the background westerly flow (e.g., Bell and Bosart 1989). Petterssen (1956, section 12.8) described midlevel cutoff cyclones as being characterized by a symmetric distribution of temperature which reaches a minimum in the core. On isobaric maps, cutoff cyclones can be identified as closed contours of geopotential height associated with a cyclonic circulation (Fig. 1.2). In addition, it is widely recognized that cutoff cyclones may also be represented as a maximum (minimum) in potential vorticity (potential temperature) on an isentropic (potential vorticity, PV) surface, and the 3

20 associated cyclonic circulation can be attributed to the presence of a positive PV anomaly at midlevels in the vicinity of the cutoff cyclone (e.g., Kleinschmidt 1957, section 47; Hoskins et al. 1985; Bell and Keyser 1993; Holton 2004, section 6.3.3). The process of cutoff cyclone intensification can be explained using the principle of PV conservation: P θ ( f ) g Constant (1.1) p where P is the isentropic coordinate form of Ertel s PV, ζ θ is relative vorticity on isentropic surfaces, f is the Coriolis parameter, g is gravity, and θ/ p may be approximated by δθ/δp, which represents the finite distance between isentropic surfaces measured in pressure units (Holton 2004, section 4.3). As an example, consider a hypothetical situation describing a mechanism for the intensification of a preexisting cutoff cyclone. In this hypothetical situation, an isolated air mass contained within a cutoff cyclone is displaced equatorward as it undergoes vertical stretching. Assuming that δθ and g are constant, f will decreases as the air mass moves equatorward and δp will increase as the column depth increases due to stretching; therefore, to maintain PV conservation ζ θ must also increase, which acts to strengthen the cyclonic circulation associated with the cutoff cyclone. Several past studies have found that common precursors for cutoff cyclone development include a large-amplitude ridge upstream and a broad trough in place where the cutoff cyclone eventually develops (e.g., Keyser and Shapiro 1986; Bell and Bosart 1993; Bell and Keyser 1993; Bell and Bosart 1994). Preceding cutoff cyclone development, a short-wave trough embedded within the large-scale trough amplifies as a northwesterly upper-level jet streak begins to move toward the base of the trough. The cutoff cyclone develops when the short-wave trough breaks off from the main flow as the 4

21 jet streak enters the base of the large-scale trough. Finally, the cutoff cyclone eventually becomes reabsorbed into the main flow as the jet moves into the southwesterly flow downstream of the large-scale trough axis. Thorncroft et al. (1993) described two nonlinear baroclinic scenarios that could lead to cutoff cyclone development. In the LC1 scenario, a positively tilted, thinning trough becomes separated from the main flow as anticyclonic wave breaking occurs and a cutoff cyclone develops equatorward of the mean jet (Fig. 1.3a). The LC2 scenario involves a negatively tilted trough that becomes wrapped up in its own cyclonic circulation, leading to a cutoff cyclone north of the jet axis (Fig. 1.3b). Cutoff cyclones are also occasionally observed to support the maintenance of atmospheric blocking (e.g., Rex 1950). Colucci (1985, 1987) documented several cases of cutoff cyclones that lead to the development of Rex (1950) blocking patterns at 500 hpa. Shutts (1983) theorized that cutoff cyclones help blocking systems persist by acting as a source of energy and PV. Northern Hemispheric and regional climatologies of cutoff cyclone frequency have been thoroughly documented. Parker et al. (1989) examined 500-hPa cyclones throughout the western half of the Northern Hemisphere and concluded that they are relatively infrequent events that occur less than 10% of the time. Bell and Bosart (1989) were among the first to investigate 500-hPa cutoff cyclone frequency and locations of genesis/lysis for the entire Northern Hemisphere. Their study identified several regions of maxima in cutoff cyclone frequency, including a region extending eastward from eastern Canada and the Northeast US across the North Atlantic. Bell and Bosart (1989) also determined that cutoff cyclone genesis and lysis regions are typically located equatorward of the main belt of westerlies, indicating that these systems are generally 5

22 slow-moving features. Nieto et al. (2002) examined 41 years of upper-level cutoff cyclones throughout the Northern Hemisphere and found that they are generally shortduration events, lasting two to three days on average, and usually have a northward or westward movement. More recently, Smith et al. (2002, 2003) examined cool-season (October May) 500-hPa cutoff cyclones in the Northern Hemisphere and identified several regions of favored cutoff cyclone activity across North America, including the southwest US, Hudson Bay, and the region encompassing the Northeast US and Canadian Maritimes. In addition, these studies found that within the Northeast US, there is an observed increase in cutoff cyclone frequency coinciding with a southward shift during the fall months Cutoff Cyclones and Precipitation Palmén (1949) recognized that midtropospheric cutoff cyclones are important weather producers in the midlatitudes. Hsieh (1949) was one of the first to document the precipitation distribution associated with a cutoff cyclone, noting that precipitation was distributed asymmetrically about the system. Light precipitation was found to occur near the core of the cyclone while heavy precipitation was observed southeast of the cyclone center within the surface warm sector (Fig. 1.4). Jorgensen et al. (1967) examined precipitation amount and location for winter season 700-hPa cutoff cyclones in the western US. Their results showed that as cyclone strength increases, the areal extent of the precipitation increases as well and precipitation tends to be focused in the eastern quadrants of the cyclone. In comparison, for weak systems the precipitation is generally 6

23 located near the center of the cyclone in the southwest quadrant. Klein et al. (1968) expanded upon this work by examining precipitation associated with cutoff cyclones at additional levels (i.e., 850, 500, and 300 hpa). The most intense cyclones at 500 hpa were found to have precipitation most commonly occurring in the southeast quadrant approximately 5 from the center of the cyclone (Fig. 1.5), while for weak systems the precipitation was generally located near the center of the cyclone in the southwest quadrant (not shown). Opitz et al. (1995) confirmed these results, determining that heavy precipitation is most likely to occur within the warm sector of extratropical surface cyclones where moisture and thermodynamic instability are maximized. Furthermore, Aiyyer and Atallah (2002) concluded that the optimal location for cutoff cyclones producing heavy precipitation in the Northeast US is to the west of this region, implying that the heavy precipitation occurs east of the cyclone center. Fracasso (2004) examined climatologies of precipitation distributions associated with 500-hPa cutoff cyclones in the Northeast US during the cool season (October May). Average daily precipitation amounts associated with cool-season cutoff cyclones were found to reach a maximum in November, decrease during the winter, and increase again slightly during the spring months. In addition, Fracasso (2004) observed enhanced precipitation amounts collocated with higher terrain, indicative of the modification of low-level flow by the topography of the Northeast US. Aiyyer and Atallah (2002) also emphasized the importance of physiographic influences on precipitation distributions, noting that during the cool season the precipitation is strongly modified by upslope flow and lake-effect enhancement. 7

24 1.2.3 Forecast Issues Associated with Cutoff Cyclones Historically, forecasters have had difficulty determining the direction and speed of movement of cool-season cutoff cyclones (Vore and McCarter 1956). Ceppa and Colucci (1989) examined the predictability of 500-hPa cutoff cyclones and found that although these systems are generally persistent, NWP models at the time were just as likely to incorrectly forecast existing systems as they were newly developing systems. These models were also found to have a tendency to overforecast geopotential heights associated with 500-hPa cutoff cyclones, with an average forecast error of +4 dam (Hawes and Colucci 1986; Ceppa and Colucci 1989). Despite these forecast issues, model forecasts of 500-hPa geopotential height fields have steadily improved over the last several decades; however, improvements in QPFs have displayed slower progress (Fig. 1.6). Jensenius (1990) evaluated the performance of past NWP models during the cool season and found that they performed poorly in forecasting precipitation amount and areal extent, especially beyond the 6-h forecast. Gyakum and Samuels (1987) found that forecasters consistently overforecasted precipitation amounts during the cool season. Olson et al. (1995) showed that overforecasting precipitation continued to be the case into the following decade, with forecasters displaying a tendency to overforecast the areal extent of one inch of precipitation by about 25%. Verification of 24-h 1-in. QPF for NWP models in recent years shows a steady improvement for both the NAM and GFS with a threat score of around 0.19 in 2000 increasing to around 0.25 by 2006 (Fig. 1.7). In addition, since 1998 the GFS has consistently shown slight improvement over the NAM, while 8

25 Hydrometeorological Prediction Center (HPC) forecasters continue to maintain greater skill than both models. Fritsch et al. (1998) recognized that continual advances in QPFs remain crucial in forecasting high-impact weather events and require improved understanding of synoptic-scale and mesoscale processes that lead to heavy precipitation. Forecasts of precipitation associated with cutoff cyclones can be especially challenging in the Northeast US. As discussed in section 1.2.2, low-level flow associated with cutoff cyclones can be modified by the complex terrain of the Northeast US (Fig. 1.8), acting to locally enhance or suppress precipitation. Opitz et al. (1995) examined cases of heavy precipitation in the eastern US and found that low-level convergence attributed to sea and lake breezes can enhance precipitation along coastlines. Their study also found that precipitation distributions can be significantly modulated by orographic effects in the vicinity of the Appalachian Mountains. Northwesterly low-level flow west of cutoff cyclones has been observed to lead to enhancement of precipitation along the mountain ranges of the Northeast US as the result of upslope flow (Smith et al. 2002; Sisson et al. 2004; St. Jean et al. 2004). Conversely, terrain shadowing is frequently observed on the leeward side of mountain ranges where downslope flow suppresses precipitation. Novak et al. (2004) found that banded precipitation can pose additional challenges in forecasting precipitation during the cool season. The study determined that frontogenesis in the presence of moist symmetric instability can force mesoscale precipitation bands, which can locally enhance precipitation. These precipitation bands were found to most commonly occur in the northwest quadrant of developing cyclones. Recently, several studies have proposed the use of forecast standardized anomaly fields to aid in identifying heavy precipitation events associated with East Coast winter 9

26 cyclones. Grumm and Hart (2001) argued that extreme weather events are often associated with significant flow departures from climatology; therefore, examination of the associated flow anomalies could provide forecasters with increased recognition of potentially high-impact systems. Standardized anomalies for fields such as 500- and 700- hpa geopotential height, 250- and 850-hPa wind components, and sea level pressure can be calculated using: N = (X μ) / σ (1.2) As defined by Grumm and Hart (2001), N is the standardized anomaly, X is a parameter value at a given grid point, μ is the 21-day running mean of the given parameter for that grid point, and σ is the grid point 21-day running standard deviation. Their study defined the term anomalous as a departure of more than ±2.5 standard deviations (σ) from climatology (i.e., the 30-year mean), indicating a situation that occurs less than 16% of the time. Using this methodology, Grumm et al. (2002) determined that heavy precipitation events in the Northeast US are associated with the strongest low-level easterly wind anomalies when compared to other regions in the US. Stuart and Grumm (2004) supported this result, finding that extreme precipitation events are more likely to occur when 850-hPa zonal wind anomalies are 4σ or below. Anomalous easterly lowlevel winds enhance precipitation by advecting Atlantic moisture into the Northeast US, by providing forcing for ascent through frictional convergence along coastlines, and by strengthening low-level frontogenesis (Stuart and Grumm 2006). Junker et al. (2008) concluded that model forecasts of standardized anomalies may be useful in indentifying extreme rainfall events in northern California. They found that the heaviest precipitation events are generally associated with large, slow-moving standardized anomalies of both 10

27 geopotential height and precipitable water. Junker et al. (2009) argued that the use of standardized anomaly forecasts will increase confidence in issuing forecasts of heavy precipitation amounts and extreme precipitation events The Influence of ENSO and the MJO Several past studies have examined seasonal teleconnections associated with tropical modes, including ENSO and the MJO, and extratropical atmospheric circulation patterns. The linkage between the tropics and extratropics often occurs as a result of the development of Rossby wave trains (e.g., Sardeshmukh and Hoskins 1988; Bladé and Hartmann 1995; Kim et al. 2006). Large-scale divergence aloft originating from deep moist convection in the tropics initiates the development of Rossby wave trains that extend poleward and eastward into the extratropics (Fig. 1.9). This extratropical response due to tropical heating is found to be strongest in the Northern Hemisphere during boreal winter (Jin and Hoskins 1995). Such Rossby wave trains have been observed throughout the globe in regions including Asia, the North Pacific, North America, and the Atlantic Ocean (e.g., Matthews et al. 2004). Noel and Changnon (1998) examined the teleconnection between ENSO and winter surface cyclone frequency in the US and concluded that during the warm phase of ENSO there is a significant increase in overall cyclone activity in New England. Hirsh et al. (2001) found that during the warm phase of ENSO, East Coast winter storms are 44% more frequent than during ENSO neutral conditions. Since 500-hPa cutoff cyclones are often linked to surface cyclones, the results of Noel and Changnon (1998) and Hirsh et al. 11

28 (2001) suggest that 500-hPa cutoff cyclones may be more frequent during the warm phase of ENSO as well. In comparison, both studies found that a relationship between surface cyclone occurrence and the cool phase of ENSO is not as evident. However, changes in the frequency of cyclones cannot be attributed entirely to ENSO conditions since other extratropical teleconnections (e.g., the North Atlantic Oscillation, the Pacific North American pattern) can affect the ENSO teleconnection with midlatitude weather (Noel and Changnon 1998). The MJO is characterized by an eastward-propagating region of deep moist convection and a time scale of days (Madden and Julian 1972, 1994). Matthews et al. (2004) found that when enhanced convection associated with the MJO is over the Indian Ocean a large-scale trough (ridge) is in place over western (eastern) North America, and when enhanced convection associated with the MJO is over the western Pacific Ocean there is a large-scale ridge (trough) in place over western (eastern) North America. This relationship suggests that there may be an increased frequency of cutoff cyclones in the Northeast US when enhanced convection associated with the MJO is located over the western Pacific Ocean and a trough is in place over the eastern US. Jones et al. (2004) found that during active phases of the MJO, predictability increased by two to three days for several synoptic fields, including 500-hPa geopotential height, suggesting that awareness of the location and strength of the MJO may improve forecasts of 500-hPa cutoff cyclones. Relatively few studies have examined the combined midlatitude influence of ENSO and the MJO. Changes in sea surface temperature associated with the various phases of ENSO modify the environment in which deep moist convection develops, thus 12

29 affecting the MJO. Pohl and Matthews (2007) found that ENSO modulates the lifetime of the MJO, with shorter MJO lifetime observed during the warm phase of ENSO due to warm sea surface temperatures extending farther east which results in faster eastward propagation of the MJO. Roundy et al. (2010) investigated changes in global circulation patterns due to the MJO during various phases of ENSO and determined that when the two modes are simultaneously active they need to be considered together to determine their effect on midlatitude weather patterns. Therefore, investigation of the combined influence of ENSO and the MJO on 500-hPa cutoff cyclone frequency in the Northeast US may provide forecasters with increased lead time as to the likelihood of cutoff cyclone occurrence. 1.3 Study Goals The primary goals of this study are to: (1) determine if ENSO and the MJO can provide increased lead time as to the likelihood of 500-hPa cutoff cyclone occurrence in the Northeast US; (2) examine five cool seasons of 500-hPa cutoff cyclones in the Northeast US and identify key synoptic-scale patterns differentiating between various precipitation distributions; and (3) determine how synoptic-scale and mesoscale features associated with 500-hPa cutoff cyclones affect precipitation distributions in the Northeast US through case study analyses of difficult-to-forecast cutoff cyclone events as well as cutoff cyclone events associated with varying precipitation distributions. The ultimate objective of this research is to incorporate the anticipated findings into operational forecasting at the National Weather Service Weather Forecast Offices throughout the 13

30 Northeast US in order to increase forecaster situational awareness and improve future precipitation forecasts associated with 500-hPa cutoff cyclones. The data and methods used in this study are described in chapter 2. A discussion of the influences of ENSO and the MJO on 500-hPa cutoff cyclone frequency in the Northeast US is presented in chapter 3. Chapter 4 focuses on composites of cool-season 500-hPa cutoff cyclones categorized by precipitation amount, tilt, structure and discusses common features differentiating between precipitation distributions. Case study analyses of three cutoff cyclone events associated with varying precipitation distributions are described in detail in chapter 5. Finally, chapters 6 and 7 include a research summary and key conclusions. 14

31 Fig Schematic representation of the development of a cold vortex as shown by the height contours (solid lines) and isotherms (dashed lines) on an isobaric surface in the middle troposphere. (Figure and caption from Hsieh 1949, Figs. 11a c.) Fig Sample 500-hPa geopotential height analyses illustrating the objective method used to identify closed circulation centers. (a) Three sample radial arms used to identify a 30 m closed contour around the cyclone center point A. Geopotential heights rise to at least 30 m larger than that of the point A before decreasing along each radial arm. Point A is therefore identified as a closed cyclonic circulation center. (b) As in (a) except that geopotential heights along the radial arm do not exceed 30 m higher than the point A before decreasing. Point A is therefore not identified as a closed circulation center. (Figure and caption from Bell and Bosart 1989, Fig. 1a,b.) 15

32 Fig Schematic of a PV-θ contour in an Atlantic storm track sharing its main characteristics with (a) an LC1-type life cycle and (b) an LC2-type life cycle. The dashed line marks the approximate position of the mean jet at each stage. (Figure and caption from Thorncroft et al. 1993, Fig. 12a,b.) Fig Schematic representation of the distribution of precipitation relative to the isobaric contours (solid lines) of the surface of the cold dome. Relatively light precipitation occurs in the stippled area, with heavier precipitation occurring in the hatched area. (Figure and caption from Hsieh 1949, Fig. 13.) 16

33 Fig Areas of maximum frequency of occurrence of measurable precipitation with the most intense lows (Class III) centered at the origin for 850-, 700-, 500-, and 300-hPa levels. The symmetrical circles represent idealized contours about the low center at any level. (Figure and caption from Klein et al. 1968, Fig. 8.) Fig Threat scores for the forecasters 0.50-, 1.00-, and 2.00-in. forecasts for day 1 from 1961 through (Figure and caption from Fritsch et al. 1998, Fig. 1.) 17

34 Fig Annual threat scores of 24-h 1-in. Day 1 QPF since 1993 for the NAM (green), GFS (blue), and HPC forecasters (red). (Figure from hpcverif.shtml.) Fig. 1.8: Key mountain ranges (blue) and valleys (yellow) of the Northeast US. 18

35 Fig Longitude latitude schematic of the day 15 σ 0.24 meridional wind perturbation for the heating on a December February zonal flow. The contour interval is 0.5 m s 1. The zero contour is not shown, and negative contours are dashed. (Figure and caption from Jin and Hoskins 1995, Fig. 8.) 19

36 2. Data and Methodology 2.1 Data Sources The Influence of ENSO and the MJO A list of all 500-hPa cutoff cyclones observed in the Northern Hemisphere from 2000 through 2007 was obtained from a dataset compiled objectively by Scalora (2009). This dataset included the date, time (0000, 0600, 1200, or 1800 UTC), location in latitude longitude coordinates, and minimum geopotential height value for each cutoff cyclone. ENSO phases for all weeks from 2000 through 2007 were determined using weekly Niño-3.4 sea surface temperature anomalies (SSTA), obtained from the Climate Prediction Center. The Niño-3.4 SSTA data are available online at The daily phase and amplitude of the MJO were determined from an index developed by Wheeler and Hendon (2004), hereafter WH04. The index is determined by considering 850-hPa zonal wind, 250-hPa zonal wind, and daily averaged total outgoing longwave radiation (OLR). WH04 identified eight phases of the MJO that are indicative of the location of enhanced convection associated with the MJO along the equator, with phase 1 representing the MJO located off the east coast of Africa and increasing values representing regions progressively farther east (Fig. 2.1). Archived data for the MJO phase and amplitude were obtained from the Bureau of Meteorology, available online at 20

37 matw/maproom/rmm/. From the daily MJO data, the weekly MJO phase and amplitude were determined for each week from 2000 through Cutoff Cyclone Composites The dates and times of cutoff cyclones for the 2004/ /09 cool seasons were obtained from the list of 500-hPa cutoff cyclones provided by Scalora (2009) and verified by manual inspection of 500-hPa geopotential height fields. The 500-hPa maps were created from four-times-daily National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) final analyses on a 1.0 latitude longitude grid (Environmental Modeling Center 2003). All of the gridded data used in this study were obtained from the in-house data archive at the Department of Atmospheric and Environmental Sciences at the University at Albany (DAES/UA), unless otherwise specified. Composite analyses of cutoff cyclones with similar precipitation amount, tilt, and structure were created using the NCEP National Center for Atmospheric Research (NCAR) reanalysis dataset. The NCEP NCAR reanalysis data are available on a grid with a 6-h temporal resolution (Kalnay et al. 1996; Kistler et al. 2001) Case Study Analyses 21

38 To investigate the synoptic-scale and mesoscale meteorological conditions for the cutoff cyclone events considered, four-times-daily NCEP GFS initialized analyses were utilized. These analyses were obtained online from NOAA s National Operational Model Archive and Distribution System ( Precipitation distributions for each case were compiled from the 6-h NCEP National Precipitation Verification Unit (NPVU) quantitative precipitation estimates (QPEs), available online at Since 2004 NPVU QPEs have been available with a horizontal resolution of 4 km and have incorporated data from rain gauges, radar-estimated precipitation amounts, and NWS Cooperative Observer Program reports (Soulliard 2007). The analyses of cutoff cyclone events also utilized hourly surface and buoy observations from NWS Automated Surface Observing System sites, obtained from the in-house data archive at the DAES/UA. Radar images were created from NEXRAD Level III base reflectivity (0.5 elevation angle) data, with approximately 8-km resolution. The radar data were obtained through the NOAAPORT datatstream and stored at the DAES/UA. 2.2 Methodology The Influence of ENSO and the MJO 22

39 From the dataset provided by Scalora (2009), only 500-hPa cutoff cyclones that were observed within the Northeast cutoff cyclone domain, defined as between N and E (Fig. 2.2), were considered. Cutoff cyclones were also required to have a minimum duration of 12 h within the Northeast cutoff cyclone domain. A total of 294 cutoff cyclones met these requirements over the eight-year period from 2000 through The Niño-3.4 SSTA and the MJO phase and amplitude were recorded for the date of the first appearance of the cutoff cyclone in the Northeast cutoff cyclone domain. To ensure an accurate estimate of the MJO phase, the dataset was filtered to remove cutoff cyclones that occurred when the MJO was weak (i.e., amplitude < 1). After removal of weak-mjo cutoff cyclones, 196 cutoff cyclones remained in the dataset. Using Niño-3.4 SSTA data, each cutoff cyclone was classified as occurring during cool, warm, or neutral ENSO conditions. Cool, warm, and neutral conditions were defined as Niño-3.4 SSTA less than or equal to 0.5 C, greater than or equal to +0.5 C, and between 0.5 and +0.5 C, respectively. In addition, the ENSO trend was determined for each cutoff cyclone. To determine the ENSO trend, the difference between the Niño-3.4 SSTA for one week before and one week after the date of the first appearance of the cutoff cyclone was calculated. Based on this calculation, cutoff cyclones were further categorized as occurring during warming, cooling, or steady ENSO conditions. To investigate the influence of the MJO and ENSO conditions on the synopticscale pattern across North America, daily composites of 500-hPa mean geopotential height, 500-hPa geopotential height anomalies, and interpolated OLR anomalies were created online through NOAA s Earth Science Research Laboratory Physical Sciences 23

40 Division webpage ( The online tool utilizes NCEP NCAR reanalyses to calculate the daily composites Standardized Anomalies As discussed in section 1.2.3, the use of standardized anomalies may be beneficial to forecasters in order to help them evaluate the degree of departure from normal for various fields and to assess the potential impact of cutoff cyclone events. In this study, standardized anomalies of several fields, including 250-hPa zonal wind, 500-hPa geopotential height, 850-hPa zonal and meridional wind, and precipitable water (PW), were calculated for the cutoff cyclone composites and for each cutoff cyclone event examined. The methodology used in this study to calculate standardized anomalies begins with computing the centered 21-day running means of the aforementioned fields over a 30-year period ( ) from the NCEP NCAR reanalysis data. Standardized anomaly fields were created from the 2.5 NCEP NCAR reanalyses (composites) and the 0.5 GFS analyses (case studies) with respect to the climatological field using Eq The General Meteorological Package (GEMPAK), version 11.1, was utilized to display the resulting standardized anomaly fields. Significance levels for various standardized anomaly values are listed in Table I and can be used to represent the rarity of a given situation. As an example, a field with a departure of ±2σ from normal represents a situation that occurs approximately 5% of the time at any given location, assuming a normal distribution. Grumm and Hart (2001) 24

41 determined that the confidence limits of a normal distribution are reasonably representative of the actual confidence limits, based on an examination of return periods for high-impact winter events Cutoff Cyclone Composites A manual inspection of 500-hPa geopotential height fields created in GEMPAK was used to verify the dates and times of cutoff cyclones in the Northeast cutoff cyclone domain during the 2004/ /09 cool seasons. For a cyclone to be considered a cutoff cyclone, it had to maintain a 30-m geopotential height rise in all directions at 500 hpa for at least three consecutive analysis periods (i.e., a 12-h period), thus ensuring that the cyclone had at least one closed geopotential height contour. For the purpose of this study, one day was defined as the 24-h period from 1200 to 1200 UTC. Each day that a cutoff cyclone was present within the Northeast cutoff cyclone domain was termed a cutoff cyclone day. If a cutoff cyclone remained within the Northeast cutoff cyclone domain for more than one day it was called a cutoff cyclone event. A precipitation domain was defined to include the states of New England in addition to New York, Pennsylvania, and New Jersey (Fig. 2.2). A total of 384 cutoff cyclone days were identified during the five cool seasons from application of the above methodology. On average, 77 cutoff cyclone days occurred per cool season and the most cutoff cyclone days (93) were observed during the 2005/06 cool season (Fig. 2.3). Monthly cutoff cyclone frequencies in the Northeast US are maximized during the fall and spring months, with the most number of cutoff cyclone 25

42 days occurring in April, followed by October (Fig. 2.4). The 384 cutoff cyclone days occurred in association with 170 cutoff cyclone events. The average duration of cutoff cyclone events was 35.6 h, with the longest-lasting cutoff cyclone remaining within the Northeast cutoff cyclone domain for 108 h (Fig. 2.5). In order to create composite analyses, each cutoff cyclone day was categorized according to precipitation amount, the tilt of the cutoff cyclone at 500 hpa, and the structure of the cutoff cyclone at 500 hpa. First, the areal extent (expressed as a percentage of the Northeast precipitation domain) of 25 mm of precipitation was determined objectively in Adobe Photoshop using 24-h NPVU QPEs. The 24-h NPVU QPEs were compiled by adding the 6-h NPVU QPEs from 1200 to 1200 UTC. These results were used to divide cutoff cyclone days into heavy precipitation (HP), light precipitation (LP), or no precipitation (NP) cutoff cyclone days. HP cutoff cyclone days were defined as cutoff cyclone days where at least five percent of the precipitation domain received 25 mm of precipitation or greater. Five percent of the precipitation domain corresponds to approximately km 2 and was used to eliminate small, localized precipitation events. LP cutoff cyclone days were days where precipitation was observed but did not meet the HP criteria. NP cutoff cyclone days were days in which a cutoff cyclone was present in the Northeast cutoff cyclone domain and precipitation was not observed at any location within the Northeast precipitation domain. Of the 384 cutoff cyclone days that occurred during the five cool seasons studied, there were 100, 250, and 34 HP, LP, and NP cutoff cyclone days, respectively. Cutoff cyclone days were further categorized by the tilt of the 500-hPa trough and embedded cutoff cyclone, using methodology similar to Scalora (2009). The tilt (i.e., 26

43 negative, neutral, or positive) was determined for each cutoff cyclone day by manual examination of the 500-hPa geopotential height field for the time preceding the 6-h maximum precipitation for that day. Negative tilt was defined as an angle, α, less than or equal to 20 between the trough axis and a line of longitude; neutral tilt was defined as α between 20 and 20 ; and positive tilt was defined as α greater than or equal to 20 (Fig. 2.6). Note that it was possible for a cutoff cyclone event spanning multiple days to have varying tilts over the event lifetime. Finally, cutoff cyclone days were again subdivided by structure, as manifested in the 500-hPa geopotential height field. The structure was designated as either cutoff or trough based on the following criteria: For a system to be considered a cutoff it had to have a 250-hPa zonal wind standardized anomaly of 2σ or below on the poleward side of the cyclone. Stuart and Grumm (2004, 2006) determined that a 250-hPa zonal wind anomaly threshold of 2.5σ or below can be used to identify slow-moving, longduration cyclones that are cut off from the main westerly flow. In the current study, a slightly lower threshold was subjectively determined to be representative of cyclones cut off from the background flow. For a system to be placed into the trough category, it did not meet the cutoff criterion and was essentially a closed low embedded within a large-scale trough. Since there were so few NP cutoff cyclone days, these days were not subdivided by structure. The resulting cutoff cyclone classification system included 15 composite categories as depicted in Fig Composite analyses were created for each category using 6-h NCEP NCAR reanalysis data. Due to data availability and time constraint issues, cutoff cyclones days in 2009 were not included in the analyses, resulting in a total 27

44 of 338 cutoff cyclone days that were composited during the 2004/ /09 cool seasons. The grid for each cutoff cyclone day was centered on the location of the 500- hpa cutoff cyclone at the time preceding the 6-h maximum precipitation. To create cyclone-relative composites, the grids for each cutoff cyclone day in a given category were averaged and centered on the centroid of all of the 500-hPa cutoff cyclones, determined objectively by the dataset provided by Scalora (2009). Composite analyses were created in GEMPAK for common tropospheric fields, including 250-hPa wind speed, 500-hPa geopotential height, 850-hPa potential temperature, mean sea level pressure (MSLP), and PW, to help facilitate further analysis Case Study Analyses Three cutoff cyclone events were chosen for in-depth examination based on their association with precipitation forecasting challenges and varying precipitation distributions throughout their lifetime in the Northeast cutoff cyclone domain. The events include: 2 3 February 2009, 1 4 January 2010, and March The 2 3 February 2009 cutoff cyclone event was associated with difficult-to-forecast precipitation. This event was considered a precipitation forecast bust; heavy precipitation (>25 mm) was forecast to occur with this event but less than 5 mm was verified at most locations. The 1 4 January 2010 and March 2010 events were associated with long-lived cutoff cyclones that produced varying daily precipitation distributions in the Northeast US. In addition, the topography of the Northeast US played a role in 28

45 modifying the precipitation distributions associated with both of these events, complicating precipitation forecasts. The three cutoff cyclone events examined are associated with several of the cutoff cyclone composite categories previously discussed. The synoptic-scale features for each day of the cutoff cyclone events will be compared to the associated composite analysis for that day to validate the use of conceptual composite summaries in operations. The following maps were produced in GEMPAK for each cutoff cyclone event: 1) Event-average 500-hPa geopotential height and track of the cutoff cyclone every 6 h to examine the location of the cutoff cyclone with respect to the precipitation domain. 2) Event-total and 24-h NPVU QPEs to illustrate the event and daily precipitation distributions, respectively, in order to locate areas of heaviest precipitation produced by the cutoff cyclone. 3) 250-hPa geopotential height, wind speed, and divergence to identify the location of the upper-level jet streak and to diagnose the associated jet dynamics with respect to the cutoff cyclone. 4) 500-hPa geopotential height, absolute vorticity, absolute vorticity advection, and wind speed and direction to determine the location and tilt of the cutoff cyclone and to illustrate its vorticity structure. 5) 700-hPa geopotential height, temperature, Q vectors, and Q-vector divergence to identify regions of favorable quasi-geostrophic (QG) forcing for ascent. Q vectors were calculated using the built-in Q-vector parameter in GEMPAK, calculated using: 29

46 (2.1) as defined in Holton (2004, section 6.4.2). Regions of forcing for upward and downward vertical motion were determined from term A of the right-hand side of the Q-vector form of the QG omega equation: (2.2) where term A represents Q-vector divergence and term B represents the β term, which is generally neglected because it is small relative to term A (Holton 2004, section 6.4.2). Here, Q-vector convergence indicates regions of forcing for ascent, not the actual vertical motion, since the left-hand side of Eq. 2.2 is not explicitly solved. 6) 850-hPa geopotential height and wind speed to determine the location and strength of low-level jets. 7) 850-hPa equivalent potential temperature, equivalent potential temperature advection, and wind speed and direction to evaluate the availability of low-level warm, moist air. 8) 925-hPa two-dimensional frontogenesis, potential temperature (θ), and wind speed and direction to locate low-level surface boundaries and low-level forcing for ascent. Frontogenesis was computed from the built-in scalar frontogenesis function in GEMPAK, defined in Martin (2006, section 7.2) as: (2.3) 30

47 Frontogenesis was calculated using the total horizontal wind and potential temperature. 9) MSLP, hPa thickness, and PW to investigate the development of a surface cyclone, areas of thermal advection, and moisture availability. 10) Radar snapshots and surface observations to identify the location of precipitation at a given time and to assess the associated surface meteorological conditions contributing to enhancement or suppression of precipitation. 11) Vertical cross sections of two-dimensional frontogenesis, potential temperature, and vertical velocity to show the vertical structure of the cutoff cyclone. 12) Maps of standardized anomalies of 250-hPa zonal wind, 500-hPa geopotential height, 850-hPa zonal and meridional wind, and PW to evaluate the degree of departure from normal for these fields. 31

48 Table I. Significance levels based on the standard deviations from normal for a normal distribution. (Table and caption from Grumm and Hart 2001, Table 1). 32

49 Fig MJO phase diagram depicting the approximate locations of the enhanced convective signal of the MJO for each phase. Weak MJO activity is represented by the inner circle. (Figure from Wheeler and Hendon 2004, Fig. 7.) Fig Northeast cutoff cyclone domain (red outline) and precipitation domain (green outline). 33

50 Fig Annual frequency of cool-season 500-hPa cutoff cyclones in the Northeast US. Fig Monthly frequency of 500-hPa cutoff cyclones in the Northeast US by cool season. 34

51 Fig Histogram of the duration of cool-season 500-hPa cutoff cyclone events occurring within the Northeast US during 2004/ /09. Fig Schematic used to assign a tilt classification to each cutoff cyclone day. (Figure from Scalora 2009, Fig. 2.3.) 35

52 Fig Number of cutoff cyclone days included in each composite category. Colors indicate the daily precipitation amount: heavy precipitation (blue), light precipitation (green), or no precipitation (red). 36

53 3. Results: The Influence of ENSO and the MJO on Cutoff Cyclone Frequency 3.1 The Influence of ENSO Regional Influence of ENSO The number of weeks characterized by the various ENSO phases from 2000 through 2007 is listed in Table II. The neutral phase of ENSO was most common during the eight years, with nearly 49% of weeks characterized by weekly Niño-3.4 SSTA between +0.5 and 0.5 C. The warm phase of ENSO occurred more often than the cool phase (126 compared to 87 weeks), suggesting that El Niño conditions, which are characterized by the warm phase of ENSO, were more prevalent during the period studied. The trend of ENSO was steady only 16% of weeks during the eight-year period, suggesting that unchanging Niño-3.4 SSTA over a two-week period was relatively infrequent. Composites of 500-hPa geopotential height anomalies for the various ENSO phases show that for all phases, except for ENSO warm and steady, there were positive geopotential height anomalies over the eastern US (Figs. 3.1a i). During ENSO warm and steady conditions, negative geopotential height anomalies over the eastern US were favored and 500-hPa geopotential heights were on average 10 to 30 m below normal; however, positive geopotential height anomalies were evident over New England (Fig. 3.1h). The strong signal for ENSO warm and steady conditions may be influenced by the relatively small sample size for this phase (n=18) as compared to the sample sizes for the 37

54 other phases, but this composite is the only one to hint at the presence of a trough over the eastern US. Therefore, it can be inferred that development of cutoff cyclones, favored by the presence of a preexisting trough, may be more likely when the weekly Niño-3.4 SSTA is greater than or equal to +0.5 C and steady. During ENSO cool and warming and ENSO neutral and warming, there appears to be a weak negative 500-hPa geopotential height anomaly over the western Atlantic Ocean (Figs. 3.1c,f), suggesting that during these ENSO phases cutoff cyclones may be favored off the East Coast, which may affect midlevel and low-level flow, and thus precipitation, across the Northeast US The Influence of ENSO on Cutoff Cyclone Frequency The frequency of 500-hPa cutoff cyclones in the Northeast US from 2000 through 2007 by ENSO condition shows a pronounced separation in the frequency of cutoff cyclones by ENSO phase (Fig. 3.2). Maxima in the number of cutoff cyclones occurred when ENSO was neutral and cooling, neutral and warming, warm and cooling, and warm and warming, with greater than 25 cutoff cyclones each, while minima occurred for all of the ENSO cool phases in addition to ENSO neutral and steady and ENSO warm and steady, with fewer than 15 cutoff cyclones each. As discussed in section 3.1.1, there was a large difference in the number of weeks characterized by the various ENSO phases, which likely affected the cutoff cyclone frequency for each ENSO phase. To remove any bias caused by the frequency of each ENSO phase, the cutoff cyclone frequency was divided by the number of weeks characterized by each ENSO phase. The resulting normalized distribution suggests that 38

55 cutoff cyclones in the Northeast US are more likely to occur during the warm phase of ENSO for any Niño-3.4 SSTA trend (Fig. 3.3). The large number of cutoff cyclones during the ENSO warm and steady phase agrees with the presence of a large negative 500-hPa geopotential height anomaly over much of the eastern US as shown in the composite for this phase (Fig. 3.1h); however, the maxima in cutoff cyclone frequency during ENSO warm and cooling and ENSO warm and warming are not supported by the composites for these phases, which indicate that positive 500-hPa geopotential height anomalies are present over the Northeast US (Figs. 3.1g,i). Statistical tests employing the bootstrap method were applied to determine statistical significance of the cutoff cyclone frequency distribution by ENSO phase (Wilks 2006, section 5.3.4). From this method, error bars were determined for cutoff cyclone frequency values and enclose the 95% confidence interval. As indicated by the overlapping error bars in Fig. 3.3, the distribution of cutoff cyclone frequency by ENSO phase is not statistically significant at the 95% confidence level. 3.2 The Influence of the MJO Regional Influence of the MJO The number of weeks characterized by each phase of the MJO from 2000 through 2007 is listed in Table III. During the eight years considered, the MJO was considered to be weak (i.e., amplitude less than 1) during 38% of weeks. On average, the MJO spent 32 weeks in each phase for , which is equivalent to approximately 4 weeks per 39

56 year in each phase. The MJO was least often in phase 4, corresponding to enhanced convection located west of the Maritime Continent. The average location of enhanced convection associated with each phase of the MJO from 2000 through 2007, as depicted in the composite interpolated OLR anomaly field, agrees fairly well with the locations determined by WH04 (Fig. 3.4). As an example, during phases 1 6 of the MJO, strong, negative composite interpolated OLR anomalies (< 25 W m 2 ) were collocated with the location of enhanced convection determined by WH04 and represented by a red X in Figures 3.4a f. For phase 7 of the MJO, the composite interpolated OLR anomaly signal in the western Pacific Ocean was weaker ( 10 to 15 W m 2 ), but was still collocated with the WH04 location of enhanced convection (Fig. 3.4g). In contrast, the composite interpolated OLR anomaly for phase 8 of the MJO was positive along the equatorial eastern Pacific Ocean, indicating lack of enhanced convection at the location determined by WH04 (Fig. 3.4h). One possible explanation for the disagreement may be attributed to the influence of ENSO on the MJO in the eastern Pacific Ocean, as discussed in section Composites of 500-hPa geopotential height anomalies for each phase of the MJO indicate the presence of negative geopotential height anomalies over the Northeast US during phases 1, 7, and 8 (Figs. 3.5a,g,h). The negative geopotential height anomalies were largest when the MJO was in phase 8 and 500-hPa geopotential height anomalies were between 20 and 40 m below normal over the eastern US (Fig. 3.5h). During this phase, the presence of a Rossby wave train extending across the Pacific Ocean eastward into North America was also apparent. As discussed in section 1.2.4, deep convection in the tropics, such as that associated with the MJO, often contributes to large-scale 40

57 divergence aloft and may initiate the development of Rossby wave trains that extend poleward and eastward into the extratropics (e.g., Jin and Hoskins 1995). Therefore, it may be inferred that the Rossby wave train observed within the 500-hPa geopotential height field composite for phase 8 of the MJO (Fig. 3.5h) was likely influenced by the presence of enhanced convection associated with the MJO. The geopotential height anomaly pattern during phases 1, 7 and 8 suggest that when enhanced convection associated with the MJO is moving across Africa or the Western Hemisphere, 500-hPa cutoff cyclones may be more likely to develop in the Northeast US due to the favored presence of midlevel troughs. Conversely, cutoff cyclones may be less likely to develop when the MJO is in phases 2 6, during which time ridging is favored over the eastern US, as indicated by the presence of positive geopotential height anomalies (Figs. 3.5b f) The Influence of the MJO on Cutoff Cyclone Frequency The frequency of 500-hPa cutoff cyclones in the Northeast US from 2000 through 2007 by MJO phase indicates that cutoff cyclones occurred most often during phase 8 of the MJO (Fig. 3.6), when enhanced convection was located in the Western Hemisphere. Conversely, cutoff cyclones occurred least often during phase 4 of the MJO, when enhanced convection was located over the Maritime Continent. When compared to the composites of 500-hPa geopotential height anomalies for each MJO phase, the maximum in cutoff cyclone frequency during phase 8 of the MJO agrees well with the presence of a large negative 500-hPa geopotential height anomaly over the eastern US during this phase (Fig. 3.5h). In addition, the minimum in cutoff cyclone frequency during phase 4 41

58 of the MJO is supported by the presence of a positive 500-hPa geopotential height anomaly over the eastern US in the composite for this phase (Fig. 3.5d), which is suggestive of unfavorable conditions for cutoff cyclone development. As with the results for the various ENSO phases, the bootstrap method was used to determine the statistical significance of the distribution of cutoff cyclone frequency by MJO phase. While the maximum in cutoff cyclone frequency for phase 8 of the MJO and the minimum in phase 4 differ by a total of 17 cutoff cyclones, the overlapping error bars in Fig. 3.6 indicate that the distribution of cutoff cyclones by MJO phase is not statistically significant at the 95% confidence level. 3.3 The Combined Influence of ENSO and the MJO Sections 3.1 and 3.2 have shown that examination of the separate influence of ENSO and the MJO on cutoff cyclone frequency in the Northeast US does not yield statistically significant results. As discussed in section 1.2.4, changes in sea surface temperature in the equatorial Pacific Ocean associated with ENSO may act to modify the development of deep convection, including the MJO. Therefore, it may be worthwhile to examine the combined influence of ENSO and the MJO on cutoff cyclone frequency to determine if the relationship is strengthened by considering these two tropical modes together. The MJO phase was determined for weeks during the cool, neutral, and warm phases of ENSO for After removal of weak amplitude MJO weeks, there were fewer than 10 weeks characterized by each MJO phase during the cool and warm phases 42

59 of ENSO. Due to the small sample sizes, the influence of the MJO by ENSO cool, neutral, and warm phases will not be discussed here; rather, the focus of this section will be on the influence of the MJO by ENSO trend on cutoff cyclone frequency in the Northeast US. The number of weeks characterized by each phase of the MJO for ENSO cooling, steady, and warming for is shown in Table IV. Once again, due to the relatively small sample sizes for MJO phases during the steady ENSO trend, the influence of the tropical modes during this ENSO trend will not be discussed Regional Influence of ENSO and the MJO Composites of 500-hPa geopotential height anomalies for each phase of the MJO during ENSO cooling indicate negative geopotential height anomalies over eastern Canada extending into the Northeast US and over the eastern US when the MJO is in phases 7 and 8, respectively (Figs. 3.7g,h); however, the sample size for phase 8 of the MJO is likely too small to be of significance, with only four weeks included within the composite. During phase 7, the presence of a Rossby wave train over the eastern Pacific Ocean extending into the US is especially evident when compared to the other phases and may be partially attributed to large-scale divergence aloft associated with the enhanced convection over the central Pacific that is associated with this phase of the MJO (e.g., Jin and Hoskins 1995). The midlevel pattern for phases 7 and 8 suggests that when the MJO is located over the central or eastern Pacific Ocean cutoff cyclones may be favored in the Northeast US, due to the presence of a preexisting trough. For the other phases of the MJO during ENSO cooling, the composites indicate positive geopotential height 43

60 anomalies over the eastern US, suggestive of the presence of a ridge over the region during these conditions, which would likely act to suppress cutoff cyclone development (Figs. 3.7a f). During weeks characterized by ENSO warming, composites of 500-hPa geopotential height anomalies for each phase of the MJO indicate strong negative geopotential height anomalies over the eastern US for phases 1 and 8 and over the extreme northeastern US during phase 6 (Figs. 3.8a,h,f). The anomalies are largest for phase 6 of the MJO, with a negatively tilted trough extending from Hudson Bay in Canada to the Gulf of Maine associated with 500-hPa geopotential height anomaly values on the order of 40 to 90 m below normal (Fig. 3.8f). The composites for phases 1, 8, and 6 suggest that cutoff cyclones in the Northeast US may be more common when ENSO is warming and enhanced convection associated with the MJO is moving across Africa, the Western Hemisphere, or the western Pacific Ocean, during which time a preexisting trough is favored over the region The Influence of ENSO and the MJO on Cutoff Cyclone Frequency In comparing the frequency of 500-hPa cutoff cyclones in the Northeast US by phase of the MJO for ENSO warming and cooling, there is a discernable difference. During ENSO cooling, the frequency peaks at 11 cutoff cyclones during phases 3 and 4 of the MJO (Fig. 3.9), when enhanced convection is located over the eastern Indian Ocean or the Maritime Continent (Figs. 3.7c,d). In addition, there is a distinct minimum in cutoff cyclones, with only one cutoff cyclone, during ENSO cooling and phase 7 of the 44

61 MJO, when enhanced convection is entering the Western Hemisphere (Fig. 3.4g). The distribution of cutoff cyclone frequency by MJO phase during ENSO warming appears to be reversed from that of ENSO cooling. During ENSO warming, there is a maximum in cutoff cyclone occurrence during phase 8 of the MJO, with 14 cutoff cyclones, and a minimum in phase 4, with only 2 cutoff cyclones. The cutoff cyclone frequency distribution by MJO phase for ENSO warming agrees reasonably well with the 500-hPa geopotential height anomaly composites. The peak in cutoff cyclones during ENSO warming and phase 8 of the MJO coincides with strong negative 500-hPa geopotential height anomalies over the eastern US (Fig. 3.8h). In addition, the minimum in cutoff cyclones during ENSO warming and phase 4 of the MJO agrees with the presence of strong positive 500-hPa geopotential height anomalies over the Northeast US (Fig. 3.8d); however, this agreement is likely related to the very small number of cutoff cyclone days (n=2) included in the composite for these conditions. In contrast, the cutoff cyclone frequency distribution by MJO phase for ENSO cooling does not agree well with the composites of 500-hPa geopotential height anomalies. While there is a minimum in cutoff cyclones for ENSO cooling and phase 7 of the MJO, the composite for this phase indicates a large negative geopotential height anomaly over eastern Canada extending into the Northeast US, suggesting favorable conditions for cutoff cyclone development (Fig. 3.7g). Similarly, the peak in cutoff cyclones during ENSO cooling and phases 3 and 4 of the MJO disagrees with the positive 500-hPa geopotential height anomalies over the Northeast US in the composite for these phases (Fig. 3.7c,d). 45

62 The bootstrap method was applied to the cutoff cyclone frequencies to determine statistical significance and the resulting error bars enclosing the 95% confidence level are shown in Figure 3.9. For ENSO cooling, the broad peak in cutoff cyclones in phases 2 6 is statistically different from the minimum in phase 7, as indicated by the lack of overlap in the error bars. Similarly, for ENSO warming, the large number of cutoff cyclones observed during phases 1 and 8 of the MJO are statistically different from the minimum observed during phase 4. When comparing the cutoff cyclone frequency distributions for ENSO cooling and warming, there is a lack of overlap in error bars during phases 4 and 7 of the MJO. The lack in overlapping error bars indicates that the maxima and minima in cutoff cyclone frequency by MJO phase for ENSO warming and cooling are statistically significant at the 95% confidence level. These results suggest that when considering the combined influence of ENSO and the MJO, 500-hPa cutoff cyclone frequency in the Northeast US is maximized when the MJO is over the Maritime Continent during ENSO cooling and when the MJO is over the Western Hemisphere during ENSO warming. 46

63 Table II. Total number and percentage of weeks characterized by each ENSO phase for Table III. Total number and percentage of weeks characterized by each phase of the MJO for Table IV. Total number of weeks characterized by each phase of the MJO for ENSO cooling, steady, and warming for A total of 159 weeks that occurred when the MJO was weak were not included. 47

64 Fig 3.1. Composites of hPa geopotential height anomaly (m, shaded) during weeks when ENSO was (a) cool and cooling, (b) cool and steady, (c) cool and warming, (d) neutral and cooling, (e) neutral and steady, (f) neutral and warming, (g) warm and cooling, (h) warm and steady, and (i) warm and warming. 48

65 Fig Number of 500-hPa cutoff cyclones by ENSO phase for Error bars were determined using the bootstrap method and enclose the 95% confidence interval. Fig As in Fig. 3.2 except normalized to account for the total number of weeks characterized by each ENSO phase for

66 Fig Composites of interpolated OLR anomaly (W m 2, shaded) during weeks when the MJO was in (a) phase 1, (b) phase 2, (c) phase 3, (d) phase 4, (e) phase 5, (f) phase 6, (g) phase 7, and (h) phase 8. The red X represents the estimated location of enhanced convection associated with each phase of the MJO as determined by Wheeler and Hendon (2004). A total of 159 weeks were removed during which the MJO was weak (i.e., amplitude < 1). 50

67 Fig Composites of hPa geopotential height anomaly (shaded, m) during weeks when the MJO was in (a) phase 1, (b) phase 2, (c) phase 3, (d) phase 4, (e) phase 5, (f) phase 6, (g) phase 7, and (h) phase 8. The red X represents the estimated location of enhanced convection associated with each phase of the MJO as determined by Wheeler and Hendon (2004). A total of 159 weeks were removed during which the MJO was weak (i.e., amplitude < 1). 51

68 Fig Number of 500-hPa cutoff cyclones by phase of the MJO for Cutoff cyclones that occurred when the MJO was weak (i.e., amplitude < 1) were removed. Error bars were determined using the bootstrap method and enclose the 95% confidence interval. 52

69 Fig As in Fig. 3.5 except for weeks when the ENSO trend was cooling. Fig As in Fig. 3.5 except for weeks when the ENSO trend was warming. 53

70 Fig Number of 500-hPa cutoff cyclones for each phase of the MJO for ENSO warming and cooling for Cutoff cyclones that occurred when the MJO was weak (i.e., amplitude < 1) were removed. Error bars were determined using the bootstrap method and enclose the 95% confidence interval. 54

71 4. Results: Cutoff Cyclone Composites 4.1 Average Location of Cutoff Cyclones The average locations of all 500-hPa cutoff cyclones within each composite category for the 2004/ /09 cool seasons are depicted in Fig A distinct difference in location of cutoff cyclones between precipitation amount categories is evident. HP cutoff cyclones tended to be located west of the Northeast precipitation domain, over the eastern Great Lakes. In comparison, LP cutoff cyclones tended to be located to the north of, or directly over, the Northeast precipitation domain, while NP cutoff cyclones were typically located to the east of the Northeast precipitation domain. The difference in average location of cutoff cyclones likely affected precipitation across the Northeast US by affecting several features, including forcing for ascent and moisture advection into the region, which will be discussed in the following sections. In comparing the average location of cutoff cyclones by structure, it is apparent that cyclones that fell into the cutoff category were on average located farther south than those categorized as trough (Fig. 4.1). The difference in location may partially be explained by the definition distinguishing between cutoff and trough. As discussed in section 2.2.3, the cutoff category was defined to include cyclones that were separated from the background westerly flow, indicating that these systems were displaced south of the mean westerly jet (e.g., Bell and Bosart 1989). In comparison, the trough category includes cyclones that were typically embedded within the background westerly flow and consequently were located farther north. 55

72 4.2 Cyclone-relative Composites HP Cutoff Cyclones 4.2.1a In-depth Examination of the HP Neutral Cutoff Category The synoptic-scale pattern for the HP neutral cutoff composite category included a 500-hPa trough and embedded cutoff cyclone located over the southwest corner of the Northeast precipitation domain and a tongue of warm air extending poleward east of the cyclone at 850 hpa (Fig. 4.2). To infer vertical motion associated with cutoff cyclones, one can apply the traditional form of the QG omega equation, (4.1) where term A represents the vertical differential advection of absolute vorticity by the geostrophic wind, term B represents the horizontal Laplacian of temperature advection, and contributions from diabatic heating and friction are ignored (e.g., Holton 2004, section 6.4.1). Forcing for upward vertical motion is therefore favored in regions of differential cyclonic vorticity advection increasing with height and in regions where the Laplacian of warm air advection is maximized. Applying Eq. 4.1 to the HP neutral cutoff cyclone-relative composite, it can be inferred that there was favorable forcing for ascent over New England where there was cyclonic vorticity advection downstream of the midlevel trough and warm air advection at 850 hpa (Fig. 4.2). At upper levels, a southwesterly jet streak greater than 90 kt (45 m s 1 ) was located south of the composite cutoff cyclone (Fig. 4.3). The cyclone-relative composite 56

73 also hints at a northern jet streak, located over the Gulf of St. Lawrence, suggestive of the presence of a dual jet streak. Dual jet streaks were frequently observed during the individual cutoff cyclone days that were included within the composite but may have been smoothed out by the averaging process. The dual jet streak configuration likely provided enhanced ascent over the precipitation domain, in associated with the poleward exit region of the southern jet streak and the equatorward entrance region of the northern jet streak (e.g., Uccellini and Kocin 1987). The composite surface cyclone was located approximately 500 km east-northeast of the composite 500-hPa cutoff cyclone (Figs. 4.2 and 4.3). Flow at 850 hpa east of the surface cyclone was generally southeasterly (Fig. 4.2), which resulted in advection of anomalously high PW values into the Northeast US. For cutoff cyclones within this composite category, PW values throughout New England were on average greater than 24 mm, which was equivalent to +1 to +3σ. The resulting widespread anomalous PW advection associated with these cutoff cyclones likely supported higher precipitation totals, allowing these cutoff cyclone days to meet the HP criteria b Schematic Diagrams of HP Cutoff Cyclones Schematic diagrams of key synoptic-scale features that affect precipitation were created for each HP composite category after examination of conventional tropospheric fields (Fig. 4.4a f), as discussed in the previous section. Features depicted in the schematics include the composite 500-hPa geopotential height field, 500-hPa absolute vorticity maxima, upper-level and low-level jet streaks, approximate locations of surface 57

74 cyclones and surface cold and warm fronts, and regions of PW greater than 20 mm. The quadrant, with respect to the 500-hPa cutoff cyclone, that received the heaviest precipitation on average is also highlighted in each schematic diagram. There was a distinct difference in the location of heaviest precipitation between HP cutoff cyclones that fell into the cutoff and trough categories. For the HP cutoff categories, the heaviest precipitation typically occurred northeast of the 500-hPa cutoff cyclone center (Figs. 4.4a,c,e). More specifically, 60% (21 of 35) of all cutoff cyclone days within the three HP cutoff categories were found to have the heaviest precipitation occurring within the northeast quadrant, while the heaviest precipitation occurred in the southeast, southwest, and northwest quadrants for 31% (11), 3% (1), and 6% (2) of cutoff cyclone days, respectively. In comparison, for the HP trough categories, the heaviest precipitation most commonly occurred within the southeast quadrant of the cutoff cyclone (Figs. 4.4b,d,f), as observed in approximately 54% (30 of 56) of cutoff cyclone days within these categories. The northeast quadrant was the second most common quadrant to receive the heaviest precipitation for the HP trough category, as observed in 36% (20 of 56) of these cutoff cyclone days. Although there was a difference in the quadrant of heaviest precipitation between structure categories, there was a general agreement that for all HP cutoff cyclones, the heaviest precipitation typically occurred east of the 500-hPa cutoff cyclone. The synoptic-scale features that contributed to precipitation for the HP cutoff categories, regardless of tilt, were similar (Figs. 4.4a,c,e), with several features contributing to the observed heavy precipitation including: (1) ascent favored within the poleward exit region of an upper-level jet streak to the south or southeast of the 500-hPa 58

75 cutoff cyclone; (2) QG forcing for ascent downstream of a midlevel cutoff cyclone in a region of inferred differential cyclonic absolute vorticity advection and warm air advection at low levels; (3) advection of PW into the region of heaviest precipitation by a southerly low-level jet; and (4) enhanced forcing for ascent in the vicinity of surface fronts. For the HP negative cutoff and HP neutral cutoff categories, the tongue of PW greater than 20 mm extended farther north and west to include areas poleward of the 500-hPa cutoff cyclone, where these PW values correspond to +1σ anomalies (Figs. 4.4a,c). The synoptic-scale features for the HP negative trough and HP neutral trough categories were comparable in that the primary upper-level and midlevel forcing mechanisms remained the same as those previously listed (Figs. 4.4b,d); however, for these cutoff cyclones, the heaviest precipitation typically occurred within the warm sector of the surface cyclone. Accordingly, warm air advection was not as important in contributing to QG forcing for ascent in comparison to differential cyclonic vorticity advection ahead of the midlevel trough. For cutoff cyclones within the HP positive trough composite category, the upper-level and midlevel features were similar to those of the HP negative trough and HP neutral trough categories; however, the surface cyclone associated with the HP positive trough category was typically located east of the Northeast precipitation domain (Fig. 4.4f), while for the HP negative trough and HP neutral trough categories the surface cyclone was typically located over western regions of the Northeast precipitation domain (Figs. 4.4b,d). The surface cyclone location for the HP positive trough category indicates that for these cutoff cyclones the surface fronts were likely located offshore, and thus could not have provided locally 59

76 enhanced forcing for ascent over the Northeast precipitation domain, as may have been the case for the other HP composite categories LP Cutoff Cyclones The synoptic-scale pattern at upper levels and midlevels for cutoff cyclone days within all six of the LP composite categories was not indicative of favorable QG forcing for ascent (Fig 4.5a f). As mentioned in section 4.1, for these composite categories the 500-hPa cutoff cyclone centers were on average located to the north of, or directly over, the precipitation domain, which likely contributed to suppression of upward vertical motion, and thus precipitation, over the Northeast US. As an example, an in-depth examination of the LP negative cutoff composite of 500-hPa geopotential height and absolute vorticity indicates that there was a lack of cyclonic vorticity advection over the Northeast US at midlevels (Fig. 4.6), suggesting that there was reduced QG forcing for ascent. In addition, the LP negative cutoff composite shows that the low-level flow throughout the precipitation domain was northwesterly or westerly, indicating slight cold air advection across the Northeast US. It is likely that the cold air advection further acted to suppress precipitation by contributing to QG forcing for descent over the region. Despite the inferred lack of synoptic-scale forcing for ascent for all LP composite categories, the heaviest precipitation for cutoff cyclones days within the LP composite categories was most commonly located to the southwest of the 500-hPa cutoff cyclone center (Figs. 4.5a f). The heaviest precipitation was observed in the southwest quadrant of the cutoff cyclone on 51% (112 of 218) of cutoff cyclone days within the LP 60

77 composite categories. In comparison, the heaviest precipitation occurred within the southeast quadrant of the cutoff cyclone 27% (58 of 218) of cutoff cyclone days and within each of the northern quadrants 11% (24 of 218) of cutoff cyclone days. Westerly or northwesterly low-level flow and cold air advection, as observed within the southwestern quadrant of the composite LP negative cutoff cyclone (Fig. 4.6), often results in lake-effect precipitation off of Lakes Erie, Ontario, and Champlain. Although very little moisture was in place across the Northeast US (PW values were less than 0.5σ for all LP composite categories, not shown), the lakes likely acted as a source of moisture for precipitation. In fact, 67% (147 of 218) of all cutoff cyclone days that were categorized as LP were associated with lake-effect precipitation off of at least one of the three aforementioned lakes. In comparison, only 25% (23 of 91) of cutoff cyclone days within the HP composite categories were associated with lake-effect precipitation. Precipitation associated with lake-effect bands may be heavy; however, the precipitation is generally localized. Therefore, while many of the LP cutoff cyclone days were associated with precipitation greater than 25 mm, they did not meet the areal extent criterion to be classified as HP NP Cutoff Cyclones The synoptic-scale patterns for cutoff cyclone days within the NP composite categories indicate that northwesterly and northerly flow at lower and upper levels were prevalent over the Northeast precipitation domain (Figs. 4.7a c). The flow over the Northeast US suggests that there was likely cold air advection into the region, which, by 61

78 application of Eq. 4.1, contributed to QG forcing for descent. Anticyclonic vorticity advection upstream of the 500-hPa trough and embedded cutoff cyclone also likely contributed to QG forcing for descent over the Northeast precipitation domain. As with the LP composite categories, the PW values for the NP composite categories were less than 0.5σ across the Northeast US. It is likely that the lack of moisture throughout the Northeast precipitation domain further contributed to the lack of precipitation observed in the region on cutoff cyclone days within the NP category. 62

79 Fig Average location of 500-hPa cutoff cyclones for each composite category. Color shading represents precipitation amount categories with blue, green, and red shading representing heavy precipitation (HP), light precipitation (LP), and no precipitation (NP) composite categories, respectively. Closed and open circles represent cutoff and trough composite categories, respectively. The brown bold lines depict the Northeast cutoff cyclone domain and the black bold line depicts the Northeast precipitation domain. 63

80 Fig Cyclone-relative composite for the HP neutral cutoff composite category showing 500-hPa geopotential height (dam, solid black contours) and absolute vorticity (10 5 s 1, shaded) and 850-hPa potential temperature (K, dashed red contours) and wind (>25 kt, barbs). The number of cutoff cyclone days (n) included within the composite is indicated in the top right corner. Fig Cyclone-relative composite for the HP neutral cutoff composite category showing MSLP (hpa, solid black contours), PW (mm, dashed red contours), standardized anomaly of PW (σ, shaded), and 250-hPa wind (kt, barbs). Green solid lines outline regions of 250-hPa wind greater than 70 and 90 kt. The number of cutoff cyclone days (n) included within the composite is indicated in the top right corner. 64

81 Fig Schematic diagrams depicting the cyclone-relative 500-hPa geopotential height (dam, solid black contours) and key synoptic-scale features that contribute to precipitation for the (a) HP negative cutoff, (b) HP negative trough, (c) HP neutral cutoff, (d) HP neutral trough, (e) HP positive cutoff, and (f) HP positive trough composite categories. The red X represents the location of the 500-hPa absolute vorticity maximum. The brown bold line depicts the Northeast precipitation domain. The number of cutoff cyclone days (n) included within each composite is indicated in the top right corner of each panel. 65

82 Fig As in Fig. 4.4 except for the (a) LP negative cutoff, (b) LP negative trough, (c) LP neutral cutoff, (d) LP neutral trough, (e) LP positive cutoff, and (f) LP positive trough composite categories. 66

83 Fig As in Fig. 4.2 except for the LP negative cutoff composite category. 67

84 Fig As in Fig. 4.4 except for the (a) NP negative, (b) NP neutral, and (c) NP positive composite categories. 68

85 5. Case Study Analyses of Three Cutoff Cyclone Events 5.1 The 2 3 February 2009 Cutoff Cyclone Event Event Overview The 2 3 February 2009 cutoff cyclone event was associated with difficult-toforecast precipitation and was considered a precipitation forecast bust for the Northeast US, since heavy precipitation greater than 25 mm was forecast to occur but most locations in the region received less than 5 mm. Forecast errors were mainly due to large disagreement between NWP models in the speed, track, and intensity of the surface cyclone (e.g., Grumm et al. 2009; Stuart 2009). As an example, the Global Ensemble Forecast System (GEFS) MSLP forecast valid 1200 UTC 3 February 2009 showed large spread among ensemble members 108 h prior to the validation time, with the largest variability (8 16 hpa) evident throughout eastern New York and western New England (Fig. 5.1a). Uncertainty in the location of the surface cyclone decreased as 1200 UTC 3 February 2009 approached; however, considerable variability among ensemble members was still apparent (Figs. 5.1c,e). In addition, the average of all ensemble members indicates that the forecast surface cyclone was initially located along the East Coast (Fig. 5.1b), but as the forecast projection decreased the location of the surface cyclone was forecasted farther east, over the Atlantic Ocean (Figs. 5.1d,f). On 2 February 2009, a large-scale trough at 500 hpa moved eastward across the northern US and stalled over the Great Lakes (not shown). Figure 5.2 shows the mean 69

86 500-hPa geopotential height field for the 24-h period (0000 UTC 3 February 0000 UTC 4 February 2009) during which the cutoff cyclone was within the Northeast cutoff cyclone domain. The track of the 500-hPa cutoff cyclone indicates that the cyclone developed and became cut off upstream of the large-scale trough at 0000 UTC 3 February 2009 (Fig. 5.2). The cutoff cyclone slowly moved southeastward around the base of the large-scale trough and then moved northeastward, remaining over the Great Lakes for the duration of its lifetime in the Northeast cutoff cyclone domain. At 0600 UTC 4 February 2009, the cyclone became reabsorbed into the large-scale flow and no longer met the criteria to be considered a cutoff cyclone. The two-day NPVU QPE for 2 3 February 2009 indicates that the precipitation associated with this cutoff cyclone event was confined to coastal regions (Fig. 5.3). Most regions received 2 10 mm, while Cape Cod and Maine received mm of precipitation. Most of the precipitation associated with this cutoff cyclone event fell in the 6-h periods following 1800 UTC 3 February 2009 and 0000 UTC 4 February 2009; therefore, in the following section the focus will be on examining the upper-level, midlevel, and low-level tropospheric conditions at these times to determine the synopticscale and mesoscale features that contributed to the observed precipitation Meteorological Conditions During the 2 3 February 2009 cutoff cyclone event, a dual jet streak was evident at upper levels, with wind speeds greater than 75 m s 1, or 146 kt (Figs. 5.4a,b). The position of the dual jet streak at 1800 UTC 3 February 2009 was consistent with 70

87 divergence over eastern Maine, in association with the poleward exit region of the southern jet streak and the equatorward entrance region of the northern jet streak (Fig. 5.4a), which provided conditions favorable for ascent over the region of heavy precipitation. The dual jet streak weakened and was located farther east by 0000 UTC 4 February 2009 and the associated region of divergence was no longer located over the Northeast US at this time (Fig. 5.4b). At midlevels, there was cyclonic absolute vorticity advection over Pennsylvania and New Jersey at 1800 UTC 3 February 2009, downstream of a lobe of moderate cyclonic absolute vorticity, on the order of s 1 (Fig. 5.5a). At 0000 UTC 4 February 2009, the lobe of cyclonic absolute vorticity moved northeastward around the cyclone center and there was cyclonic absolute vorticity advection over southern New York and Connecticut (Fig. 5.5b). By application of the traditional form of the QG omega equation (Eq. 4.1), differential cyclonic vorticity advection, inferred from the cyclonic vorticity advection at 500 hpa, contributed to favorable QG forcing for ascent throughout Pennsylvania, New Jersey, New York, and southern New England, and likely acted to support precipitation in those regions. Further examination of Fig. 5.3 reveals that the regions of 500-hPa cyclonic absolute vorticity advection were in fact collocated with the band of light precipitation (2 10 mm) extending from eastern Pennsylvania and New Jersey northeastward into western Massachusetts. However, there was little, if any, contribution to QG forcing for ascent by the Laplacian of temperature advection, as inferred from the 850-hPa temperature and wind fields at 1800 UTC 3 February 2010 (Fig. 5.6). For example, across Pennsylvania, New Jersey, New York, and southern New England, the northeasterly flow at 850 hpa was weak (<20 kt or 10 m s 1 ), resulting in 71

88 little or no temperature advection in those regions. The 700-hPa Q-vector analyses at 1800 UTC 3 February 2009 and 0000 UTC 4 February 2009 confirm there was QG forcing for ascent, indicated by Q-vector convergence, across regions of Pennsylvania, New Jersey, and New York (Figs. 5.7a,b). In addition, there was also Q-vector convergence over Cape Cod and Maine at 1800 UTC 3 February 2009 (Fig. 5.7a), where the heaviest precipitation was observed with this event. During the 2 3 February 2009 cutoff cyclone event, there was very little support for precipitation at low levels. The observed surface cyclone was located over the western North Atlantic Ocean, southeast of Cape Cod (Fig. 5.8), suggesting that surface fronts did not play a role in enhancing precipitation across the Northeast US during this event. PW values throughout the entire Northeast US were less than 12 mm, except in Cape Cod where PW values were mm (Fig. 5.8), which corresponded to +0.5 to +1.0σ (not shown). In addition, the northeasterly flow across the Northeast US resulted in little PW transport into the region. The lack of low-level forcing for ascent and low PW values likely contributed to the low precipitation amounts that were observed with this cutoff cyclone event Conceptual Summary A schematic diagram of the key synoptic-scale features that contributed to the precipitation associated with the 2 3 February 2009 cutoff cyclone event is presented in Fig The precipitation observed in eastern Maine was supported by ascent associated with a dual jet streak at 250 hpa. Light precipitation extending from eastern 72

89 Pennsylvania and New Jersey northeastward into western Massachusetts was located in a region of inferred differential cyclonic vorticity advection ahead of a lobe of cyclonic absolute vorticity southeast of the 500-hPa cutoff cyclone center, which likely contributed to QG forcing for ascent. There was little or no low-level support for precipitation, with little or no temperature advection at 850 hpa and low PW values observed throughout the Northeast US. The cutoff cyclone would have been placed into the LP positive trough composite category throughout the duration of its lifetime in the Northeast cutoff cyclone domain, since it involved a cutoff cyclone embedded within a large-scale trough and less than 25 mm of precipitation was observed in the Northeast precipitation domain. Comparing the event schematic diagram (Fig. 5.9) to the schematic for the LP positive trough category (Fig. 4.5f), it is evident that there was a difference in the location of the 500-hPa cutoff cyclone. At 1800 UTC 3 February 2009, the cutoff cyclone was located farther south and west than the composite cutoff cyclone. In addition, the surface cyclone at 1800 UTC 3 February 2009 was located a greater distance to the east of the midlevel cyclone than the LP positive trough composite surface cyclone The 1 4 January 2010 Cutoff Cyclone Event Event Overview The 1 4 January 2010 cutoff cyclone event was a long duration event with varying daily precipitation distributions throughout its lifetime. This cutoff cyclone event 73

90 was also associated with record-breaking snowfall observed in Burlington, VT, with total of 37.6 in. (95.5 cm) observed for 1 3 January 2010 (e.g., Sisson 2010). Numerical models showed considerable variability in forecasting the precipitation distribution leading up to the event (e.g., Stuart 2010a). The NAM apparently forecast the precipitation best, capturing the terrain enhancement just prior to the event, but QPF amounts were higher than observed precipitation amounts. The 1 4 January 2010 cutoff cyclone originated from a preexisting trough over central Canada and entered the Northeast cutoff cyclone domain at 0000 UTC 2 January 2010 (Fig. 5.10). Throughout the cutoff cyclone event, a highly amplified ridge associated with a large-scale blocking pattern and the negative phase of the North Atlantic Oscillation (NAO) was in place over Greenland (e.g., Sisson 2010). Studies have found that the negative phase of the NAO is typically associated with a meridional flow regime across the Northeast US, associated with blocking over the North Atlantic Ocean, which results in troughing and colder than normal temperatures along the East Coast (e.g., Bradbury et al. 2002, Stuart and Grumm 2006). During the 1 4 January 2010 cutoff cyclone event, the highly amplified ridge likely contributed to the cutoff cyclone stalling over the western Atlantic Ocean at 0000 UTC 3 January 2010 and retrograding into the Gulf of Maine (Fig. 5.10). At 0000 UTC 4 January 2010 the cutoff cyclone began moving northeastward before finally exiting the Northeast cutoff cyclone domain at 1800 UTC on the same day. The four-day NPVU QPE for 1 4 January 2010 indicates that heavy precipitation (>25 mm) occurred throughout most of Maine in addition to the western slopes of the Green Mountains and the Berkshires (Fig. 5.11). Lake-effect snow off of Lakes Erie and 74

91 Ontario and Lake Champlain contributed to the heavy precipitation observed in western New York and the Champlain Valley, respectively. During this cutoff cyclone event, the precipitation distributions varied considerably from one cutoff cyclone day to the next (Figs. 5.12a d). On 1 January 2010, light precipitation (5 10 mm) occurred throughout New England, with mm of precipitation observed in southern Maine (Fig. 5.12a). The heaviest 24-h precipitation associated with this event occurred on 2 January 2010 (Fig. 5.12b). On this day, mm of precipitation was observed throughout most of Maine, while mm of precipitation was observed in regions of northern New York, Vermont, and New Hampshire. Precipitation throughout the Northeast US was once again light on 3 January 2010, with 5 10 mm of precipitation observed across New England, western New York, and the Champlain Valley (Fig. 5.12c). On 4 January 2010, the persistent precipitation was nearing an end, with little or no precipitation associated with the cutoff cyclone observed in the Northeast US (Fig. 5.12d). The focus of the following sections will be on examining the synoptic-scale and mesoscale features that contributed to the varying precipitation distributions on 2 and 3 January 2010, while 1 and 4 January 2010 will not be discussed Meteorological Conditions: 2 January 2010 The heaviest precipitation on 2 January 2010 occurred in the 6-h periods following 0000 UTC and 0600 UTC 3 January Therefore, in order to identify synoptic-scale and mesoscale features that contributed to the heavy precipitation (15 30 mm) in Maine and the moderate precipitation (10 15 mm) in northern New York, 75

92 Vermont, and New Hampshire (Fig. 5.12b), the upper-level, midlevel, and low-level tropospheric conditions at these times will be discussed. At upper levels, the large-scale trough and embedded cutoff cyclone was negatively tilted at 0000 UTC 3 January 2010 and a developing easterly jet was evident poleward of the cutoff cyclone (Fig. 5.13a). At 0600 UTC 3 January 2010, the region of heaviest precipitation in Maine was located in a region of divergence associated with the equatorward exit region of the easterly jet streak (Fig. 5.13b), which likely contributed to ascent over this region. In addition, the 250-hPa zonal wind poleward of the cutoff cyclone exceeded 3σ at 0000 UTC 3 January 2010 (Fig. 5.14), therefore satisfying the 2.5σ threshold determined by Stuart and Grumm (2004, 2006) to be representative of cyclones that are purely cut off from the background westerly flow. At midlevels, a weak lobe of cyclonic absolute vorticity ( s 1 ) extending along the Massachusetts coast, west of the 500-hPa cutoff cyclone center, was evident at 0000 UTC 3 January 2010 (Fig. 5.15a). At 0600 UTC 3 January 2010, the lobe of cyclonic absolute vorticity had strengthened considerably and had moved slightly westward (Fig. 5.15b). The precipitation in New York, Vermont, and New Hampshire occurred downstream of this lobe in a region of inferred differential cyclonic absolute vorticity advection, which likely contributed to favorable QG forcing for ascent over the region. At 0000 UTC 3 January 2010, the 850-hPa northeasterly and easterly flow poleward of the cutoff cyclone advected warm air into regions of northern Maine, Vermont, and New Hampshire (Fig. 5.16). It can be inferred that the Laplacian of warm air was maximized in this region of warm air advection, which would have further contributed to QG forcing for ascent in Vermont and New Hampshire. Q-vector analyses 76

93 indicate that 700-hPa Q-vector convergence was maximized ( Pa m 2 s 1 ) over Maine and northern regions of New Hampshire and Vermont at 0000 UTC and 0600 UTC 3 January 2010, respectively (Figs. 5.17a,b), further confirming that precipitation in these regions was supported by QG forcing for ascent. At 0000 UTC 3 January 2010, a region of 925-hPa frontogenesis along a warm front was evident in eastern Maine (Fig. 5.18a). At 0600 UTC 3 January 2010, the region of frontogenesis had strengthened and moved southwestward across Maine (Fig. 5.18b). A cross section at 0600 UTC 3 January 2010 shows two regions of frontogenesis located between 500 and 600 hpa and near the surface (Fig. 5.18c), suggesting that there was forcing for ascent associated with frontogenesis at both midlevels and low levels, which likely enhanced the precipitation over Maine. Also at low levels, anomalous PW (+1 to +2σ) was advected into northern Maine from the east by the low-level northeasterly flow poleward of the cutoff cyclone (Fig. 5.19). The advection of anomalous PW from the east likely further contributed to the larger precipitation amounts observed Maine, as compared to other regions of the Northeast US. Finally, enhanced precipitation (10 15 mm) was also observed along the Green Mountains and the Berkshires on 2 January 2010 (Fig. 5.12b). Surface observations at 1000 UTC 3 January 2010 show west northwesterly surface winds throughout the Hudson and Champlain Valleys (Fig. 5.20). The direction of the low-level flow resulted in upslope flow and enhanced precipitation along the western slopes of the Green Mountains and the Berkshires, as indicated by a local maximum in base reflectivity values (15 30 dbz) in Fig

94 5.2.3 Meteorological Conditions: 3 January 2010 The heaviest precipitation on 3 January 2010 occurred in the 6-h periods following 1200 and 1800 UTC 3 January Therefore, the focus of this section will be on examining the upper-level, midlevel, and low-level tropospheric conditions at these times to identify the synoptic-scale and mesoscale features that contributed to the light precipitation in New England, western New York, and the Champlain Valley (Fig. 5.12c). At 1200 UTC 3 January 2010 the lobe of cyclonic absolute vorticity at 500 hpa that contributed to precipitation in northern New England on the previous day was nearly out of the region (Fig. 5.21a). This lobe of cyclonic absolute vorticity had weakened over the Atlantic Ocean at 1800 UTC 3 January 2010 and a second lobe of cyclonic absolute vorticity extending westward from the 500-hPa cutoff cyclone center was present over Massachusetts (Fig. 5.21b). Inferred differential cyclonic absolute vorticity advection downstream of this lobe likely supported the precipitation observed throughout New England on 3 January 2010 by contributing to favorable QG forcing for ascent over the region. Persistent advection of warm air into the Northeast US from the north and east at 850 hpa (Fig. 5.22), and thus an inferred maximum in the Laplacian of warm air advection, also contributed to QG forcing for ascent. The precipitation in New England was likely further supported by a quasi-stationary region of 925-hPa frontogenesis associated with the southwestward moving warm front that had stalled over southern New Hampshire (not shown). 78

95 At the surface, PW values throughout the Northeast US at 1200 UTC 3 January 2010 were less than 12 mm (Fig. 5.23), which likely contributed to the low precipitation observed on this day as they likely did on 1 January The 850-hPa flow across the Northeast US was north-northwesterly and cold air (below 12 C) was in place over New York (Fig. 5.22), both of which provided favorable conditions for the development of lake-effect snow bands. Radar and surface observations at 1300 UTC 3 January 2010 show a prominent lake-effect precipitation band in upstate New York collocated with low-level northwesterly flow over Lake Ontario (Fig. 5.24). Surface winds at Burlington, VT, were north northwesterly as compared to light westerly or southwesterly winds in surrounding areas, suggesting that the low-level flow was being channeled through the Champlain Valley. The north northwesterly winds across Lake Champlain provided favorable conditions for the ongoing support of a lake-effect snow band and contributed to the record-breaking snowfall observed at Burlington, VT. Therefore, the lakes acted as a moisture source, despite low PW values throughout the Northeast US, contributing to the light precipitation observed in western New York and the Champlain Valley on 3 January Conceptual Summary Schematic diagrams of the key synoptic-scale features that contributed to the precipitation distributions on 2 and 3 January 2010 in association with the 1 4 January 2010 cutoff cyclone event are presented in Figs. 5.25a,b. On 2 January 2010, precipitation across regions of northern New York, Vermont, and New Hampshire, and 79

96 throughout Maine, was enhanced by: (1) ascent associated with divergence within the equatorward exit region of an easterly jet streak at upper levels; (2) QG forcing for ascent associated with inferred differential cyclonic vorticity advection west of the 500-hPa cutoff cyclone and an inferred maximum in the Laplacian of warm air advection at 850 hpa; (3) a region of frontogenesis along a southwestward moving warm front; and (4) advection of anomalous PW from the east poleward of the cutoff cyclone (Fig. 5.25a). On 3 January 2010, light precipitation in New England occurred in a region of persistent QG forcing for ascent associated with inferred differential cyclonic vorticity advection and warm air advection, while light precipitation in western New York and the Champlain Valley was attributed to lake-effect precipitation (Fig. 5.25b). On 2 January 2010, the cutoff cyclone would have been placed into the HP neutral cutoff composite category. Comparing the schematic diagram for 2 January 2010 (Fig. 5.25a) to the schematic diagram for the HP neutral cutoff category (Fig. 4.4c), the synoptic-scale features contributing to precipitation are comparable in that the primary features include: (1) ascent favored within the exit region of an upper-level jet streak; (2) inferred differential cyclonic absolute vorticity advection downstream of a 500-hPa absolute vorticity maxima; (3) enhancement of precipitation along a surface warm front; and (4) advection of anomalous PW into the Northeast US by the low-level flow poleward of the surface cyclone. While the features remain similar, the two schematic diagrams appear to differ by a rotation of approximately 90. For instance, rather than a southwesterly upper-level jet streak as depicted in the HP neutral cutoff schematic, there was an easterly upper-level jet streak poleward of the cutoff cyclone on 2 January This difference is largely due to the presence of a highly amplified ridge 80

97 associated with a large-scale blocking pattern and the negative phase of the NAO on 2 January 2010 that acted to increase the geopotential height gradient north and east of the cutoff cyclone and contributed to the development of an easterly jet streak poleward of the cutoff cyclone. The cutoff cyclone on 3 January 2010 would have been placed into the LP neutral cutoff composite category. The schematic diagrams for 3 January 2010 (Fig. 5.25b) and the LP neutral cutoff composite category (Fig. 4.5c) are similar, with northwesterly low-level flow west of the cutoff cyclone contributing to light precipitation observed in the southwest quadrant of the cutoff cyclone, in association with lake-effect snow bands. The schematic diagrams differ in that there is an upper-level easterly jet and a southwestward-moving warm front that contributed to light precipitation in New England on 3 January 2010, whereas these features are not evident in the composite schematic diagram. 5.3 The March 2010 Cutoff Cyclone Event Event Overview The March 2010 cutoff cyclone event was a long duration event, with the cutoff cyclone remaining within the Northeast cutoff cyclone domain for approximately 84 h. The event was associated with widespread flooding throughout southern New England and high winds across New Jersey and southern New York. For instance, at 0000 UTC 14 March 2010 a wind gust of 64 kt (33 m s 1 ) was recorded at Kennedy 81

98 International Airport (Grumm 2010). Leading up to the event, NWP models did well forecasting that precipitation would occur; however, the forecast precipitation amounts were much lower than observed and they did not capture the terrain influences well (e.g., Grumm 2010; Stuart 2010b). The March 2010 cutoff cyclone developed from a broad trough in place over the central US and entered the Northeast cutoff cyclone domain at 1200 UTC 13 March 2010 (Fig. 5.26). The track of the cutoff cyclone indicates that the cyclone remained south of Pennsylvania as it traveled eastward toward the Atlantic coast. On 15 March 2010, the cutoff cyclone stalled over the Atlantic Ocean and retrograded southeast of New Jersey becoming reabsorbed into the background westerly flow at 0000 UTC 17 March Extremely heavy precipitation was observed with the March 2010 cutoff cyclone event, as indicated by the four-day NPUV QPE (Fig. 5.27). Locations in eastern Massachusetts and coastal New Hampshire received mm of precipitation, with a second precipitation maximum ( mm) observed in New Jersey. This cutoff cyclone event was associated with precipitation distributions that varied considerably from one cutoff cyclone day to the next (Figs. 5.28a d). The heaviest precipitation associated with this cutoff cyclone event occurred on 13 and 14 March The precipitation distribution for 13 March 2010 indicates that over 25 mm of precipitation was observed throughout coastal regions of the Northeast precipitation domain, while lower precipitation amounts (5 20 mm) were observed throughout the Hudson Valley in eastern New York (Fig. 5.28b). The heaviest precipitation on this day was observed in southern New England and northern New Jersey, with over 100 mm of precipitation. On 82

99 14 March, over 80 mm of precipitation was observed in northern Massachusetts and coastal New Hampshire, while lighter precipitation (>25 mm) persisted across regions of New Jersey (Fig. 5.28c). The focus of the following sections will be on examining 13 and 14 March 2010 to determine what features contributed to the observed precipitation distributions on these respective days Meteorological Conditions: 13 March 2010 The heaviest precipitation on 13 March 2010 occurred in the 6-h periods following 1800 UTC 13 March 2010 and 0000 UTC 14 March Upper-level, midlevel, and low-level tropospheric conditions at these times will be discussed to identify the synoptic-scale and mesoscale features that contributed to the heavy precipitation in southern New England and northern New Jersey (Fig. 5.28b). At upper levels, an easterly jet greater than 35 m s 1 (68 kt) was forming poleward of the cutoff cyclone at 1800 UTC 13 March 2010 and extended farther west across Pennsylvania by 0000 UTC 14 March 2010 (Figs. 5.29a,b). Divergence was evident within the poleward entrance region of the easterly jet streak over Pennsylvania, New Jersey, and southern New England, which likely contributed to ascent in these regions of observed heavy precipitation. The 250-hPa zonal wind anomaly at 0000 UTC 14 March 2010 exceeded 3σ poleward of the cutoff cyclone (Fig. 5.30), indicating that the cyclone was purely separated from the background westerly flow (e.g., Stuart and Grumm 2004, 2006). 83

100 At 700 hpa, Q-vector convergence was evident east of the cutoff cyclone (Fig. 5.31), indicative of favorable QG forcing for ascent that likely contributed to the heavy precipitation observed in New Jersey on 13 March At low levels, a strong (>60 kt) southeasterly jet was in place across the Northeast US (Fig. 5.32), corresponding to zonal wind between 3 and 5σ (not shown). The onshore low-level flow was favorable for advection of moist air from the east, resulting in +1 to +3σ PW values across the Northeast US (Fig. 5.33). The anomalous PW advection into the Northeast US likely further contributed to the heavy precipitation observed in southern New England and New Jersey on 13 March Despite the widespread heavy precipitation observed on 13 March 2010, there were also regions of suppressed precipitation associated with downslope flow. For example, the Hudson Valley in eastern New York received only 5 15 mm of precipitation, while surrounding areas received greater than 25 mm (Fig. 5.28b). Surface observations at 1800 UTC 13 March 2010 indicate that there were easterly surface winds throughout the Northeast US (Fig. 5.34). The direction of the low-level flow resulted in downslope flow and suppression of precipitation throughout the Hudson Valley in eastern New York, as indicated by a local minimum in base reflectivity values Meteorological Conditions: 14 March 2010 The heaviest precipitation on 14 March 2010 occurred in the 6-h periods following 1200 and 1800 UTC 14 March Hence, the focus of this section will be on examining upper-level, midlevel, and low-level tropospheric conditions at these time 84

101 periods in order to identify the synoptic-scale and mesoscale features contributing to the heavy precipitation across northern Massachusetts, coastal New Hampshire, and New Jersey (Fig. 5.28c). At upper levels, the easterly jet streak poleward of the cutoff cyclone persisted and divergence within the entrance and exit regions of this jet streak continued to provide favorable conditions for ascent over the region of heaviest precipitation in northern Massachusetts (not shown). At 500 hpa, a lobe of cyclonic absolute vorticity northeast of the cutoff cyclone center was evident at 1200 UTC 14 March 2010 (Fig. 5.35a) and at 1800 UTC this lobe of cyclonic absolute vorticity had moved westward across southern New England and over New Jersey (Fig. 5.35b). In addition, at low levels, southeasterly flow east of the cutoff cyclone resulted in advection of warm air into New England (Fig. 5.36). By application of Eq. 4.1, favorable QG forcing for ascent, associated with inferred differential cyclonic vorticity advection downstream the lobe of 500-hPa cyclonic absolute vorticity and an inferred maximum in the Laplacian of warm air advection at 850 hpa over New England, likely contributed to the precipitation observed. The orientation of the southeasterly low-level jet continued to favor advection of anomalous warm, moist air into the coastal regions, as indicated by the 850-hPa equivalent potential temperature field at 1200 UTC 14 March 2010 (Fig. 5.37). At this time, advection of warm, moist air was maximized [>20 K (3 h) 1 ] along coastal Massachusetts. The strong equivalent potential temperature advection was collocated with the region of heaviest precipitation observed across northern Massachusetts and coastal New Hampshire on this day. In addition, a region of 925-hPa frontogenesis developed in southern New England at 1200 UTC 14 March 2010 (Fig. 5.38), in 85

102 association with the equivalent potential temperature advection. This region of frontogenesis remained quasi-stationary over the following 24 h (not shown), likely acting to further enhance precipitation in Massachusetts and coastal New Hampshire Conceptual Summary Schematic diagrams of the key synoptic-scale features that contributed to the precipitation distributions on 13 and 14 March 2010 are presented in Figs. 5.39a,b. On 13 March 2010, the heavy precipitation in southern New England was supported by favorable conditions for ascent in association with divergence within the entrance region of an easterly jet streak at upper levels (Fig. 5.39a). Advection of anomalous warm, moist air by a strong southeasterly low-level jet also contributed to the heavy precipitation amounts observed on this day in both southern New England and New Jersey. On 14 March 2010, support for precipitation in northern Massachusetts and coastal New Hampshire was provided by persistent upper-level and low-level jet streaks in addition to a region of frontogenesis that had developed along the coast (Fig. 5.39b). Precipitation in New Jersey on 14 March 2010 was lighter than on 13 March 2010; however, because of the proximity to the midlevel cutoff cyclone center, precipitation continued due to favorable QG forcing for ascent provided by inferred differential cyclonic absolute vorticity advection in the lower troposphere and an inferred maximum in the Laplacian of warm air advection at 850 hpa. Both the 13 and 14 March 2010 cutoff cyclone days would have been placed into the HP neutral cutoff composite category. The 13 and 14 March 2010 schematic 86

103 diagrams (Figs. 5.39a,b) compare well with the HP neutral cutoff composite schematic diagram (Fig. 4.4c). Several contributing features to heavy precipitation depicted in the composite schematic diagram were confirmed by the examination of 13 and 14 March 2010: (1) the location of the cutoff cyclone was to the southeast of the Northeast precipitation domain; (2) the heaviest precipitation occurred in the northeast quadrant of the cyclone; (3) forcing for ascent was supported by divergence associated with an upper level jet streak; (4) southeasterly low-level flow advected Atlantic moisture into the region; and (5) a surface warm front acted to locally enhance precipitation. 87

104 Fig NCEP Global Ensemble Forecast System valid at 1200 UTC 3 February 2009 and initialized at (a,b) 0000 UTC 30 January, (c,d) 1200 UTC 30 January, and (e,f) 1800 UTC 30 January. Left panels show 1008 and 1020 hpa MSLP isobars of each member (hpa, colored contours), the mean of all members (hpa, black contour), and the spread about the mean (hpa, shaded). Right panels show the average MSLP isobars of all members (hpa, green contours) and the standardized anomalies computed from the mean (σ, shaded). (Figure modified from Grumm et al. 2009, Fig. 13.) 88

105 Fig Mean 500-hPa geopotential height (dam, black contours) for 0000 UTC 3 February 0000 UTC 4 February 2009 and the track of the 500-hPa cutoff cyclone center every 6 h (red contours). The brown bold lines depict the Northeast cutoff cyclone domain. Fig Two-day NPVU QPE (mm, shaded) ending 1200 UTC 4 February The black bold line depicts the Northeast precipitation domain. 89

106 Fig hPa geopotential height (dam, black solid contours), wind speed (m s 1, shaded), and divergence (10 5 s 1, red dashed contours) at (a) 1800 UTC 3 February 2009 and (b) 0000 UTC 4 February Fig hPa geopotential height (dam, black solid contours), absolute vorticity (10 5 s 1, shaded), cyclonic absolute vorticity advection [10 5 s 1 (3 h) 1, blue dashed contours] and wind (kt, barbs) at (a) 1800 UTC 3 February 2009 and (b) 0000 UTC 4 February

107 Fig hPa geopotential height (dam, black solid contours), temperature ( C, shaded), and wind (>30 kt, barbs) at 1800 UTC 3 February Fig hPa geopotential height (dam, black solid contours), temperature ( C, green dashed contours), Q vectors (>5 x 10 7 Pa m 1 s 1, arrows), and Q-vector convergence (10 12 Pa m 2 s 1, shaded) at (a) 1800 UTC 3 February 2009 and (b) 0000 UTC 4 February

108 Fig MSLP (hpa, black solid contours), hpa thickness (m, red dashed contours), and PW (mm, shaded) at 1800 UTC 3 February Fig Schematic depicting the 500-hPa geopotential height (dam, black contours) at 1800 UTC 3 February 2009 and key synoptic-scale features that contribute to precipitation for the 2 3 February 2009 cutoff cyclone event. The brown bold line depicts the Northeast precipitation domain. 92

109 Fig Mean 500-hPa geopotential height (dam, black contours) for 0000 UTC 2 January 1200 UTC 4 January 2010 and the track of the 500-hPa cutoff cyclone center every 6 h (red contours). The brown bold lines depict the Northeast cutoff cyclone domain. Fig Four-day NPVU QPE (mm, shaded) ending 1200 UTC 5 January The black bold line depicts the Northeast precipitation domain. 93

110 Fig h NPVU QPE (mm, shaded) ending (a) 1200 UTC 2 January 2010, (b) 1200 UTC 3 January 2010, (c) 1200 UTC 4 January 2010, and (d) 1200 UTC 5 January The black bold line depicts the Northeast precipitation domain. Fig As in Fig. 5.4 except at (a) 0000 UTC 3 January 2010 and (b) 0600 UTC 3 January

111 Fig hPa geopotential height (dam, black contours) and standardized anomalies of 250-hPa zonal wind (σ, shaded) at 0000 UTC 3 January Fig As in Fig. 5.5 except at (a) 0000 UTC 3 January 2010 and (b) 0600 UTC 3 January

112 Fig As in Fig. 5.6 except at 0000 UTC 3 January Fig hPa geopotential height (dam, black solid contours), temperature ( C, green dashed contours), Q vectors (>5 x 10 7 Pa m 1 s 1, arrows), and Q-vector convergence (10 12 Pa m 2 s 1, shaded) at (a) 0000 UTC 3 January 2010 and (b) 0600 UTC 3 January

113 Fig hPa frontogenesis [K (100 km) 1 (3 h) 1, shaded], potential temperature (K, black solid contours), and wind (kt, barbs) at (a) 0000 UTC 3 January 2010 and (b) 0600 UTC 3 January 2010; (c) cross section of 925-hPa frontogenesis [K (100 km) 1 (3 h) 1, shaded], potential temperature (K, black solid contours), and omega (μb s 1, dashed contours; upward is indicated in red, downward is indicated in blue) at 0600 UTC 3 January The blue dashed line in (b) indicates the approximate location of the cross section in (c). 97

114 Fig hPa geopotential height (dam, black solid contours), wind (kt, barbs), PW (mm, red dashed contours), and standardized anomalies of PW (σ, shaded) at 0000 UTC 3 January Fig Base reflectivity (dbz) and surface observations at 1000 UTC 3 January

115 Fig As in Fig. 5.5 except at (a) 1200 UTC 3 January 2010 and (b) 1800 UTC 3 January Fig As in Fig. 5.6 except at 1200 UTC 3 January

116 Fig As in Fig. 5.7 except at 1200 UTC 3 January Fig As in Fig except at 1300 UTC 3 January

117 Fig As in Fig. 5.8 except for (a) 2 January 2010 and (b) 3 January

118 Fig Mean 500-hPa geopotential height (dam, black contours) for 0600 UTC 13 March 1800 UTC 16 March 2010 and the track of the 500-hPa cutoff cyclone center every 6 h (red contours). The brown bold lines depict the Northeast cutoff cyclone domain. Fig Four-day NPVU QPE (mm, shaded) ending 1200 UTC 16 March The black bold line depicts the Northeast precipitation domain. 102

119 Fig h NPVU QPE (mm, shaded) ending (a) 1200 UTC 13 March 2010, (b) 1200 UTC 14 March 2010, (c) 1200 UTC 15 March 2010, and (d) 1200 UTC 16 March The black bold line depicts the Northeast precipitation domain. Fig As in Fig. 5.4 except at (a) 1800 UTC 13 March 2010 and (b) 0000 UTC 14 March

120 Fig At in Fig except at 0000 UTC 14 March Fig As in Fig except at 0000 UTC 14 March

121 Fig hPa geopotential height (dam, black contours), wind speed (m s 1, shaded), and wind (kt, barbs) at 0000 UTC 14 March Fig As in Fig except at 0000 UTC 14 March

122 Fig As in Fig except at 1800 UTC 13 March Fig As in Fig. 5.5 except at (a) 1200 UTC 14 March 2010 and (b) 1800 UTC 14 March

123 Fig As in Fig. 5.6 except at 1200 UTC 14 March Fig hPa equivalent potential temperature (K, black contours), equivalent potential temperature advection [K (3 h) 1, shaded], and wind (m s 1, barbs) at 1200 UTC 14 March

124 Fig As in Fig. 5.17a except at 1200 UTC 14 March

125 Fig As in Fig. 5.8 except for (a) 13 March 2010 and (b) 14 March