Cyclone Wind Field Asymmetries during Extratropical Transition in the Western North Pacific

ABSTRACT

Risk-assessment systems for wind hazards (e.g., hurricanes or typhoons) often rely on simple parametric wind field formulations. They are built using extensive observations of tropical cyclones and make assumptions about wind field asymmetry. In this framework, maximum winds are always simulated to the right of the cyclone, but analysis of the Climate Forecast System Reanalysis database for the western North Pacific Ocean suggests that wind fields from cyclones undergoing extratropical transition around Japan often present features that cannot be adequately simulated under these assumptions. These "left-hand-side contribution" (LHSC) wind fields exhibit strong winds on both sides of the moving cyclone with the maximum magnitude often located to the left. Classification of cyclones in terms of their most frequent patterns reveals that 67% of cases that make a transition around Japan are dominantly LHSC. They are more likely in autumn and have more intense maximum winds. The results from this study show the need for a new approach to the modeling of transitioning wind fields in the context of risk-assessment systems.

(ProQuest: ... denotes formula omitted.)

1. Introduction

Given its meridional extent and location within the Pacific Ocean basin, Japan is regularly impacted by strong winds from mature tropical cyclones (TCs), ''transition- ing'' systems (storms that are making the transition from tropical to extratropical status), or extratropical (ET) systems. To quantify the potential impact of extreme events (e.g., crop/property damage and population safety) risk-assessment models aim to simulate surface wind fields from thousands of stochastic events that ex- trapolate historical records (Powell et al. 2005; Vickery et al. 2009). Risk models use dynamically based ap- proaches rather than statistical extrapolations of history because they are better suited for the design of risk- assessment scenarios. In this context, a compromise is required between the need to simulate wind field var- iability around the cyclone during its entire evolution and run-time constraints.

Satellite products (Hoffman and Leidner 2005; Knaff et al. 2011) have been used along with other instruments such as dropsondes and flight observations (Powell et al. 1998) to study TC wind structure (Shea and Gray 1973; Willoughby and Rahn 2004; Kossin et al. 2007; Chan and Chan 2012). These show strong left/right asymmetries in the wind fields of mature tropical systems [e.g., Fig. 1, from the Climate Forecast System Reanalysis (CFSR) database; Saha et al. 2010]. Consistent with these ob- servations, a range of parametric wind models was de- veloped under the assumption that this asymmetry results from the superposition of the cyclone forward motion onto an axisymmetric wind field (Phadke et al. 2003; MacAfee and Pearson 2006). In this framework the location and magnitude of the maximum wind relative to the cyclone center are usually a model input while a range of parameters controls the shape of the wind field (DeMaria et al. 1992; Houston and Powell 1994; Willoughby et al. 2006). A correction factor that is based on the cyclone's translational speed then enables pro- jection of the storm motion-relative wind field onto the fixed frame of reference (Jelesnianski 1966; MacAfee and Pearson 2006). Note that similar assumptions are often made when deriving TC wind structure from infrared satellite imagery (Mueller et al. 2006). The magnitude of the left/right asymmetry depends on the model for- mulation and forward velocity correction, but in each case the maximum modeled winds are on the right of the track. Although very simple, these models have shown an ability to capture the key aspects of tropical wind fields and are often used in risk-assessment systems for which the limited number of inputs and fast run times are major advantages.

One of the shortcomings of the type of data listed above is that coverage is often discontinuous in space and time because of various instrument limitations or gaps in satellite swaths (Mueller et al. 2006; Brennan et al. 2009; Chan and Chan 2012). For studies that in- vestigate the climatological behavior of TCs, reanalyses are increasingly used (Maloney and Hartmann 2000; Hart 2010) because they provide consistent datasets over lon- ger time periods (Thorne and Vose 2010).

As with every data analysis study, there is a degree of sensitivity to the dataset selected. For example, Schenkel and Hart (2012) highlight discrepancies in TC position and intensity from five reanalysis products and dem- onstrate that intensity estimates improve with TC size; they suggest that the expansion of the wind field during transitioning makes the storms easier to resolve. Their analysis identifies CFSR and the Japanese 25-year Reanalysis Project (JRA-25; Onogi et al. 2007) as the two best products. For the transitioning/ET phase where observations are scarce, reanalyses therefore offer an alternative (Evans and Hart 2003; Kitabatake and Fujibe 2009). Archambault et al. (2013) show that the 0.58- resolution CFSR fields are consistent with satellite images for a selected transitioning cyclone in the western North Pacific Ocean (WNP). In particular they demon- strate that the interaction of the cyclone with the extra- tropical flow is well captured. Chan and Chan (2013) also use the 0.58 CFSR to characterize the synoptic flow patterns that influence TCs in the WNP. For the higher- resolution (;38 km) CFSR surface winds, recent eval- uations against buoy observations show good skill in the Pacific (Chawla et al. 2013), although they link the poorest performances to TCs. Following these recent studies, we here base our analysis on the high-resolution CFSR data.

In a study using JRA-25, Kitabatake (2011) estimated that 40% of WNP cyclones transition. From a risk- assessment point of view, it is critical to assess whether tropical wind models are suitable to represent the as- sociated wind fields. In particular, the validity of the asymmetry assumption is questionable because historical events such as Tokage 2004 have highlighted the poten- tial for damaging winds on both sides of a storm track, and Fujibe and Kitabatake (2007) and later Kitabatake and Fujibe (2009) showed that such types of wind fields are common for cyclones transitioning around Japan. Note that the recent study by Sousounis et al. (2012) tackles a similar issue from the point of view of asym- metries in the JRA-25 pressure field.

In this paper we use the CFSR dataset (1979-2010) to directly quantify the left/right asymmetry in wind mag- nitude. The objective is to determine whether tropical wind field formulations are suitable for the transitioning/ ET phase. A key finding is that for two-thirds of the cy- clones transitioning around Japan they appear not to be.

2. CFSR wind field asymmetry

Six-hourly CFSR wind snapshots (;38-km resolution; e.g., Fig. 1) are used along with Japan Meteorological Agency (JMA) information on track and wind parame- ters as reported by the International Best Track Archive for Climate Stewardship (IBTrACS; Knapp et al. 2010). Only cyclones with at least two snapshots with CFSR winds . 17.2 m s21 in the Fig. 1 domain are kept. This threshold matches the one from Fujibe and Kitabatake (2007), who quantified the right/left wind asymmetry from station observations during typhoon landfall in Japan. One of the five wind patterns in their cluster analysis (cluster ''C4'') exhibited a very weak right/left asymmetry relative to the others, and Kitabatake and Fujibe (2009) later showed it to be mainly composed of cyclones undergoing ET.

Twowindmaximaareusedheretoquantifythis asymmetry, Vmax.rhs and Vmax.lhs, which represent the peak winds on the right- and left-hand sides of the track-heading direction. They are computed within an 800km 3 800 km box centered on the cyclone and ori- ented along the heading direction (see the black dashed boxinFig.1).ThetermVmax.rhs is the maximum magni- tude inthe 400km3 800km box tothe right ofthe black arrow (Fig. 1); it represents the peak wind modeled by parametric formulations. The term Vmax.lhs is defined as the maximum value recorded at least 100 km to the left of the track (the gray box in Fig. 1). This setting is designed to test the limitations of parametric formulations: if peak winds at a distance of at least 100 km to the left of the track can reach a magnitude that is similar to (or larger than) the ones to the right of it, they would not be ap- propriately modeled with a tropical formulation such as the ones that are listed in the introduction.

Using the IBTrACS end-of-transition time stamp for each cyclone and estimates of transition duration from Kitabatake (2011), the CFSR database is split in two groups: 1) tropical cases are characterized by all 6-hourly snapshots that are at least 96 h before the cyclone is flagged as ET and 2) transitioning/ET cases are char- acterized by all snapshots that are 24 h before the flag or later (note that changing this criterion to 18 or 30 h has little effect on the results). When one plots Vmax.lhs as a function of Vmax.rhs for the two groups (Fig. 2), the distinct behavior of transitioning/ET wind fields becomes obvious. In the tropical case (Fig. 2a), the majority of points are below the 1:1 line and the slope of the linear regression (dashed line) is 0.747. Most important for risk- assessment applications, the strongest winds recorded (.40 m s21) are on the right of the track and are re- sponsible for the largest differences relative to their corresponding left-hand-side value (i.e., these should be well represented by tropical formulations). On the other hand, Fig. 2b shows that for transitioning/ET cases the magnitude of the winds on both sides of the track is much more similar, with the slope of the regression line being 0.945. Stronger winds are often observed to the left of the track, especially for the largest values recorded (.35 m s21). Hereinafter, our focus is on transitioning/ ET storms.

In the rest of the study we consider the following ratio to characterize left/right asymmetry:

...

it is similar to the ratio from Fujibe and Kitabatake (2007) except that we use maximum winds instead of mean values to better represent the focus of risk-assessment tools (i.e., to capture the extreme). We define a threshold value of DVmax 5 0.20 (the blue dashed line in Fig. 3) and distinguish two cases:

1) The first case is wind fields with strong left/right asymmetry, characterized by DVmax $ 0.20 (i.e., Vmax.lhs is at least 20% weaker than Vmax.rhs). These are designated as ''right-hand side only'' (RHSO) and provide a similar framework to the one in which tropical models apply.

2) The second case is wind fields with a significant left- hand-side contribution, characterized by DVmax , 0.20 (i.e., Vmax lhs is within 20% of Vmax.rhs or is larger than Vmax.rhs). These are designated as ''left-hand-side contribution'' (LHSC). Given the relative magnitude of the left-hand-side winds, the forward velocity correction would need to be limited, suppressed, or reversed in any attempt to simulate this type of wind field with a tropical formulation. The corresponding modeled winds would result in a (close to) axisym- metric field with large circular areas of strong winds. This situation has implications for risk-assessment models, and a closer look at typical LHSC wind fields (section 3) will show how inadequate the above assumption is.

Note that this definition is based on an arbitrary choice of a DVmax threshold that is designed to split the wind fields according to how well they match tropical-model assumptions. Other similar thresholds could be selected (i.e., one could change the slope of the blue dashed line; Fig. 3) to reflect different views of what is an acceptable relative magnitude to assume left/right asymmetry.

3. Climatological description of transitioning cyclones

A total of 189 cyclones were selected with at least two CFSR snapshots in the domain formed by the solid box ofFig.11)24hbeforetheIBTrACSETflag orlaterand 2)withCFSRwinds.17.2ms21.Thischoiceleaves1091 snapshots. Note that cases in which another approaching cyclone (or other weather system) contaminates the pro- cedure were filtered out. In what follows, each cyclone is assigned the category (RHSO/LHSC) of its most frequent wind field (Fig. 3). With 86% of selected cases having more than 75% of their snapshots in one category, these results suggest that a cyclone is dominantly of one type or the other. This result is emphasized by Fig. 3, where the colors/symbols refer to the cyclone category: although some blue triangles are above the blue dashed line (i.e., LHSC snapshots belonging to a cyclone classified as RHSO) and red circles are below it (RHSO snapshots belonging to an LHSC cyclone), these are relatively un- common. According to that classification scheme, 67% of transitioning cyclones are LHSC and 33% are RHSO. This result confirms the need for a careful analysis of the LHSC type and the specificities of its wind field.

Recent research on transitioning has focused on the interaction between TCs and the midlatitude circulation (McTaggart-Cowan et al. 2001; Jones et al. 2003; Agusti- Panareda et al. 2004; Ritchie and Elsberry 2007). From analysis of the 250-hPa winds available from CFSR (0.58 resolution), specific patterns appear for LHSC and RHSO storms.

For LHSC, the area of strong winds extends from the right of the track to the left-rear quadrant. This type of wind field is also described as ''horseshoe'' (e.g., Edson 2004) or ''comma shape'' (Kitabatake et al. 2007) in the literature and would belong to the C4 cluster from Kitabatake and Fujibe's (2009) classification. Maximum winds tend to occur either to the right of the track or in the left-rear quadrant (e.g., Fig. 4a) and typically flip from one side of the track to the other within a given storm. Figure 4 shows a typical example with Rammasun 2008 18 h before it was flagged as ET. The cyclone enters a strong baroclinic region (Fig. 4b) with the trough in the upper-level jet in a configuration that enhances the development of the transitioning system at the surface (Fig. 4c) as a result of strong upper-level divergence (right entrance region of the jet; Hanley et al. 2001).

For RHSO, the area of stronger winds is to the right of the track and the left-hand-side contribution is much weaker. In this configuration, the upper-level jet is not conducive to any enhancement of the transitioning cy- clone and the surface system is swept away. An example of such a configuration is presented in Fig. 5 for Halong 2002.

This classification allows simplification of very complex phenomena (see Fig. 1 from Kitabatake 2011) in a way that is useful for risk-assessment applications. Wind fields from RHSO cyclones can be approximated by using tropical formulations (after calibration), but this is not the case for the horseshoe patterns, and alternative formu- lations are needed to directly parameterize the horseshoe shape. With this approach, a decision can be made on the proportion of transitioning cyclones for which each model should be applied in a stochastic event set (33%: recalibrated tropical; 67%: horseshoe). Note that the development of a horseshoe parametric model would also help to characterize better the transitioning/ET wind structure from infrared satellite data [i.e., following the method of Mueller et al. (2006)].

As a final step, we briefly introduce some potential indicators of the storm type on the basis of their likely interaction with the jet. These four potential predictors are taken from JMA reported values 12 h before the cyclone is flagged as ET (Fig. 6). 1) LHSC storms are more frequent in the months of September and October than earlier in the year, and RHSO cases are more fre- quent during the summer (Fig. 6a): stronger and with an increased meridional extent in September/October (Archambault et al. 2013), the jet is more likely to in- teract and enhance the formation of left-hand-side winds. 2) Figure 6b suggests that tracks heading north northeast during transitioning are more likely to end up in an LHSC pattern (higher potential for a positive interaction with the westerly jet) than tracks heading east northeast. 3) Con- firming the initial results from Fig. 3, the analysis of JMA values of maximum winds (Fig. 6c) shows that the most intense cyclones tend to be LHSC types. 4) Although the trend in the data is very weak, it would suggest that faster- moving cyclones are more likely to be RHSO (Fig. 6d). Also, a two-sample Kolmogorov-Smirnov test shows that the RHSO and LHSC histograms are statistically different for the decimal-month and Vmax cases (Figs. 6a,c) but not for the other two.

4. Conclusions

Analysis of 31 years of CSFR data suggests that wind fields from cyclones transitioning around Japan often present features that cannot be adequately represented by the parametric wind formulations that are used in risk-assessment models. These left-hand-side contribu- tion wind fields exhibit strong winds on both sides of the moving cyclone with the maximum magnitude located either to the right or to the left in a pattern often de- scribed as a horseshoe. Classification of cyclones in terms of their most frequent patterns reveals that 67% of cases are dominantly LHSC. A first-order analysis of cyclone characteristics 12 h before it completes its transition highlights some potential predictors for the likelihood of an LHSC case.

This study shows that, in the context of risk assess- ment, a new approach is needed for two-thirds of cy- clones transitioning in the vicinity of Japan. They can be primarily characterized as more likely in autumn and as having more intense maximum winds. This has some potential to determine when a new horseshoe model should be used. For the other one-third of cases, tropical formulations are applicable, provided that they are calibrated accordingly.

Acknowledgments. The authors thank Dr. John Knaff for his useful comments and advice as well as three anonymous reviewers. We also thank NCEP for pro- vision of the CFSR reanalysis and JMA/IBTrACS for the track data.

REFERENCES

Agusti-Panareda, A., C. D. Thorncroft, G. C. Craig, and S. L. Gray, 2004: The extratropical transition of Hurricane Irene (1999): A potential vorticity perspective. Quart. J. Roy. Meteor. Soc., 130, 1047-1074.

Archambault, H. M., L. F. Bosart, D. Keyser, and J. M. Cordeira, 2013: A climatological analysis of the extratropical flow re- sponse to recurving western North Pacific tropical cyclones. Mon. Wea. Rev., 141, 2325-2346.

Brennan, M. J., C. C. Hennon, and R. D. Knabb, 2009: The oper- ational use of QuikSCAT ocean surface vector winds at the National Hurricane Center. Wea. Forecasting, 24, 621-645.

Chan, K. T. F., and J. C. L. Chan, 2012: Size and strength of tropical cyclones as inferred from QuikSCAT data. Mon. Wea. Rev., 140, 811-824.

_____, and _____, 2013: Angular momentum transports and synoptic flow patterns associated with tropical cyclone size change. Mon. Wea. Rev., 141, 3985-4007.

Chawla, A., D. M. Spindler, and H. L. Tolman, 2013: Validation of a thirty year wave hindcast using the Climate Forecast System Reanalysis winds. Ocean Modell., 70, 189-206.

DeMaria, M., S. Aberson, K. V. Ooyama, and S. J. Lord, 1992: A nested spectral model for hurricane track forecasting. Mon. Wea. Rev., 120, 1628-1643.

Edson, R. T., 2004: Tropical cyclone analysis techniques from QuikSCAT NRCS, wind and ambiguity data and microwave imagery. Preprints, 26th Conf. on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., P1.8. [Available online at https://ams.confex.com/ams/pdfpapers/75828.pdf.]

Evans, J. L., and R. E. Hart, 2003: Objective indicators of the onset and completion of ET for Atlantic tropical cyclones. Mon. Wea. Rev., 131, 909-925.

Fujibe, F., and N. Kitabatake, 2007: Classification of surface wind fields of tropical cyclones at landfall on the Japan main islands. J. Meteor. Soc. Japan, 85, 747-765.

Hanley, D., J. Molinari, and D. Keyser, 2001: A composite study of the interactions between tropical cyclones and upper- tropospheric troughs. Mon. Wea. Rev., 129, 2570-2584.

Hart, R., 2010: An inverse relationship between aggregate North- ern Hemisphere tropical cyclone activity and subsequent winter climate. Geophys. Res. Lett., 38, L01705, doi:10.1029/ 2010GL045612.

Hoffman, R. N., and S. M. Leidner, 2005: An introduction to the near-real-time QuikSCAT data. Wea. Forecasting, 20, 476-493.

Houston, S. H., and M. D. Powell, 1994: Observed and modeled wind and water-level response from tropical storm Marco (1990). Wea. Forecasting, 9, 427-439.

Jelesnianski, C. P., 1966: Numerical computations of storm surges without bottom stress. Mon. Wea. Rev., 94, 379-394.

Jones, S. C., and Coauthors, 2003: The extratropical transition of tropical cyclones: Forecast challenges, current understanding, and future directions. Wea. Forecasting, 18, 1052-1092.

Kitabatake, N., 2011: Climatology of extratropical transition of tropical cyclones in the western North Pacific defined by using cyclone phase space. J. Meteor. Soc. Japan, 89, 309-325.

_____, and F. Fujibe, 2009: Relationship between surface wind fields and three-dimensional structures of tropical cyclones land- falling in the main islands of Japan. J. Meteor. Soc. Japan, 87, 959-977.

_____, S. Hoshino, K. Bessho, and F. Fujibe, 2007: Structure and intensity change of Typhoon Songda (0418) undergoing ex- tratropical transition. Pap. Meteor. Geophys., 58, 135-153.

Knaff, J. A., M. DeMaria, D. A. Molenar, C. R. Sampson, and M. G. Seybold, 2011: An automated, objective, multiple-satellite- platform tropical cyclone surface wind analysis. J.Appl. Meteor. Climatol., 50, 2149-2166.

Knapp,K.R., M.C.Kruk,D.H.Levinson,H. J.Diamond,andC.J. Neumann, 2010: The International Best Track Archive for Climate Stewardship (IBTrACS): Unifying tropical cyclone best track data. Bull. Amer. Meteor. Soc., 91, 363-376.

Kossin, J. P., J. A. Knaff, H. I. Berger, D. C. Herndon, T. A. Cram, C. S. Velden, R. J. Murnane, and J. D. Hawkins, 2007: Esti- mating hurricane wind structure in the absence of aircraft reconnaissance. Wea. Forecasting, 22, 89 101.

MacAfee, A. W., and G. M. Pearson, 2006: Development and testing of tropical cyclone parametric wind models tailored for midlatitude application-Preliminary results. J. Appl. Meteor. Climatol., 45, 1244-1260.

Maloney, E., and D. Hartmann, 2000: Modulation of eastern North Pacific hurricanes by the Madden-Julian oscillation. J. Climate, 13, 1451-1460.

McTaggart-Cowan, R., J. R. Gyakum, and M. K. Yau, 2001: Sensi- tivitytesting of extratropical transitions usingpotential vorticity inversions to modify initial conditions: Hurricane Earl case study. Mon. Wea. Rev., 129, 1617-1636.

Mueller, K. J., M. DeMaria, J. A. Knaff, J. P. Kossin, and T. H. Vonder Haar, 2006: Objective estimation of tropical cyclone wind structure from infrared satellite data. Wea. Forecasting, 21, 990-1005.

Onogi, K., and Coauthors,2007: The JRA-25 Reanalysis. J. Meteor. Soc. Japan, 85, 369-432.

Phadke, A. C., C. D. Martino, K. F. Cheung, and S. H. Houston, 2003: Modeling of tropical cyclone winds and waves for emer- gency management. Ocean Eng., 30, 553-578.

Powell, M. D., S. H. Houston, L. R. Amat, and N. Morisseau-Leroy, 1998: The HRD real-time hurricane wind analysis system. J. Wind Eng. Ind. Aerodyn., 77-78, 53-64.

_____, G. Soukup, S. Cocke, S. Gulati, N. Morisseau-Leroy, S. Hamid, N. Dorst, and L. Axe, 2005: State of Florida hurri- cane loss projection model: Atmospheric science component. J. Ind. Aerodyn., 93, 651-674.

Ritchie, E. A., and R. L. Elsberry, 2007: Simulations of the extra- tropical transition of tropical cyclones: Phasing between the upper-level trough and tropical cyclones. Mon. Wea. Rev., 135, 862-876.

Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015-1057.

Schenkel, B. A., and R. E. Hart, 2012: An examination of tropical cyclone position, intensity, and intensity life cycle within at- mospheric reanalysis datasets. J. Climate, 25, 3453-3475.

Shea, D. J., and W. M. Gray, 1973: The hurricane's inner core re- gion: Symmetric and asymmetric structure. J. Atmos. Sci., 30, 1544-1564.

Sousounis, P. J., J. Butke, and K. Hill, 2012: Tropical cyclone risk models-circular symmetry revisited. Preprints, 30th Conf. on Hurricanes and Tropical Meteorology, Ponte Vedra Beach, FL, Amer. Meteor. Soc., 7D.7. [Available online at https:// ams.confex.com/ams/30Hurricane/webprogram/Manuscript/ Paper205892/AMS.2012.30th.Tropical.ExtAbs.PS%26JB%26 KH.2.pdf.]

Thorne, P., and R. Vose, 2010: Reanalyses suitable for character- izing long-term trends. Bull. Amer. Meteor. Soc., 91, 353-361.

Vickery, P. J., F. J. Masters, M. D. Powell, and D. Wadhera, 2009: Hurricane hazard modeling: The past, present, and future. J. Wind Eng. Ind. Aerodyn., 97 (7-8), 392-405.

Willoughby, H. E., and M. E. Rahn, 2004: Parametric representa- tion of the primary hurricane vortex. Part I: Observations and evaluation of the Holland (1980) model. Mon. Wea. Rev., 132, 3033-3048.

_____, R. W. R. Darling, and M. E. Rahn, 2006: Parametric represen- tation of the primary hurricane vortex. Part II: A family of sec- tionally continuous profiles. Mon. Wea. Rev., 134, 1102-1120.

T. LORIDAN,E.SCHERER,M.DIXON,E.BELLONE, AND S. KHARE

Risk Management Solutions, London, United Kingdom

(Manuscript received 13 August 2013, in final form 25 October 2013)

Corresponding author address: Thomas Loridan, Risk Manage- ment Solutions, Model Development, Peninsular House, 30 Mon- ument St., London, EC3R 8NB, United Kingdom.

E-mail: [email protected]