JANUARY 2016 GUHA ET AL. 393 Modeling Rossby Wave Breaking in the Southern Spring Stratosphere ANIRBAN GUHA* AND CARLOS R. MECHOSO Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, Los Angeles, California CELAL S. KONOR AND ROSS P. HEIKES Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado (Manuscript received 17 March 2015, in final form 29 September 2015) ABSTRACT Rossby wave breaking (RWB) plays a central role in the evolution of stratospheric flows. The generation and evolution of RWB is examined in the simple dynamical framework of a one-layer shallow-water system on a sphere. The initial condition represents a realistic, zonally symmetric velocity profile corresponding to the springtime southern stratosphere. Single zonal wavenumber Rossby waves, which are either stationary or traveling zonally with realistic speeds, are superimposed on the initial velocity profile. Particular attention is placed on the Lagrangian structures associated with RWB. The Lagrangian analysis is based on the calculation of trajectories and the application of a diagnostic tool known as the ‘‘M’’ function. Hyperbolic trajectories (HTs), produced by the transverse intersections of stable and unstable invariant manifolds, may yield chaotic saddles in M. Previous studies associated HTs with ‘‘cat’s eyes’’ generated by planetary wave breaking at the critical levels. HTs, and hence RWB, are found both outside and inside the stratospheric polar vortex (SPV). Significant findings are as follows: (i) stationary forcing produces HTs only outside of the SPV and (ii) eastward-traveling wave forcing can produce HTs both outside and inside of the SPV. In either case, HTs appear at or near the critical latitudes. RWB was found to occur inside the SPV even when the forcing was located completely outside. In all cases, the westerly jet remained impermeable throughout the simulations. The results suggest that the HT inside the SPV observed by de la Cámara et al. during the southern spring 2005 was due to RWB of an eastward-traveling wave of wavenumber 1. 1. Introduction The southern stratosphere during the winter and spring seasons is characterized by a primarily westerly flow that defines the polar night vortex [or stratospheric polar vortex (SPV)], which gradually weakens with time (Mechoso et al. 1988). This westerly flow is perturbed by planetary-scale disturbances representing Rossby waves. Stratospheric flow features observed during these seasons are broadly consistent with the ideas of Charney and Drazin (1961) and Matsuno (1970), according to which only waves with very long wavelengths (typically zonal * Current affiliation: Department of Mechanical Engineering, Indian Institute of Technology Kanpur, Kanpur, Uttar Pradesh, India. Corresponding author address: Dr. Anirban Guha, Department of Mechanical Engineering, Indian Institute of Technology Kanpur, SL 109, Kanpur 208016 UP, India. E-mail: anirbanguha.ubc@gmail.com DOI: 10.1175/JAS-D-15-0088.1 Ó 2016 American Meteorological Society wavenumbers 1–3) can propagate upward from the troposphere, provided the mean flow is westerly and not too strong. Among the planetary-scale disturbances in the southern stratosphere, the quasi-stationary wave of zonal wavenumber 1 (S1 ) is the most apparent (Quintanar and Mechoso 1995). Another prominent wave feature, especially in October, is an eastward-propagating Rossby wave of zonal wavenumber 2 (T2 ) (Manney et al. 1991; Hio and Yoden 2004). The planetary-scale waves induce quasi-horizontal tracer advection, which is a trademark of the winter and spring seasons in the stratosphere. Rossby wave propagation on isentropic surfaces is associated with reversible deformation of potential vorticity (PV) contours. Rossby wave breaking (RWB), which occurs when the wave amplitude becomes sufficiently large, is associated with irreversible deformation. RWB in the ‘‘surfzone,’’ where material filaments are pulled out of the SPV edge and mixed with the exterior flow (McIntyre and Palmer 1983, 1984, 1985), sharpens the PV gradient surrounding the SPV, making the SPV 394 JOURNAL OF THE ATMOSPHERIC SCIENCES edge a barrier to horizontal transport of air parcels (Juckes and McIntyre 1987). The discovery of the Antarctic ozone hole, and the great interest in understanding its generation and evolution, motivated many researchers in the last few decades to study tracer advection in the stratosphere. Particular emphasis has been given on the permeability of the SPV. In the southern spring stratosphere, the transport barrier mentioned in the previous paragraph is robust in the presence of waves with slower phase speeds (Bowman 1996). The strong SPV retains its material identity, and the air in its interior remains chemically isolated from the rest of the stratosphere (Juckes and McIntyre 1987; Haynes 2005). Therefore, dynamical processes have important implications with respect to the formation and dissipation of the ‘‘Antarctic ozone hole’’ (Shepherd 2007; de la Cámara et al. 2012). Significant advancements have been made in the understanding of isentropic transport in stratospheric flows using Lagrangian methods along with the application of concepts from dynamical systems theory (Pierrehumbert 1991a,b; Bowman 1993; Ngan and Shepherd 1997, 1999; Mizuta and Yoden 2001; Koh and Legras 2002; Rypina et al. 2007; Beron-Vera et al. 2010). Equivalent studies have been performed in oceanography as well; see Mendoza and Mancho (2010) and Olascoaga et al. (2013). Recently, de la Cámara et al. (2012, 2013) examined the trajectories of superpressure balloons released during the springs of 2005 and 2010 from McMurdo, Antarctica, by the Vorcore and Concordiasi components of the Stratéole project [see Hertzog et al. (2007) and Rabier et al. (2010), respectively]. The balloons drifted in the lower stratosphere for several months, providing approximations to quasi-isentropic fluid trajectories. For analysis of balloon behavior, de la Cámara et al. (2012, 2013) applied a global Lagrangian descriptor, ‘‘M’’ (Jiménez Madrid and Mancho 2009; Mancho et al. 2013), to European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim) velocity data. Previous work by Mendoza and Mancho (2010) have shown that M displays stable and unstable manifolds from all possible hyperbolic trajectories (HTs, which are the traces of material hyperbolic points) in the neighborhood of a region, without the need to identifying these trajectories a priori. Using the descriptor M, de la Cámara et al. (2013) identified HTs in the flow field. These HTs were interpreted as evidence of Kelvin’s ‘‘cat’s eye’’ patterns (Stewartson 1977; Warn and Warn 1978). Cat’seyes form around critical layers, which develop in shear flows at locations where the wave phase speed matches the background velocity. These concepts are further discussed in section 3 below. VOLUME 73 Several numerical studies on RWB and associated stratospheric transport have been performed using shallow-water equations on a sphere using realistic initial and boundary conditions (Juckes 1989; Polvani et al. 1995; Ngan and Shepherd 1999; Rong and Waugh 2004). However, these studies concentrated on the processes and mechanisms at work outside the SPV (specifically, the surfzone dynamics). By contrast, only a few studies have addressed RWB inside the SPV, for example, see Mizuta and Yoden (2001) and Nakamura and Plumb (1994). These particular studies used either highly idealized velocity fields (Nakamura and Plumb 1994) or implemented a dynamically consistent model (Mizuta and Yoden 2001). Our objective in this paper is to systematically investigate RWB both outside and inside of the SPV. Our motivation for studying RWB inside the SPV is primarily derived from the recent work of de la Cámara et al. (2013), who found evidence of HT inside the SPV during the southern spring of 2005. De la Cámara et al. (2012) also discovered that ‘‘lobes’’ associated with HTs (see Koh and Plumb 2000) outside and inside of the SPV can intersect, thereby establishing routes of transport across the SPV edge and challenging its permeability. The authors conjectured that the HT observed inside the SPV is also due to RWB, but they did not determine the properties of the breaking wave (i.e., its wavenumber and frequency). In addition, de la Cámara et al. (2013) mentioned that associated transports may bring ozone-rich air from the vortex periphery to its inside. We undertake this investigation of RWB during southern winter/spring conditions in the simple dynamical framework provided by a single-layer shallow-water numerical model with a free surface on a sphere. Tessellation of the domain is done using geodesic (icosahedral) grids, which avoid the ‘‘pole problem’’ of longitude–latitude grids. The model’s initial conditions correspond to a realistic, zonally symmetric balanced flow. The bottom topography provides a perturbation with a single zonal wavenumber, for which different amplitude, phase speed, and meridional structures are considered. Depending on the background flow and phase speed, critical latitudes can exist both outside and inside the SPV. For analysis, we apply the Lagrangian descriptor M to the model’s results and search for HTs in the flow field. Finally, we associate the HTs to cat’s-eyes and hence to RWB. The outline of the paper is as follows. In section 2 we describe the shallow-water model and present the prescribed initial conditions and bottom topography. In section 3, we discuss the Lagrangian analysis performed through offline integration of particle trajectories. We also provide a brief description of the Lagrangian descriptor M and its implementation in our context. In JANUARY 2016 395 GUHA ET AL. section 4 we present the results of several simulations. The cases examined can be broadly divided into two categories: (i) stationary forcing and (ii) eastwardtraveling forcing. We perform a Lagrangian analysis on the velocity data obtained from each test case. In particular, we emphasize the contour plots of j=Mj, from which HTs clearly emerge at the critical latitudes, signifying RWB. Finally, conclusions are drawn in section 5, and implications of our study with respect to observed stratospheric features are discussed. 2. Shallow-water model a. Model equations An important technical difficulty in setting up a numerical study of the flow inside the polar vortex is the need to resolve motions near or over the pole. The use of spherical coordinates for discretization of the model’s equations leads to the standard pole problem (Heikes and Randall 1995a). This occurs because in a longitude– latitude-based coordinate system with constant resolution in latitude, the distance between adjacent grid points along a latitude circle continuously decreases toward the poles. Thus, a very small time step must be used in order to avoid linear computational instability. For the present study we selected the shallow-water system on the surface of a sphere described by Heikes and Randall (1995a). This model is free from the pole problem and provides quasi-homogeneous and quasiisotropic resolution over the entire sphere (Heikes and Randall 1995a,b; Heikes et al. 2013). In the following we give a brief model description that highlights recent model updates. The model’s prognostic variables are vorticity, (velocity) divergence, and fluid depth. The absolute vorticity za is predicted by ›za 5 2= (za y) , ›t (1) ^ = 3 y is relative vorticity, f is where za [ z 1 f , z [ k ^ is the the Coriolis parameter, t is time, y is velocity, and k vertical unit vector. The (velocity) divergence d [ = y is predicted by ›d 5 = (za =c) 1 J(za , x) 2 =2 [g(h 1 hb ) 1 K], ›t (2) ^ (=a 3 =b) is where c is the streamfunction; J(a, b) [ k the Jacobian operator; x is the velocity potential; =2 [ = = is the Laplacian operator; g is acceleration due to gravity; h is fluid depth; hb is the height of the bottom topography; and K is kinetic energy, defined as follows: FIG. 1. The icosahedral grid of the shallow-water model. For illustration purposes, we show G2 resolution yielding 162 grid cells and approximately 1908-km grid distance. Dots and ‘‘P’’ indicate cell centers and pentagons, respectively. Cell centers, walls, and corners are marked with thin arrows. Normal velocities for the walls of a hexagon are marked with thick arrows. 1 K [ [= (c=c) 2 c=2 c 1 = (x=x) 2 x=2 x] 1 J(c, x) . 2 (3) The fluid depth is predicted by ›h 5 2= (hy) . ›t (4) The streamfunction and velocity potential are obtained from the relative vorticity and divergence through solving the following elliptic equations: =2 c 5 z, =2 x 5 d. (5) Finally, the velocity is obtained from the streamfunction and velocity potential through ^ 3 =c 1 =x . y[k (6) b. Discretization of equations on an icosahedral grid The model equations are discretized on the hexagon– pentagon icosahedral grid sketched in Fig. 1. The grid generation, optimization, and performance of finitedifference operators are discussed by Heikes et al. (2013). In the integrations performed for this study, we use the tweaked G6 grid with 40 962 grid cells, yielding approximately 120-km grid distance. Regardless of resolution, 12 of the cells are pentagons. Note that the icosahedral grid system used here is not twisted. The geographic poles are placed in diametrically opposite pentagons— for example, the one marked with ‘‘P’’ in 396 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 73 Fig. 1. Following the Z-grid convention introduced by Randall (1994), the vorticity, divergence, and fluid height are predicted at the cell centers (marked with dots in Fig. 1). An important advantage of the Z grid over the A and C grids is the exclusion of computational modes. The model uses the multigrid solver described in Heikes et al. (2013) to obtain the streamfunction and velocity potential at the cell centers. The only velocity y defined in the discrete system is the one normal to the cell walls pointing from one cell center to the next one as shown for a selected hexagon in Fig. 1. To obtain this velocity, the streamfunction is interpolated to the cell corners from cell centers using a quadratic interpolation technique requiring information from six surrounding cell centers. The absolute vorticity flux za y and mass flux hy are computed at the cell walls using a third-order upstream finite-difference scheme. c. Initial conditions The initial wind profile is zonally symmetric and is prescribed as follows: f 1 C2 u(f) 5 C1 cos(f) sech C3 C 2f . 2 C4 cos(f) sech 5 C6 (7) Here C1 5 118 m s 21 , C2 5 1:09 rad, C3 5 0:185, C4 5 20 m s21, C5 5 0:5 rad, C6 5 0:7, and f is latitude (rad). This velocity profile yields a maximum jet speed of umax 5 55:4 m s21 (at approximately 608S), and a zerowind line at 288S; see Fig. 2a. The velocity profile represented by Eq. (7) is similar to the one used by Ngan and Shepherd (1999); however, the constants are suitably chosen so as to mimic the zonal-mean zonal wind distribution at the 475-K isentropic surface on 17 September 2005. The choice is representative of the zonal flow in the lower stratosphere during the Vorcore campaign. The initial height field is obtained by using the gradient wind balance equation, h(f) 5 h0 2 ðf Ru(f0 ) tan(f0 ) f1 u(f0 ) df0 , (8) g R where h0 5 5985 m is the free-surface height at the South Pole and R is Earth’s radius. Following Polvani et al. (1995) the average shallow-water height is chosen to be 8 km. Physically this height represents the vertical scale of the Rossby waves that would be produced when the flow is perturbed (hence this height does not necessarily correspond to the actual height of the 475-K isentropic surface). Figure 2b shows the variation of h with latitude. FIG. 2. (a) Initial zonal wind profile. The gray dashed line corresponds to the zonal-mean zonal wind distribution on 17 Sep 2005 at the 475-K isentropic surface. The thick solid black line, given by Eq. (7), provides a simple analytical representation of the real profile. Thin solid black lines are Rossby wave phase speeds [computed from Eq. (10)] for different simulation test cases; see Table 1. The intersection of thin and thick solid lines occurs at critical latitudes, marked by open circles. The dashed–dotted gray line below represents the latitudinal variation of the topography given by Eq. (9). (b) The solid line represents the latitudinal variation of the initial shallow-water height, which is obtained from Eq. (8). The dashed line is the mean shallow-water height. d. Topography The height of the bottom topography is prescribed by # " # f 2 f0 2 cos(f) exp 2 hb 5 2B cos(f0 ) Df t cos(kl 2 vt) , 3 1 2 exp 2 tS " (9) where B 5 392 m (it corresponds to a 400-m-high mountain), f0 5 2p/3 (unless mentioned otherwise), Df 5 p/18, tS 5 10 days, l is longitude (rad), k is a nondimensional zonal wavenumber, and v is frequency (day21). The value of B for the given shallow-water height is chosen such that the forcing amplitude is relatively weak. Stronger forcing simply accelerates the JANUARY 2016 397 GUHA ET AL. TABLE 1. Simulation test cases. The values of fcrit are predicted from linear theory. Case Label k f0 (8) T (days) finside (8) crit foutside (8) crit Stationary wave 1 Stationary wave 2 Stationary wave 3 Traveling wave 1 Traveling wave 2 Traveling wave 3 Traveling wave 2 (fast) Traveling wave 2 (fast, different f0 ) S1 S2 S3 T1 (7:72) T2 (7:72) T3 (7:72) T2 (3:86) T^2 (3:86) 1 2 3 1 2 3 2 2 260 260 260 260 260 260 260 245 ‘ ‘ ‘ 7.72 7.72 7.72 3.86 3.86 — — — 275 283 287 275 275 228 228 228 250 243 239 250 250 evolution of the events. Our preference is to stay close to linearity and delay the appearance of chaotic flow. The phase speed of Rossby waves induced by this topography is given as follows: c(f) 5 vR cos(f) . k (10) Our choices of v and k are given in Table 1. We select k 5 1, 2, and 3 because these are the zonal wavenumbers typically observed in the winter/springtime stratosphere. The point(s) of intersection of the u(f) profile given by Eq. (7) and the c(f) profiles given by Eq. (10) for the selected values of v and k, respectively, correspond to the critical latitudes; see Fig. 2a. There is only one critical latitude when the wave is stationary, and two critical latitudes when the wave is eastward traveling at a speed of 0 , c , umax . Therefore, we expect RWB outside of the SPV in the stationary forcing cases, and both outside and inside of the SPV when the wave is traveling. The simulation cases listed in Table 1 are integrated for 60 days, except for one (S1 ), which is integrated for 70 days. In all cases the model’s time step is 1 min and the output is saved every 1 h. 3. Lagrangian analysis The trajectory of the tracer parcels are calculated as follows: dx(t) 5 y(x, t) , dt (11) where the velocity field is provided by the output of the shallow-water model, which is available hourly. For these calculations, the velocity data are interpolated to parcel locations using a bicubic spline interpolation scheme. Parcels march forward in time using a fourthorder Runge–Kutta scheme, with a time step of 1 h. The results obtained are largely insensitive to halving the time step. The Lagrangian descriptor M is defined as follows (Jiménez Madrid and Mancho 2009): Mt (x0 , t0 ) 5 ð t0 1t t0 2t jy[x(t), t]j dt . (12) Here x0 5 x(t0 ) is the location of a parcel at t 5 t0 . Thus, the values of M correspond to the length of the trajectories in the phase space (which, in this case, is same as the physical space) passing through x0 at t 5 t0 over the time interval [t0 2 t, t0 1 t]. Therefore, calculation of M implies integration of backward trajectories from t0 2 t to t0 and forward trajectories from t0 to t0 1 t. The function M, as defined in Eq. (12), provides a global dynamical picture of any arbitrary time-dependent flow. As noted in Mancho et al. (2013), convergence of the structure of M toward the stable and unstable manifolds requires sufficiently large values of t. We have chosen t 5 15 days for stationary forcing cases and t 5 20 days when forced via traveling waves. It is important to note here that M depends on t. For instance, for low t the appearance of M is rather simple, almost without structure and resembling that of Eulerian currents. For larger t, the structure of M becomes more and more refined and more details of the invariant manifolds are observed. This is similar to what is observed in Lyapunov exponents or in the direct calculation of the manifolds: the longer the integration period, the more detailed and complex is the structure of the manifold (Mendoza and Mancho 2010, 2012; Mancho et al. 2013; Lopesino et al. 2015). The chaotic saddles in M portray the transverse intersections of stable and unstable invariant manifolds, which provide the evidence of HTs (see Fig. 2 of de la Cámara et al. 2013). Proving the existence of chaos in the current setting is beyond the scope of this article. However. we have noted similarities between the expected evolution of trajectories within the domains bounded by transverse intersections (in our numerical experiments) and those of the chaotic invariant set obtained from the ‘‘Smale horseshoe map’’ [see Mendoza and Mancho (2010), Mancho et al. (2013), and Mendoza and Mancho (2012) for details on a rigorous proof for the existence of chaos on autonomous and nonautonomous maps]. These analogies suggest the presence 398 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 73 FIG. 3. PV contours for (a) S1 on day 50, (b) S2 on the day 25, and (c) S3 on day 30. of chaotic advection, which occurs in temporally aperiodic systems when well-defined heteroclinic structures associated with aperiodic HTs exist in the flow field. Ngan and Shepherd (1999) proposed a necessary condition for the existence of chaotic advection in dynamical flows—there should be an ‘‘organizing structure’’ (which includes hyperbolic points) in the flow field and that should be robust. These structures are also known as Lagrangian coherent structures (LCS) in the literature— for example, see Beron-Vera et al. (2010) and references therein. De la Cámara et al. (2013) suggested that chaotic saddles in M, and hence HTs, are representative of cat’seye LCSs associated with planetary wave breaking at the critical levels. The flow structures referred to as cat’seyes form around ‘‘critical layers,’’ which develop in shear flows at locations where the wave phase speed matches the background velocity. In a reference frame traveling with the Rossby wave phase speed, the flow near a critical layer exhibits Kelvin’s cat’s-eye patterns (Stewartson 1977; Warn and Warn 1978). Hyperbolic points are at the locations where the cats’ eyelids meet. Perturbation of the cat’s-eyes results in irreversible deformation of material contours, signifying RWB. This can lead to chaotic advection of tracers, which can be understood using Hamiltonian chaos theory (Pierrehumbert 1991a,b; Wiggins 1992; del-Castillo-Negrete and Morrison 1993; Samelson and Wiggins 2006; Rypina et al. 2007). 4. Results and discussion a. Critical layer outside the SPV We start by considering stationary forcing cases with k 5 1, 2, and 3. These cases are referred to as S1 , S2 , and S3 , respectively; see Table 1. According to the predictions of linear theory, stationary forcing results in the formation of a critical layer at the zero-wind line. For the background zonal flow prescribed in Eq. (7), the zerowind line is located in the subtropics at 288S (see Fig. 2a) and is therefore far outside the SPV (note that the SPV extends up to ’608S). Figure 3 shows contour plots of PV (equal to za /h) for each of the three cases considered. Each panel in Fig. 3 displays anticyclonic features around the critical layer (the surfzone), signifying RWB. Initially, the number of anticyclones (or cat’s-eyes) is equal to the k corresponding to the forcing. As the simulation progresses, secondary cat’s-eyes may appear, for example, see Fig. 3a. Another outstanding feature is the deformation of the SPV, which is most prominent for the S3 case. The time evolution of the tracer field corresponding to each k is shown in Fig. 4. Although significant latitudinal transport is observed in the surfzone, there is virtually no such transport across the SPV boundary. This observation is in agreement with a large body of literature that examined the impermeability of the SPV boundary under stationary (especially k 5 1) forcing, for example, Juckes (1989), Polvani et al. (1995), and Ngan and Shepherd (1999). RWB and the appearance of a complex surfzone occur earlier in time as k is increased (note that the forcing amplitude B has been held constant in all cases). Merging of secondary and primary cat’s-eyes is also observed—see day 31 and day 40 plots for the S2 case in Fig. 4b. We next apply the Lagrangian descriptor M discussed in section 3 to visualize the invariant manifolds and locate the HTs. In the following, we have opted to plot j=Mj instead of M as in the previous papers. The term =M approximately aligns with the manifolds, and its magnitude provides direct information on the regions where M has strong horizontal variations. In this context, HTs are captured by the intersection of contour lines where j=Mj has local maxima. Figure 5 shows JANUARY 2016 GUHA ET AL. 399 FIG. 4. Tracer evolution for (a) S1 , (b) S2 , and (c) S3 . Color contours represent the initial latitudinal position of tracers. The day corresponding to each plot appears to its left. contours of j=Mj for S1 , S2 , and S3 . The HTs in these cases are labeled ‘‘H’’ in the different panels of the figure. As expected, these HTs are close to the zero-wind lines, and their locations are practically independent of time for a few days. Noting that our flow is fully nonlinear, deviations from stationary conditions in space predicted by linear theory can be expected. The latitude f at which HTs appear correspond to c ’ u, and the number of HTs corresponds to the zonal wavenumber of the perturbation k. Hence, the frequency v can be calculated from Eq. (10). Secondary cat’s-eyes may form during the later stages of integration, producing additional or secondary HTs. These are labeled ‘‘H’’ in gray in Figs. 5a and 5b. 400 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 73 FIG. 5. The j=Mj contours for (a) S1 with t0 5 day 50, (b) S2 with t0 5 day 25, and (c) S3 with t0 5 day 30. The black ‘‘H’’ denotes HTs from the primary cat’s-eyes, while the gray one denotes HTs from the secondary cat’s-eyes. Primary HTs are related to stronger gradients in M; hence, they are defined by darker shades. b. Critical layers both outside and inside the SPV To examine these cases, the model is forced with eastward-traveling Rossby waves by setting v . 0 in Eq. (9). Similar to the stationary case, we take k 5 1, 2, and 3 but choose v corresponding to a period of 7.72 days. The justification for this choice is that waves with zonal wavenumbers 1–3 and a period of 7–8 days have been observed in the Southern Hemisphere during winter/ springtime (Hio and Yoden 2004). We refer to these cases as T1 (7:72), T2 (7:72), and T3 (7:72); see Table 1. According to Fig. 2a, the c profile for each of these cases intersects the u velocity profile at two locations. Hence, there are two critical layers, one outside and one inside the SPV. The outer critical layer is located in the midlatitudes, as shown in Fig. 6. The signature of the inner critical layer is the deformation of PV contours, which is prominent in Figs. 6a and 6b but not in Fig. 6c. We find that the critical layer(s) for traveling disturbances appear(s) to be weaker/thinner than that of stationary disturbances. The narrow width of the cat’s-eyes for traveling waves is a direct consequence of the nature of the streamfunction created by superimposing a wave on a background shear. The width of a cat’s-eye in the steady case is proportional to the amplitude of the wave and the value of the shear. In the unsteady case, the width of the chaotic layer will depend on the size of the unperturbed cat’s-eye, as well as on the amplitude and nature of the transient parts of the flow. These findings are in agreement with the observations of Ngan and Shepherd (1999). From the tracer plots shown in Fig. 7, it is difficult to discern whether the clearly seen flow structures represent cat’s-eyes. For proper visualization, one should FIG. 6. PV contours for (a) T1 (7:72) on day 30, (b) T2 (7:72) on day 30, and (c) T3 (7:72) on day 25. JANUARY 2016 GUHA ET AL. 401 FIG. 7. Tracer evolution for (a) T1 (7:72), (b) T2 (7:72), and (c) T3 (7:72). Plot properties are similar to that of Fig. 4. work in a comoving reference frame in which the wave appears to be stationary. Structures inside the SPV are not visible in the T3 case as shown in both Figs. 6c and 7c. One possible reason for this absence is that the critical latitude in this case is inside the SPV finside crit , which is at about 878S. Therefore, the effect of topographic forcing is considerably (which is centered near 608S) at finside crit weak, and it does not lead to RWB. Another possibility is that the model’s horizontal resolution is insufficient at these high latitudes. Similar to the stationary forcing case, we apply j=Mj to visualize invariant manifolds, especially HTs. The presence of HTs will indicate whether the observed structures are indeed LCSs, and therefore signify RWB. In Fig. 8a, HTs clearly emerge both outside and inside the SPV. In other words, the ‘‘8-day wave’’ of zonal 402 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 73 FIG. 8. The j=Mj contours for (a) T1 (7:72) with t0 5 day 30, (b) T2 (7:72) with t0 5 day 30, and (c) T3 (7:72) with t0 5 day 25. wavenumber 1 breaks both inside and outside of the SPV. Unlike the T1 case, HTs are not visible inside in the other two cases; see Figs. 8b and 8c. It is important to note here that structures are observed in the PV field for the T2 case. The critical latitude in both T2 and T3 are close to the South Pole (see Table 1). For T2 case we believe those structures to be LCSs but were unable to obtain HTs. We expect that a higher-resolution simulation and a larger value of t might reveal HTs in this case. However it is unlikely for the T3 case to have HTs inside the SPV. To visualize HTs for a higher wavenumber, we performed another simulation with k 5 2 and a period of 3.86 days. This case is referred to as T2 (3:86) in Table 1, to occur at 758S. Physand linear theory predicts finside crit ically, T2 (3:86) corresponds to the ‘‘4-day wave’’ seen in the southern stratosphere during wintertime (Mizuta and Yoden 2001). Figure 9a reveals four HTs, two outside and two inside the SPV. We also carried out a model integration with the mountain (nearly) centered at 458S. This case is referred to as T^2 (3:86) in Table 1. Although the mountain in this case is (almost) completely outside the SPV, the forcing is strong enough to deform its periphery and trigger Rossby waves inside the vortex. Evidence of HTs in the T^2 (3:86) case is given in Fig. 9b. We find that there are four primary HTs, of which two are outside and two are FIG. 9. The j=Mj contours on t0 5 day 25 for the cases (a) T2 (3:86) and (b) T^2 (3:86). The black ‘‘H’’ denotes HTs from the primary cat’s eyes, while the gray one denotes HTs from the secondary cat’s-eyes. JANUARY 2016 GUHA ET AL. FIG. 10. HTs outside and inside the SPV for the case T1 (7:72). Successive markers are 1 day apart. inside the SPV. Although Figs. 9a and 9b have been plotted for the same t0 , the outcome in each case is substantially different from the other. There are also two secondary HTs (labeled with gray ‘‘H’’) outside for the T^2 (3:86) case, as indicated by the grayscale shading in the figure. Since the peak amplitude of the forcing is at , secondary cat’s-eyes are easily 458S and close to foutside crit formed. On the other hand, the inner critical layer is far from the location predicted by linear theory. This is probably due to the nonlinearity in the background flow, which makes the velocity field less zonal with time, especially at high latitudes. Figures 8a and 9a resemble Fig. 10 of Mizuta and Yoden (2001). Note that the procedure for visualizing HTs described by these authors requires a corotating frame of reference in which the wave appears to be stationary. Although this is a standard technique for visualizing the breaking of a single wave in a Cartesian system [e.g., see Samelson and Wiggins (2006) and references therein], implementing it in a realistic scenario is challenging. It is unclear at which phase speed the frame should rotate when multiple waves are present. Also, unlike the b-plane approximation, phase speed in the spherical coordinate system is a function of latitude according to Eq. (10). The function M, by contrast, is frame independent and does not require a priori knowledge of the underlying dynamical system. HTs can be directly used to obtain the phase speed and frequency of the breaking Rossby waves. As an example, we have shown the HTs of T1 (7:72) in Fig. 10. The figure shows that during the period investigated, the outer HT is located very close to 508S, while the inner HT is located near 758S. These two latitudes are foutside crit and finside crit , respectively, predicted by the linear critical layer theory (see Table 1), which gives an excellent prediction in this case. Two successive markers along each HT are 1 day apart; hence, phase speed can be easily evaluated. It is important to note here that T1 (7:72) is an ideal case in which the phase speed can 403 also be well approximated from the background zonal flow (c 5 u at the critical layer, where u is the zonalmean zonal wind). In realistic flows, neither the background flow nor the Rossby wave phase speed is strictly zonal. Even in our idealized simulations, we found a large departure from the linear theory predictions for RWB inside the SPV; see case T^2 (3:86). Therefore, HTs captured using M provide an accurate, frame-independent way for evaluating the phase speed of a breaking Rossby wave. Once the phase speed is known, the frequency can be obtained from Eq. (10).1 We confirmed that the wave period obtained from each HT is approximately 7.72 days. Although forcing via traveling waves can eventually produce chaotic regions both outside and inside of the SPV, in our simulations these regions are always separated by the westerly jet. The jet acts like an effective barrier to lateral transport and remains impermeable for the entire length of our simulations. Similar behavior was also observed by Rypina et al. (2007), who studied the permeability of a Bickley jet. The fact that the jet is a strong transport barrier was explained by Kolmogorov– Arnold–Moser (KAM) theory—the jet is composed of an invariant KAM tori (Rypina et al. 2007; Beron-Vera et al. 2010). We are currently performing several analyses in order to gain insight into these issues, which may require a more complex framework for research. 5. Summary and conclusions Recently de la Cámara et al. (2013), on the basis of results obtained with reanalysis products for the southern winter/spring of 2005, found HTs both outside and inside the SPV (see Fig. 2 in their paper). On the basis of linear critical layer theory, they argued that such flow structures are indicative of RWB. The present paper exploits this argument in an idealized framework, in which a circumpolar vortex with a realistic velocity profile was forced by perturbations with a single zonal wavenumber. Our methodology was based on a spherical single-layer shallow-water model with free surface in a geodesic (icosahedral) grid system. Numerical instabilities at the poles (known as the ‘‘pole problem’’) do not occur when such a grid system (see Fig. 1) is used. The initial velocity profile, given by Eq. (7), approximated the zonal flow in the lower stratosphere on a given day in the southern spring of 2005. The 1 Secondary cat’s-eyes, if present, can produce a higher number of HTs. In a real scenario, k is unknown,and is found by counting HTs. This may lead to an erroneous determination of v, which can be avoided by performing two checks: (i) the value of c is close to u and (ii) the HTs are related to the intersection of darker j=Mj contours. 404 JOURNAL OF THE ATMOSPHERIC SCIENCES perturbations were provided by a bottom topography consisting of a single wavenumber in the zonal direction and a (nearly) Gaussian profile in the meridional direction [see Eq. (9)]. The perturbations at the model’s lower boundary had the longest planetary waves (k 5 1–3), which could be either stationary or eastward traveling. With this system we performed a systematic series of numerical experiments (see Table 1). First we focused on cases in which the forcing was stationary. In all of these cases, critical layers develop in the tropics at around 288S—that is, far outside the SPV edge that is approximately at 608S. Three outstanding features were observed: (i) RWB around the critical layer leads to the formation of ‘‘cat’s eyes’’, which are LCSs signifying chaotic advection (see Fig. 4). (ii) Initially the number of HTs corresponded to the zonal wavenumber with which the system was forced. At later times, however, secondary cat’s-eyes appeared in the flow, which led to the formation of additional HTs. The j=Mj plots in Fig. 5 clearly depicted all these features. (iii) Forcing could distort the SPV, and the distortion bore the signature of the forcing wavenumber. This was particularly true whenever k . 2. For example, a triangularshaped SPV was observed in the S3 case; see Fig. 3c. Next, we turned to eastward-traveling cases. Unlike the stationary forcing cases, two critical layers were formed, one outside the SPV (in the midlatitudes) and the other inside of it; see Fig. 2a. Based on observation data, we chose the forcing frequency to correspond to a period of approximately 8 days. These cases were given by T1(7.72), T2(7.72), and T3(7.72) in Table 1. The significant findings in this regard are as follows: (i) Typical cat’s-eye structures were not detected in either PV plots (Fig. 6) or tracer plots (Fig. 7). This is because appropriate visualization of such structures requires the observer to be in a comoving reference frame where the breaking wave appears to be stationary. Nevertheless, structures were seen outside, and in some cases, inside the SPV as well. Lagrangian analysis was performed to search for HTs in order to confirm that these structures were indeed cat’s-eye LCSs. (ii) Plots of j=Mj in Fig. 8 showed HTs outside the SPV in all cases but inside the SPV only for T1 (7:72). Thus, we concluded that the 8-day wave with k 5 1 breaks both outside and inside of the SPV. (iii) In the T3 (7:72) case, RWB did not occur inside the SPV because the location of the critical latitude was very close to the South Pole, where wave amplitude is very small and model resolution may not suffice. Relatively coarse resolution and smaller length of integration t probably contributed to the difficulties in not observing HTs for the T2 (7:72) case. We made additional numerical experiments with traveling waves by either changing the wave period or VOLUME 73 the location of the peak of the bottom topography. Our findings are as follows: (iv) By forcing with a (nearly) 4-day wave of k 5 2 [case T2 (3:86) in Table 1], we were able to observe HTs both outside and inside of the SPV; see Fig. 9a. (v) By making f0 5 458S in Eq. (9), the location of the mountain was made outside of the SPV. Without changing the mountain amplitude, we were able to generate HTs both outside and inside of the SPV as shown in Fig. 9b. Thus, a traveling perturbation localized outside the SPV could still produce RWB inside of it. In both stationary and traveling forcing cases, the Lagrangian descriptor M was found to be a powerful diagnostic tool for RWB. In fact, wave properties like frequency, wavenumber, and phase speed could be evaluated from the HTs; see Fig. 10. In all simulations, we found the westerly jet to be an impermeable transport barrier. Even though RWB and chaotic transport might occur on both sides of the jet, strong KAM stability makes it impermeable. This result is in agreement with the findings of Rypina et al. (2007). Transport across the jet, as observed by de la Cámara et al. (2013), might be the result of more complex events, which need to be investigated in the future. Finally, we conclude that the HT inside the SPV observed during early spring of 2005 by de la Cámara et al. (2013) was due to the breaking of an eastward-traveling Rossby wave of the zonal wavenumber 1. A more detailed justification for this assertion will be given in future work. Acknowledgments. We are grateful to Alvaro de la Cámara and Ana Mancho for the useful discussions, and their generous help regarding the ‘‘M’’ function. We also thank the anonymous referees for their helpful suggestions and constructive comments. Anirban Guha and Carlos R. Mechoso were supported by the U.S. NSF Grant AGS-1245069. Anirban Guha is also grateful to the initiation grant provided by IIT Kanpur (IITK/ME/2014338) for partial support of this research work. Celal S. Konor and Ross P. Heikes were partially funded by the U.S. DOE under Cooperative Agreement DE-FC02-06ER64302 to Colorado State University and the DOE under Grant DE-SC07050. Also, Celal S. Konor was partially funded by the NSF Grant AGS-1062468. REFERENCES Beron-Vera, F. J., M. J. Olascoaga, M. G. Brown, H. Koçak, and I. I. Rypina, 2010: Invariant-tori-like Lagrangian coherent structures in geophysical flows. Chaos, 20, 017514, doi:10.1063/ 1.3271342. JANUARY 2016 GUHA ET AL. Bowman, K. P., 1993: Barotropic simulation of large-scale mixing in the Antarctic polar vortex. J. Atmos. Sci., 50, 2901–2914, doi:10.1175/1520-0469(1993)050,2901:BSOLSM.2.0.CO;2. ——, 1996: Rossby wave phase speeds and mixing barriers in the stratosphere. Part I: Observations. J. Atmos. Sci., 53, 905–916, doi:10.1175/1520-0469(1996)053,0905:RWPSAM.2.0.CO;2. Charney, J. G., and P. Drazin, 1961: Propagation of planetary-scale disturbances from the lower into the upper atmosphere. J. Geophys. Res., 66, 83–109, doi:10.1029/JZ066i001p00083. de la Cámara, A., A. M. Mancho, K. Ide, E. Serrano, and C. R. Mechoso, 2012: Routes of transport across the Antarctic polar vortex in the southern spring. J. Atmos. Sci., 69, 741–752, doi:10.1175/JAS-D-11-0142.1. ——, C. R. Mechoso, A. M. Mancho, E. Serrano, and K. Ide, 2013: Isentropic transport within the Antarctic polar-night vortex: Rossby wave breaking evidence and Lagrangian structures. J. Atmos. Sci., 70, 2982–3001, doi:10.1175/JAS-D-12-0274.1. del-Castillo-Negrete, D., and P. Morrison, 1993: Chaotic transport by Rossby waves in shear flow. Phys. Fluids, 5A, 948–965, doi:10.1063/1.858639. Haynes, P., 2005: Stratospheric dynamics. Annu. Rev. Fluid Mech., 37, 263–293, doi:10.1146/annurev.fluid.37.061903.175710. Heikes, R. P., and D. A. Randall, 1995a: Numerical integration of the shallow-water equations on a twisted icosahedral grid. Part I: Basic design and results of tests. Mon. Wea. Rev., 123, 1862–1880, doi:10.1175/1520-0493(1995)123,1862: NIOTSW.2.0.CO;2. ——, and ——, 1995b: Numerical integration of the shallow-water equations on a twisted icosahedral grid. Part II: A detailed description of the grid and an analysis of numerical accuracy. Mon. Wea. Rev., 123, 1881–1887, doi:10.1175/1520-0493(1995)123,1881: NIOTSW.2.0.CO;2. ——, ——, and C. S. Konor, 2013: Optimized icosahedral grids: Performance of finite-difference operators and multigrid solver. Mon. Wea. Rev., 141, 4450–4469, doi:10.1175/MWR-D-12-00236.1. Hertzog, A., and Coauthors, 2007: Stratéole/vorcore—Longduration, superpressure balloons to study the Antarctic lower stratosphere during the 2005 winter. J. Atmos. Oceanic Technol., 24, 2048–2061, doi:10.1175/2007JTECHA948.1. Hio, Y., and S. Yoden, 2004: Quasi-periodic variations of the polar vortex in the Southern Hemisphere stratosphere due to wave– wave interaction. J. Atmos. Sci., 61, 2510–2527, doi:10.1175/ JAS3257.1. Jiménez Madrid, J. A., and A. M. Mancho, 2009: Distinguished trajectories in time dependent vector fields. Chaos, 19, 013111, doi:10.1063/1.3056050. Juckes, M. N., 1989: A shallow water model of the winter stratosphere. J. Atmos. Sci., 46, 2934–2956, doi:10.1175/ 1520-0469(1989)046,2934:ASWMOT.2.0.CO;2. ——, and M. E. McIntyre, 1987: A high-resolution one-layer model of breaking planetary waves in the stratosphere. Nature, 328, 590–596, doi:10.1038/328590a0. Koh, T.-Y., and R. A. Plumb, 2000: Lobe dynamics applied to barotropic Rossby-wave breaking. Phys. Fluids, 12, 1518– 1528, doi:10.1063/1.870400. ——, and B. Legras, 2002: Hyperbolic lines and the stratospheric polar vortex. Chaos, 12, 382–394, doi:10.1063/1.1480442. Lopesino, C., F. Balibrea, S. Wiggins, and A. M. Mancho, 2015: Lagrangian descriptors for two dimensional, area preserving, autonomous and nonautonomous maps. Commun. Nonlinear Sci. Numer. Simul., 27, 40–51, doi:10.1016/j.cnsns.2015.02.022. Mancho, A. M., S. Wiggins, J. Curbelo, and C. Mendoza, 2013: Lagrangian descriptors: A method for revealing phase space 405 structures of general time dependent dynamical systems. Commun. Nonlinear Sci. Numer. Simul., 18, 3530–3557, doi:10.1016/j.cnsns.2013.05.002. Manney, G., J. Farrara, and C. Mechoso, 1991: The behavior of wave 2 in the Southern Hemisphere stratosphere during late winter and early spring. J. Atmos. Sci., 48, 976–998, doi:10.1175/ 1520-0469(1991)048,0976:TBOWIT.2.0.CO;2. Matsuno, T., 1970: Vertical propagation of stationary planetary waves in the winter Northern Hemisphere. J. Atmos. Sci., 27, 871–883, doi:10.1175/1520-0469(1970)027,0871: VPOSPW.2.0.CO;2. McIntyre, M. E., and T. N. Palmer, 1983: Breaking planetary waves in the stratosphere. Nature, 305, 593–600, doi:10.1038/305593a0. ——, and ——, 1984: The ‘surf zone’ in the stratosphere. J. Atmos. Terr. Phys., 46, 825–849, doi:10.1016/0021-9169(84)90063-1. ——, and ——, 1985: A note on the general concept of wave breaking for Rossby and gravity waves. Pure Appl. Geophys., 123, 964–975, doi:10.1007/BF00876984. Mechoso, C. R., A. O’Neill, V. D. Pope, and J. D. Farrara, 1988: A study of the stratospheric final warming of 1982 in the southern hemisphere. Quart. J. Roy. Meteor. Soc., 114, 1365–1384, doi:10.1002/qj.49711448402. Mendoza, C., and A. M. Mancho, 2010: Hidden geometry of ocean flows. Phys. Rev. Lett., 105, 038501, doi:10.1103/ PhysRevLett.105.038501. ——, and ——, 2012: Review article: ‘‘The Lagrangian description of aperiodic flows: A case study of the Kuroshio Current,’’ Nonlinear Processes Geophys., 19, 449–472, doi:10.5194/ npg-19-449-2012. Mizuta, R., and S. Yoden, 2001: Chaotic mixing and transport barriers in an idealized stratospheric polar vortex. J. Atmos. Sci., 58, 2616–2629, doi:10.1175/1520-0469(2001)058,2616: CMATBI.2.0.CO;2. Nakamura, M., and R. A. Plumb, 1994: The effects of flow asymmetry on the direction of Rossby wave breaking. J. Atmos. Sci., 51, 2031–2045, doi:10.1175/1520-0469(1994)051,2031: TEOFAO.2.0.CO;2. Ngan, K., and T. G. Shepherd, 1997: Chaotic mixing and transport in Rossby-wave critical layers. J. Fluid Mech., 334, 315–351, doi:10.1017/S0022112096004363. ——, and ——, 1999: A closer look at chaotic advection in the stratosphere. Part I: Geometric structure. J. Atmos. Sci., 56, 4134– 4152, doi:10.1175/1520-0469(1999)056,4134:ACLACA.2.0.CO;2. Olascoaga, M., and Coauthors, 2013: Drifter motion in the Gulf of Mexico constrained by altimetric Lagrangian coherent structures. Geophys. Res. Lett., 40, 6171–6175, doi:10.1002/2013GL058624. Pierrehumbert, R. T., 1991a: Chaotic mixing of tracer and vorticity by modulated traveling Rossby waves. Geophys. Astrophys. Fluid Dyn., 58, 285–319, doi:10.1080/03091929108227343. ——, 1991b: Large-scale horizontal mixing in planetary atmospheres. Phys. Fluids, 3A, 1250–1260, doi:10.1063/1.858053. Polvani, L. M., D. W. Waugh, and R. A. Plumb, 1995: On the subtropical edge of the stratospheric surf zone. J. Atmos. Sci., 52, 1288–1309, doi:10.1175/1520-0469(1995)052,1288: OTSEOT.2.0.CO;2. Quintanar, A. I., and C. R. Mechoso, 1995: Quasi-stationary waves in the Southern Hemisphere. Part I: Observational data. J. Climate, 8, 2659–2672, doi:10.1175/1520-0442(1995)008,2659: QSWITS.2.0.CO;2. Rabier, F., and Coauthors, 2010: The Concordiasi project in Antarctica. Bull. Amer. Meteor. Soc., 91, 69–86, doi:10.1175/2009BAMS2764.1. Randall, D. A., 1994: Geostrophic adjustment and the finitedifference shallow-water equations. Mon. Wea. Rev., 406 JOURNAL OF THE ATMOSPHERIC SCIENCES 122, 1371–1377, doi:10.1175/1520-0493(1994)122,1371: GAATFD.2.0.CO;2. Rong, P.-P., and D. W. Waugh, 2004: Vacillations in a shallowwater model of the stratosphere. J. Atmos. Sci., 61, 1174–1185, doi:10.1175/1520-0469(2004)061,1174:VIASMO.2.0.CO;2. Rypina, I., M. Brown, F. Beron-Vera, H. Koçak, M. Olascoaga, and I. Udovydchenkov, 2007: On the Lagrangian dynamics of atmospheric zonal jets and the permeability of the stratospheric polar vortex. J. Atmos. Sci., 64, 3595–3610, doi:10.1175/JAS4036.1. Samelson, R. M., and S. Wiggins, 2006: Lagrangian Transport in Geophysical Jets and Waves: The Dynamical Systems VOLUME 73 Approach. Interdisciplinary Applied Mathematics, Vol. 31, Springer, 150 pp. Shepherd, T. G., 2007: Transport in the middle atmosphere. J. Meteor. Soc. Japan, 85B, 165–191, doi:10.2151/jmsj.85B.165. Stewartson, K., 1977: The evolution of the critical layer of a Rossby wave. Geophys. Astrophys. Fluid Dyn., 9, 185–200, doi:10.1080/ 03091927708242326. Warn, T., and H. Warn, 1978: The evolution of a nonlinear critical level. Stud. Appl. Math., 59, 37–71, doi:10.1002/sapm197859137. Wiggins, S., 1992: Interdisciplinary Applied Mathematics. Vol. 2, Chaotic Transport in Dynamical Systems, Springer, 301 pp.