Shear-Driven Instabilities as the Origin of Multi-Banded Cloud and Precipitation Structures in an Extratropical Cyclone

Generated using the official AMS L A T EX template v6.1 Shear-Driven Instabilities as the Origin of Multi-Banded Cloud and1 Precipitation Structures in an Extratropical Cyclone2 Stephen R. Guimonda 3 aDepartment of Atmospheric and Planetary Sciences and Severe Weather Research Center, Hampton University, Hampton, VA, USA 4 5 Corresponding author: Stephen R. Guimond, [email protected]6 1 ABSTRACT: This paper investigates the dynamics governing multi-banded cloud and precipitation in extratropical cyclones through a case study from the NASA Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS) field campaign. On 1 February 2020, a low-pressure system emerged off the North Carolina coast at 1200 UTC, deepening by 7 hPa in six hours as it accelerated northeast over the Atlantic. High-resolution GOES visible imagery revealed multiple bands of high-reflectance cloud to the north/northeast of the center, along with clusters of convective cells closer to the core. Wavelet analysis identified a dominant multi-band wavelength of 30 km and a secondary peak at 15-20 km. Airborne radar measurements from IMPACTS flights showed deep convection near the center and narrow, elevated reflectivity bands linked to the multi-band features farther out. Numerical simulations reproduced the multi-bands, enabling exploration of their dynamical origin. Intrinsic phase speed calculations revealed that, contrary to several previous studies, the dominant multi-bands were not gravity waves. Instead, the features were identified as dynamic instabilities (Kelvin–Helmholtz instability) arising from vertical wind shear and low Richardson numbers near the upper-level outflow. Gravity waves were present in the low to mid levels (0–6 km) generated by convection, but they did not account for the strong perturbations in the mid to upper levels (6–10 km). This study presents new scientific insight into the governing dynamics of multi-banded structures in extratropical cyclones that highlights the role of shear-driven instabilities. 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 2 SIGNIFICANCE STATEMENT: Extratropical cyclones along the United States East Coast can25 produce intense bands of snow and other types of precipitation that cause travel chaos and potential26 loss of life. In this work, a storm of this type was studied using NASA aircraft and satellite27 data as well as computer simulations. The results show that the storm bands were not caused by28 atmospheric waves, as scientists often thought, but by sharp changes in the wind speed with height29 called shear. This shear created unstable layers that rolled and mixed the air, forming organized30 cloud and precipitation patterns. Understanding these fundamental physical processes can help31 forecasters better predict when and where heavy snow bands will form, improving warnings and32 public safety.33 1. Introduction34 Banded structures in extreme weather systems are perturbations to a balanced, background35 flow that can organize and concentrate variables such as moisture, momentum, and energy. In36 extratropical cyclones (ETCs), the focus of this paper, the concentration of these variables can37 lead to intense bands of multi-phase precipitation at the surface that are difficult to measure,38 model and predict with significant consequences for society. For example, snowfall associated39 with ETCs in the winter months can often organize into multiple bands that drop large amounts40 of snow in a short time causing vehicle crashes, flight cancellations and shutdowns of schools and41 businesses. The current understanding of the dynamical processes controlling the formation and42 evolution of these precipitation multi-bands (as opposed to a single, large-scale band) and their43 representation/predictability in numerical models is very limited. These limitations were part of44 the motivating factors for the recently completed Investigation of Microphysics and Precipitation45 for Atlantic Coast-Threatening Snowstorms (IMPACTS; McMurdie and Coauthors 2022) field46 experiment, which sought to improve the understanding of precipitation multi-bands in ETCs from47 various perspectives.48 Ganetis et al. (2018) used radar observations, soundings, and reanalysis data to examine the49 environments containing banded structures in a large set of ETC winter storms. They found50 that multi-bands were not well correlated with frontogenetical forcing and associated deformation51 zones. This result is consistent with the notion that deformation alone cannot explain the regular,52 oscillatory nature of the various fields connected to multi-bands. The authors also examined53 3 the presence of conditional symmetric instability (CSI) in the environment of various types of54 banded structures. While CSI was present in the vast majority of banded cases, there was no clear55 separation between single, large-scale bands and the smaller-scale, multi-band structures. This56 suggests that CSI may not be the fundamental property underlying the formation and evolution of57 multi-bands.58 In their review paper, Schultz and Schumacher (1999) noted that the banding of clouds and59 precipitation in ETCs appears to be weakly related to CSI. Xu (1989) used a viscous form of60 the Sawyer-Eliassen equation to understand banding mechanisms associated with frontogenetical61 forcing. For broad forcing and negative moist potential vorticity, which is a condition that reflects62 the presence of CSI, multiple bands develop with an intensity that scales with the degree of63 instability. However, it is not clear from the theory of Xu (1989) exactly how CSI coupled with64 moisture and lift, associated with frontogenesis, would produce the regularly spaced bands of65 clouds and precipitation observed in the real atmosphere. In an unstable environment, parcels are66 accelerated exponentially in the direction of the displacement and do not maintain an oscillatory67 behavior.68 Seltzer et al. (1985) and Schultz and Schumacher (1999) described several conditions that should69 be met to associate precipitation multi-bands with CSI. One of the key requirements is that the70 bands have no intrinsic propagation as they should be moving directly with the environmental flow.71 However, several studies of multi-bands do show intrinsic propagation, which when coupled with72 the description above, question the role of CSI in the band dynamics.73 Another possible culprit for multi-bands is the presence of a wave phenomenon to organize the74 oscillations in the state variables and provide the lift necessary to release any type of instability75 in the environment. Several previous studies have focused on the role of gravity waves as the76 mechanism for organizing the cloud and precipitation fields into multi-band structures in ETCs77 (e.g., Bosart et al. 1998; Koch and Siedlarz 1999; Ruppert et al. 2022). These waves can be78 generated from flow imbalances (e.g., associated with upper-level jets), convective perturbations,79 and/or flow over topography. Koch and Siedlarz (1999) documented large-amplitude mesoscale80 gravity waves (wavelengths of 200 - 260 km) during several winter storms using time series of81 surface pressure and wind measurements. These gravity waves were found to originate at upper82 levels from jet streak imbalances and the associated geostrophic adjustment process (e.g., Uccellini83 4 and Koch 1987). For large-amplitude waves, a surface reflection of the upper-level perturbation84 energy was detected and hypothesized to become trapped at lower-levels due to strong static stability85 (“wave ducting”; Lindzen and Tung 1976). This wave ducting allows for a longer residence time of86 the gravity wave energy in the lower levels of the atmosphere resulting in the potential for organized87 clouds and precipitation as shown by satellite and radar instruments.88 Zhang et al. (2001) also highlighted the role of mesoscale (∼200 km) gravity waves using89 numerical simulations of an ETC using various resolutions down to 4 km spacing. The authors90 proposed a conceptual model for the generation and evolution of mesoscale gravity waves based91 on these simulations. First, the waves were initiated in the upper levels of the system from the92 geostrophic adjustment process. Then, convection associated with the frontal features and local93 instability transported some of the gravity wave energy towards the surface where it can be ducted94 within a statically stable layer. This ducted wave then interacts with the moisture field to create a95 convectively-coupled wave that can be maintained and possibly amplified over a significant amount96 of time.97 Several questions and uncertainties come to mind regarding the applicability of these previous98 gravity wave studies to the general problem of ETC multi-bands. How does convective activity99 preferentially transport the wave energy initiated at upper levels (near the tropopause) downward100 to low levels and keep the wave intact? The presence of a strong static stability layer at lower101 levels and reflecting layer above to enable wave ducting appears to be a special set of circumstances102 and it is not clear how this layer can be formed and maintained in regions of the system that are103 convectively active.104 The goals of the present study are: (1) To determine the key structures and spatial/temporal scales105 of multi-banded precipitation features in ETCs and (2) To determine the origin and dynamical106 processes associated with these bands. To address the above goals and develop potentially new107 understanding of ETC multi-bands, a case study from the NASA IMPACTS field campaign is108 analyzed with multi-scale numerical simulations and a variety of remote sensing measurements.109 Case studies are an important first step towards developing a deep, holistic understanding of a110 physical process, which then allows that understanding to be tested more broadly on a larger111 collection of systems. The technical novelties of the present work are in the study of modern112 5 Fig. 1. Synoptic maps of mean sea level pressure and frontal analysis on 1 February, 2020 at (a) 1200 UTC and (b) 1800 UTC. remote sensing measurements and numerical models along with some new methods of analysis to113 the problem of ETC multi-bands.114 2. Brief overview of extratropical system115 During the winter of 2020, NASA organized the first phase of a multi-year effort called IMPACTS116 to study the microphysics and dynamics of ETCs with a focus on understanding banded regions of117 precipitation. On 1 February 2020, a double surface low-pressure system along a stationary front118 formed off the coast of the Carolinas with the southern low starting off at ∼1006 hPa on 1200119 UTC 1 February (Fig.1a). During the next 6 hours, the southern low intensified to 999 hPa at120 1800 UTC 1 February (Fig.1b) and to 998 hPa at 0000 UTC 2 February (not shown). During this121 period, multiple bands of convective clouds and precipitation formed to the north and northeast122 of the southern low-pressure center, which allowed detailed study of their characteristics. Flights123 from the NASA P-3 and ER-2 aircraft occurred during ∼1200 - 1800 UTC 1 February sampling124 the environment and banded features associated with the southern low.125 6 3. Data and processing126 a. EXRAD127 The ER-2 X-band Doppler Radar (EXRAD) is an X-band, downward-pointing airborne radar128 that measures radar reflectivity and Doppler velocity from hydrometeors with two beams and ∼20129 m gate spacing. The first beam is fixed and points nominally at nadir, while the second beam scans130 conically at 20 revolutions per minute at a nominal incidence angle of 32°. From the altitude of131 the NASA ER-2 aircraft (∼20 km), the swath width at the first level of useful data is ∼23 km.132 The EXRAD scanning beam mainand side-lobes interact strongly with the ocean surface and133 contaminate the precipitation signal below about 1 km height. The sampling of precipitation from134 the scanning beam is ∼500 m along-track and ∼2°in azimuth while the nadir beam along-track135 sampling is ∼50 m.136 Calculations of the three-dimensional (3D) wind field using the EXRAD scanning beam are137 performed using the 3D variational algorithm described in Guimond et al. (2014). The retrieval138 grid is set to 500 m in the horizontal dimensions and 250 m in the vertical dimension. Finer grid139 spacing in the vertical is possible, but was not deemed necessary. A two to three grid point running140 mean filter is applied to the raw retrievals in post-processing to remove numerical noise. The141 effective resolution of the wind fields produced from the algorithm has been analyzed with large142 eddy simulations in Guimond et al. (2018). Results show that scales of 5 Δ𝑥and larger are fully143 resolved, which translates to 2.5 km for the present study. Scales below this threshold, down to the144 grid scale, are subject to increasing kinetic energy attenuation.145 Quality control has been performed on the wind fields to remove data with low signal-to-noise146 ratios and high uncertainties. While a combination of parameters have been studied, the best quality147 fields were found by removing data with standard deviations larger than 6 m/s (error statistics are148 computed as part of the wind algorithm, see Guimond et al. (2014)). The data presented in this149 paper have this threshold applied. Validation of the EXRAD 3D winds with flight level in-situ data150 from the NASA ER-2 aircraft during IMPACTS 2020 have been performed. This validation showed151 zonal wind root mean square errors (RMSEs) of 3.99 m/s with a correlation coefficient of 0.92.152 For the meridional wind, the RMSEs are 4.53 m/s with a correlation coefficient of 0.89. These153 statistics are collected at various locations across the radar swath and for different time offsets.154 7 Further information about these error statistics and the EXRAD scanning beam wind retrievals can155 be found in (Guimond 2023).156 For scientific analysis, the EXRAD scanning beam is used to present the horizontal wind field,157 while the nadir beam is used to display the vertical wind field, since it has higher quality. Several158 corrections to the nadir beam Doppler velocities are needed before the vertical velocity can be159 analyzed for science. The expression for the nadir beam Doppler velocity (𝑉𝑑) is given by160 𝑉𝑑=® 𝑉𝑤·ˆ𝑒−® 𝑉𝑔·ˆ𝑒+𝑉𝑛(1) where ˆ𝑒=𝑥ˆ 𝑖+𝑦ˆ 𝑗+𝑧ˆ 𝑘 𝑟and161 ® 𝑉𝑤=𝑢ˆ 𝑖+𝑣ˆ 𝑗+ (𝑤+𝑣𝑡)ˆ 𝑘, (2) ® 𝑉𝑔=(𝐺𝑆ℎ∗𝑠𝑖𝑛𝑇)ˆ 𝑖+ (𝐺𝑆ℎ∗𝑐𝑜𝑠𝑇)ˆ 𝑗+ (𝐺𝑆𝑣)ˆ 𝑘(3) and 𝑉𝑛represents the effects of non-uniform beam filling.162 In these equations x,y,z are the Earth-relative coordinates of the radar pulse volumes (Guimond163 et al. 2014), ris the range, u,v,w are the components of the Earth-relative wind, 𝑣𝑡is the hydrometeor164 fallspeed, 𝐺𝑆ℎ,𝐺𝑆𝑣are the horizontal and vertical components of the aircraft ground speed and T165 is the aircraft track angle.166 Solving for the vertical velocity yields,167 𝑤=(𝑉𝑑−𝑉𝑛+® 𝑉𝑔·ˆ𝑒)𝑟−𝑢𝑥 −𝑣𝑦 𝑧−𝑣𝑡.(4) The calculation of hydrometeor fallspeeds from the reflectivity measurements follows the studies168 of Ulbrich and Chilson (1994) and Heymsfield et al. (2009). The non-uniform beam filling effects169 are removed from the Doppler velocities following Walker McLinden et al. (2021). Aircraft motion170 can result in antenna pointing angles that intercept the horizontal wind field, contaminating the171 vertical winds. The horizontal wind fields computed from the EXRAD scanning beam, described172 above, are used to remove this contamination from the nadir beam following equation 4.173 8 b. GOES-16174 High spatial and temporal resolution imagery of clouds from the advanced baseline imager on175 the Geostationary Operational Environmental Satellite (GOES) - 16 satellite at visible and infrared176 wavelengths are used to track banded structures. Specifically, reflectance data from the visible177 (channel 2) band at 500 m pixel spacing and 60 s time updates is utilized. This fine sampling in178 space and time provides the detailed structure of the cloud lifecycle as well as the organization and179 evolution into multi-bands.180 c. Numerical Simulations181 The Weather Research and Forecasting (WRF) model version 4.2 with the advanced research182 dynamic core is utilized to provide context for the observations. Four domains are utilized with183 a large, parent domain at 2 km grid spacing (domain 1) covering the full movement of the low-184 pressure system and banded features on 1 February 2020. Three nested domains at 0.667 km185 (domain 2), 0.222 km (domain 3) and 0.074 km (domain 4) grid spacing were placed to the North186 and Northeast of the low center to try and capture the finer scales of the multi-bands. All domains187 utilize 121 stretched vertical levels with a spacing of ∼75 m at the surface, ∼200 m at 10 km188 height and ∼900 m near the model top at 20 km height.189 Portions of the multi-bands were captured in domains 2 and 3, but not domain 4. The focus of190 the paper is on domain 1 for a few reasons. First, this domain captures the entirety of the system,191 which allows both large-scale and mesoscale features to be analyzed. Second, the wavelengths192 of dominant features found in the observations are 15-20 km and ∼30 km, which should be well193 resolved by the 2 km domain given the ∼7 - 8 Δ𝑥numerical dissipation range of WRF (e.g.,194 Guimond et al. 2016). The higher resolution domains were analyzed as part of this research and195 will be noted where appropriate.196 The setup of the model is as follows. The NCEP Global Data Assimilation System (GDAS)/Final197 (FNL) operational global analyses at 0.25 °spacing and 6 h temporal spacing are used as the initial198 and boundary conditions for the simulation. This widely used system combines the Global Forecast199 System (GFS) model with various synoptic-scale observations to achieve an optimal state of the200 atmosphere. For the 2 km parent domain the following sub-grid physics schemes were chosen:201 Thompson for microphysics, YSU for boundary layer (vertical diffusion), Smagorinsky-2D for202 9 is noted. At upper-levels (Fig. 7c) there are a few vorticity oscillations to the east of the along-track316 axis that appear to have connections with the corresponding reflectivity field (Fig. 6c).317 Two hours later at 1600 UTC, the same fields have changed significantly. The low-level reflectivity318 (Fig. 8a) reveals three bands oriented at different angles and denoted with dashed, black circles.319 The mid-level reflectivity (Fig. 8b) shows some signatures of the low-level bands, such as the320 southernmost circled region, but other bands to the north do not have the same orientation. These321 bands are oriented approximately perpendicular to the along-track axis that denotes the axis of322 band propagation. The upper-level reflectivity (Fig. 8c) shows narrow signatures of multi-bands323 that extend from near the system center out to approximately 300 km radius. The wavelengths324 of the reflectivity bands in Fig. 8 are ∼30 km, which is very similar to those documented from325 the GOES data, shown by the wavelet analysis in Fig. 5. The multi-bands present in the model326 reflectivity appear similar to those observed in the GOES data (Fig. 2).327 Figure 9 shows the absolute vertical vorticity field at 1600 UTC. The low-level (Fig. 9a) vorticity336 shows similar structure to the field at 1400 UTC. The mid-level (Fig. 9b) vorticity shows noticeable337 curvature in the bands with positive/negative oscillations more prominent to the east of the along-338 track axis. The upper-level (Fig. 9c) vorticity field shows a vibrant multi-banded structure to the339 northeast of the system center with vorticity oscillations up to ±1×10−3𝑠−1and wavelengths of ∼340 30 km. This structure is consistent with the reflectivity field shown in Fig. 8c.341 To examine the multi-bands more closely, the model data is output at two minute intervals and352 interpolated to a track-relative grid, centered on the black line in the preceding figures, with a grid353 spacing of 2 km in the across-track and along-track dimension while keeping the native model354 vertical spacing. In addition, the along-track (𝑈𝑎) and across-track (𝑈𝑥) velocities were computed355 on this grid,356 𝑈𝑎=𝑢cos(T) + 𝑣sin(T)(5) 𝑈𝑥=𝑢sin(T) − 𝑣cos(T)(6) where 𝑢and 𝑣are the zonal and meridional velocities and 𝑇is the grid track angle of ∼75 °.357 A positive along-track velocity is moving towards increasing along-track values (looking down358 16 Fig. 6. Horizontal cross sections of the simulated reflectivity (dBZ) in domain 1 (2.0 km grid spacing) on 1 February, 2020 at 1400 UTC. Panels (a), (b) and (c) show the 0 - 3 km, 4 - 7 km and 7 - 10 km height averaged fields, respectively. The black line in all panels denotes the along-track axis (normal to the phase lines) of multi-bands observed in subsequent figures. 328 329 330 331 the track), while a positive across-track velocity is moving towards increasing across-track values359 (from left to right looking down the track). One snapshot of the bands at different levels is shown in360 the next two figures. This snapshot is representative of the multi-band structure under investigation361 and thus, multiple snapshots are not shown.362 Figure 10 shows the absolute vorticity and vertical velocity averaged over the 6 - 10 km height363 range on 1 February, 2020 at 1530 UTC. In these figures, the raw model fields are filtered with a ∼364 20 km across-track running mean and a ∼10 km along-track running mean to reduce small-scale365 variability and pull out the larger scale features that were documented in the observations. The366 sensitivity of the multi-bands to model grid spacing was analyzed by comparing the structures367 17 Fig. 7. Horizontal cross sections of the simulated absolute vertical vorticity (s−1) in domain 1 (2.0 km grid spacing) on 1 February, 2020 at 1400 UTC. Panels (a), (b) and (c) show the 0 - 3 km, 4 - 7 km and 7 - 10 km height averaged fields, respectively. The black line in all panels denotes the along-track axis (normal to the phase lines) of multi-bands observed in subsequent figures. 332 333 334 335 in domain 1 (2 km) and domain 2 (0.67 km). The raw fields, before filtering, clearly show368 more oscillations and larger magnitudes in several variables on the higher resolution grid, which369 is expected. However, after filtering the fields to the same scales noted above, the multi-band370 structure looks similar in both domains with dominant horizontal wavelengths of ∼30 km. Since371 the observations also show dominant wavelengths of ∼30 km, the fields simulated in domain 1 are372 deemed sufficient to examine the dynamics.373 The vertical velocity (Fig. 10b) displays a similar multi-banded structure as the vorticity field378 (Fig. 10a), although the vorticity perturbations are more vibrant and continuous when compared379 to the vertical velocity field. The message from Fig. 10 is that the positive phase lines of vorticity380 18 Fig. 8. Horizontal cross sections of the simulated reflectivity (dBZ) in domain 1 (2.0 km grid spacing) on 1 February, 2020 at 1600 UTC showing the southern low pressure system with embedded multi-bands. Panels (a), (b) and (c) show the 0 - 3 km, 4 - 7 km and 7 - 10 km height averaged fields, respectively. The black, dashed circles highlight the locations of several bands identified in panel (a). These circles are copied onto the panels in (b) and (c). The black line in all panels denotes the along-track axis (normal to the phase lines) of multi-bands observed in panels (b) and (c). 342 343 344 345 346 347 in the multi-bands are largely uncorrelated with the positive phase lines of vertical velocity. To381 clarify, the dashed, white lines in Fig. 10a are generally out of phase with the peaks in vertical382 velocity shown in Fig. 10b, which indicates that the stretching of pre-existing vorticity is not the383 primary driver of the vorticity oscillations.384 Figure 11a highlights a series of vorticity oscillations in the mid-levels (3 - 6 km height average)385 at the same time as Fig. 10 that have some similarities to the upper-level features, such as the386 strong anomaly at ∼75 km along-track. However, the vorticity oscillations in the mid-levels387 19 Fig. 9. Horizontal cross sections of the simulated absolute vertical vorticity (s−1) in domain 1 (2.0 km grid spacing) showing the southern low pressure system with embedded multi-bands. Panels (a), (b) and (c) show the 0 - 3 km, 4 - 7 km and 7 - 10 km height averaged fields, respectively. The black line in all panels denotes the along-track axis (normal to the phase lines) of multi-bands shown in Figure ??. 348 349 350 351 are substantially weaker in magnitude and not as well-defined as those in the upper levels. The388 corresponding vertical velocity field (Fig. 11b) continues to be out of phase with the vorticity389 with the exception of the strong anomaly at ∼75 km along-track, which is a band of convection390 closer to the system center. While there are several bands of vertical velocity between 50 - 250391 km along-track, these bands do not seem to be as organized in the across-track direction as those392 shown in the upper-levels (Fig. 10b).393 Figure 12 shows vertical cross sections of the multi-bands averaged between ±50 km across-394 track. Distinct oscillations in vorticity (Fig. 12a) are visible in the 6 - 10 km layer that extend395 from ∼50 - 300 km along track. Below 6 km, the vorticity pattern is not obvious, although some396 20 Fig. 10. Horizontal cross sections, averaged over the 6 - 10 km height range, of simulated data on the track-relative grid revealing properties of multi-bands. Panels (a) and (b) show the simulated absolute vorticity (s−1) and vertical velocity (m s−1), respectively. The white, dashed lines highlight the positive perturbations in vorticity, which are copied onto the vertical velocity plot. 374 375 376 377 oscillations that are tilted down the track with height are visible in the ∼2 - 6 km layer from ∼397 125 - 200 km along-track. The vertical velocity field (Fig. 12b) shows similar oscillations to the398 vorticity in the 6 - 10 km layer, most apparent in the negative perturbations, but the vertical velocity399 and vorticity are mostly out of phase as previously discussed. The main exception is at 75 km400 along-track, where a strong, positive vertical velocity anomaly is collocated with a strong, positive401 vorticity anomaly. This feature is part of the main rotating convective band located just northeast402 of the system center highlighted in Fig. 8 and Fig. 9. Positive correlations in vorticity and vertical403 velocity are also apparent at ∼125 km along-track below 6 km height.404 What are the dynamics governing the oscillations observed in the upper-levels of the system?405 One might suspect the perturbations are gravity waves and we evaluate that potential here. The406 21 Fig. 11. The same as in Figure 10, only for data averaged over the 3 - 6 km height range. Note the color bar has been expanded on the vertical velocity panel compared to Fig. 10b. total phase speed of the bands is calculated by using the two-minute model output to track the lines407 of constant phase using the across-track velocity field (shown later in Fig. 14b) averaged over ±408 50 km across-track and 6 - 10 km height. Tracking these phase lines resulted in mean total phase409 speeds of 26.67 m s−1. In order to estimate the possibility of intrinsic propagation, the mean flow410 in the direction of wave propagation must be removed from the total phase speed. The vertical411 cross sections of the bands (Fig. 12) clearly show that they are present within the 6 - 10 km layer.412 Sensitivity tests in the across-track averaging interval were performed for 100 km, 200 km and413 300 km track-relative grids. These tests showed that the mean along-track velocity, averaged over414 the appropriate across-track distance and the 6 - 10 km layer were ∼27.0 ±0.5 m s−1. Thus, the415 measured intrinsic phase speed of the multi-bands in this layer is ∼0ms−1.416 For completeness, the theoretical gravity wave speed for this environment is also calculated. The417 dispersion relation for internal gravity waves is418 22 Fig. 12. Vertical cross sections of simulated data on 1 February, 2020 at 1530 UTC, averaged over the ±50 km across-track range showing (a) absolute vorticity and (b) vertical velocity. (𝜔−¯𝑢𝑘)2𝑘2+𝑚2−𝑁2𝑘2=0 (7) where 𝜔is the angular frequency, ¯𝑢is the large-scale, averaged horizontal windspeed in the419 direction of wave propagation, 𝑁is the Brunt–V¨ ais¨ al¨ a frequency, 𝑘is the horizontal wavenumber,420 and 𝑚is the vertical wavenumber.421 Rearranging Eq. (7) for the intrinsic horizontal phase speed on the left-hand-side yields422 𝜔/𝑘−¯𝑢=𝑁/√︁𝑘2+𝑚2.(8) The right-hand-side of Eq. (8) is evaluated using the data from the simulation. Note that the423 buoyancy frequency was estimated to be 50 times larger than the Coriolis frequency and thus, the424 potential gravity waves are not significantly affected by the Earth’s rotation. Taking an average425 of data over the system at two different time periods and in the 6 - 10 km layer produced values426 of 𝑁around 10−2s−1. The dominant horizontal wavelength in the model is ∼30 km and the427 23 Fig. 13. Vertical cross sections of simulated data on 1 February, 2020 at 1530 UTC, averaged over the ±50 km across-track range showing (a) perturbation vertical velocity (m s−1) and (b) perturbation reflectivity (dBZ). The black arrows in panel (a) denote features discussed in the text. 435 436 437 vertical wavelength is taken to be twice the depth of the perturbations, which equates to 8 km.428 Entering these numbers produces a horizontal phase speed of 12.30 m s−1, which is representative429 of the bands propagating down the track-relative grid to the northeast of the system center. It is430 clear from this calculation that the measured intrinsic phase speeds have a large mismatch with431 the theoretical intrinsic phase speeds for internal gravity waves, even with significant uncertainty432 bounds in various parameters. Thus, the observed multi-bands in the upper-levels cannot be gravity433 waves.434 However, there are more subtle oscillations in several variables in the low to middle levels of438 the system that are revealed through examining perturbation fields. In this analysis, perturbations439 are defined as deviations of the total variables from the 50 km along-track filtered variables. The440 perturbation vertical velocity at 1530 UTC (Fig. 13a) shows the prominent multi-bands in the 6441 24 - 10 km layer along with weaker perturbations in the ∼0 - 6 km layer denoted by black arrows.442 The weaker oscillations exhibit amplitudes approximately three to four times lower than those in443 the upper layer, making them clearly distinguishable despite the similar wavelengths of ∼30 km.444 Animations of vertical velocity (not shown) appear to show these waves emanating from the deep445 convection present at ∼75 km along-track. The perturbation reflectivity (Fig. 13b) is largely446 consistent with the vertical velocity except the low to middle level oscillations are about ten times447 lower in amplitude than the upper layer waves. This is why the multi-bands are not as visible in448 the low-level precipitation field as shown by the reflectivity (Fig. 8a).449 The low to middle level waves (0 - 6 km layer) were tracked in the perturbation vertical velocity450 field with the two minute model output and the total phase speeds were measured. This procedure451 produced total wave phase speeds of ∼30 ±3 m/s. The along-track velocity was averaged across-452 track (±50 km), along-track (0 - 342 km) and height (0 - 6 km) to represent the mean flow moving453 the waves, which resulted in values of ∼14 ±0.5 m/s. Thus, the measured intrinsic phase speeds of454 the waves were ∼16 m/s. The theoretical intrinsic phase speeds for gravity waves were computed455 using the same inputs as before with the exception of a 12 km vertical wavelength (two times the456 6 km depth of the vertical velocity perturbations). These inputs produced values of 17.73 m/s,457 which is close to the measured intrinsic phase speed of ∼16 m/s. Thus, these low to middle level458 waves can be identified as internal gravity waves that are being generated by the convective activity459 closer to the system center.460 If the dominant, upper-level multi-bands are not gravity waves, what is driving their dynamics?464 Figure 14 shows vertical cross sections of the along-track velocity and across-track velocity. The465 along-track velocity (Fig. 14a) shows a jet centered at ∼11 km height with large values of vertical466 shear (maximum values of 0.01 s−1) located between 8 - 10 km height and most notably between467 50 - 225 km along-track. The locations of the large vertical wind shear values match well with468 the locations of the wave motions associated with the multi-bands. For example, the across-track469 velocity (Fig. 14b) reveals clear wave motions in the 6 - 10 km layer and between 50 - 250 km470 along-track. There are perhaps extensions of the wave motions down to ∼4 - 5 km in some regions,471 but overall, the oscillations become less detectable below 6 km height.472 The along-track and across-track velocity fields shown in Fig. 14 match well with the EXRAD473 observations shown in Fig. 4, despite the differences in spatial/temporal coverage (also note the474 25 Acknowledgments. The IMPACTS field campaign and the work of author Guimond are supported599 by the NASA Earth Venture Suborbital (EVS) program and NASA grant 80NSSC24K0302. Author600 Guimond thanks the IMPACTS team members for organizing the campaign, providing forecasting601 support, coordinating the flights and collecting the data. Many thanks are given to Dr. Paul Reasor602 for multiple, lengthy discussions on the work that improved this paper considerably. Feedback on603 the gravity wave analysis from Dr. Steven Koch is also greatly appreciated.604 Data availability statement. The data sets needed to reproduce the core results of this paper can605 be found at the following link: https://doi.org/10.5281/zenodo.17253490606 References607 Barnes, H. C., J. P. Zagrodnik, L. A. McMurdie, A. K. Rowe, and J. Houze, R. A., 2018:608 Kelvin–helmholtz waves in precipitating midlatitude cyclones. Journal of the Atmospheric Sci-609 ences,75 (8), 2763–2785, https://doi.org/10.1175/JAS-D-17-0365.1.610 Bosart, L. F., W. E. Bracken, and A. Seimon, 1998: A study of cyclone mesoscale structure with em-611 phasis on a large-amplitude inertia–gravity wave. Monthly Weather Review,126 (6), 1497–1527,612 https://doi.org/10.1175/1520-0493(1998)126⟨1497:ASOCMS⟩2.0.CO;2, URL https://journals.613 ametsoc.org/view/journals/mwre/126/6/1520-0493 1998 126 1497 asocms 2.0.co 2.xml.614 Frank, W. M., and E. A. Ritchie, 2001: Effects of vertical wind shear on the intensity and structure615 of numerically simulated hurricanes. Monthly Weather Review,129, 2249–2269.616 Galperin, B., S. Sukoriansky, and P. S. Anderson, 2007: On the critical richardson number in stably617 stratified turbulence. Atmospheric Science Letters,8 (3), 65–69, https://doi.org/10.1002/asl.153,618 URL https://doi.org/10.1002/asl.153.619 Ganetis, S. A., B. A. Colle, S. E. Yuter, and N. P. Hoban, 2018: Environmental conditions620 associated with observed snowband structures within northeast u.s. winter storms. Mon. Wea.621 Rev.,146, 3675–3690, https://doi.org/https://doi.org/10.1175/MWR-D-18-0054.1.622 Guimond, S., 2023: Er-2 x-band radar (exrad) 3d winds impacts. NASA Global Hydrometeorol-623 ogy Resource Center DAAC, Huntsville, Alabama, U.S.A., URL http://dx.doi.org/10.5067/624 IMPACTS/EXRAD/DATA201, dataset available online, https://doi.org/10.5067/IMPACTS/625 EXRAD/DATA201.626 32 Guimond, S. R., 2007: A diagnostic study of the effects of trough interactions on tropical cyclone627 qpf. Master’s thesis, Florida State University, Tallahassee, FL.628 Guimond, S. R., J. M. Reisner, S. Marras, and F. X. Giraldo, 2016: The impacts of dry dynamic629 cores on asymmetric hurricane intensification. Journal of the Atmospheric Sciences,73 (12),630 4661–4684, https://doi.org/10.1175/JAS-D-16-0055.1.631 Guimond, S. R., S. Sroka, and D. Protzko, 2018: A large eddy simulation of hurricane inten-632 sification. 33rd Conference on Hurricanes and Tropical Meteorology, American Meteorolog-633 ical Society, 17, URL https://science.gsfc.nasa.gov/sed/content/uploadFiles/publication files/634 large eddy 33hurr guimond.pdf, available online.635 Guimond, S. R., L. Tian, G. M. Heymsfield, and S. J. Frasier, 2014: Wind Retrieval Algo-636 rithms for the IWRAP and HIWRAP Airborne Doppler Radars with Applications to Hurri-637 canes. Journal of Atmospheric and Oceanic Technology,31 (6), 1189–1215, https://doi.org/638 10.1175/JTECH-D-13-00140.1.639 Hanley, D., J. Molinari, and D. Keyser, 2001: A composite study of the interactions between640 tropical cyclones and upper-tropospheric troughs. Monthly Weather Review,129 (10), 2570–641 2584, https://doi.org/10.1175/1520-0493(2001)129⟨2570:ACSOTI⟩2.0.CO;2.642 Heymsfield, G. M., L. Tian, A. J. Heymsfield, L. Li, and S. Guimond, 2009: Characteristics of deep643 tropical and subtropical convection from nadir-viewing high-altitude airborne doppler radar.644 Journal of the Atmospheric Sciences,67, 285–308, https://doi.org/10.1175/2009JAS3132.1,645 submitted March 2009, revised June 2009.646 Houser, J. L., and H. B. Bluestein, 2011: Polarimetric doppler radar observations of kelvin–647 helmholtz waves in a winter storm. Journal of the Atmospheric Sciences,68 (7), 1676–1702,648 https://doi.org/10.1175/2011JAS3566.1.649 Koch, S. E., and L. M. Siedlarz, 1999: Mesoscale gravity waves and their environment in the650 central united states during storm-fest. Mon. Wea. Rev.,127, 2854– 2879, https://doi.org/https:651 //doi.org/10.1175/1520-0493(1999)127%3C2854:MGWATE%3E2.0.CO;2.652 33 Lindzen, R., and K. Tung, 1976: Banded convective activity and ducted gravity waves. Mon. Wea.653 Rev.,104, 1602– 1617, https://doi.org/10.1175/1520-0493(1976)104⟨1602:BCAADG⟩2.0.CO;654 2.655 Liu, Y., X. San Liang, and R. H. Weisberg, 2007: Rectification of the bias in the wavelet power656 spectrum. Journal of Atmospheric and Oceanic Technology,24 (12), 2093–2102, https://doi.org/657 10.1175/2007JTECHO511.1.658 Lonfat, M., F. D. Marks, and S. S. Chen, 2004: Precipitation distribution in tropical cyclones659 using the tropical rainfall measuring mission (trmm) microwave imager: A global perspective.660 Monthly Weather Review,132, 1645–1660.661 McMurdie, L., and Coauthors, 2022: Chasing snowstorms: The investigation of microphysics and662 precipitation for atlantic coast-threatening snowstorms (impacts) campaign. Bull. Amer. Meteor.663 Soc.,103, E1243–E1269, https://doi.org/https://doi.org/10.1175/BAMS-D-20-0246.1.664 Medina, S., and J. Houze, R. A., 2016: Kelvin-helmholtz waves in extratropical cyclones passing665 over mountain ranges. Quarterly Journal of the Royal Meteorological Society,142 (697), 1311–666 1319, https://doi.org/10.1002/qj.2755.667 Merrill, R. T., 1988: Characteristics of the upper tropospheric environmental flow around668 hurricanes. Journal of the Atmospheric Sciences,45, 1665–1677, https://doi.org/10.1175/669 1520-0469(1988)045⟨1665:COTUTE⟩2.0.CO;2.670 Molinari, J., and D. Vollaro, 1989: External influences on hurricane intensity. part i: Outflow671 layer eddy angular momentum fluxes. Journal of the Atmospheric Sciences,46 (8), 1093–1105,672 https://doi.org/10.1175/1520-0469(1989)046⟨1093:EIOHIP⟩2.0.CO;2.673 Nolan, D. S., and M. J. Onderlinde, 2022: The representation of spiral gravity waves in a mesoscale674 model with increasing horizontal and vertical resolution. Journal of Advances in Modeling675 Earth Systems,14 (10), e2022MS003 067, https://doi.org/10.1029/2022MS003067, URL https:676 //doi.org/10.1029/2022MS003067.677 Protzko, D. E., S. R. Guimond, C. R. Jackson, J. W. Sapp, Z. Jelenak, and P. S. Chang, 2023:678 Documenting coherent turbulent structures in the boundary layer of intense hurricanes through679 34 wavelet analysis on iwrap and sar data. IEEE Transactions on Geoscience and Remote Sensing,680 61, 1–16, https://doi.org/10.1109/TGRS.2023.3305998.681 Ruppert, J. H., S. E. Koch, X. Chen, Y. Du, A. Seimon, Y. Q. Sun, J. Wei, and L. F.682 Bosart, 2022: Mesoscale gravity waves and midlatitude weather: A tribute to fuqing683 zhang. Bulletin of the American Meteorological Society,103 (1), E129–E156, https://doi.org/684 10.1175/BAMS-D-20-0005.1.685 Schultz, D. M., and P. N. Schumacher, 1999: The use and misuse of conditional symmetric instabil-686 ity. Mon. Wea. Rev.,127, 2709–2732, https://doi.org/https://doi.org/10.1175/1520-0493(1999)687 127,2709:TUAMOC.2.0.CO;2.688 Seltzer, M., R. Passarelli, and K. Emanuel, 1985: The possible role of symmetric instability689 in the formation of precipitation bands. J. Atmos. Sci.,42, 2207–2219, https://doi.org/https:690 //doi.org/10.1175/1520-0469(1985)042%3C2207:TPROSI%3E2.0.CO;2.691 Torrence, C., and G. P. Compo, 1998: A practical guide to wavelet analysis. Bulletin of the Ameri-692 can Meteorological Society,79 (1), 61–78, https://doi.org/10.1175/1520-0477(1998)079⟨0061:693 APGTWA⟩2.0.CO;2.694 Uccellini, L. W., and S. E. Koch, 1987: The synoptic setting and possible energy sources for695 mesoscale wave disturbances. Mon. Wea. Rev.,115, 721–729, https://doi.org/https://doi.org/10.696 1175/1520-0493(1987)115%3C0721:TSSAPE%3E2.0.CO;2.697 Ulbrich, C. W., and P. B. Chilson, 1994: Effects of variations in precipitation size distribution698 and fallspeed law parameters on relations between mean doppler fallspeed and reflectivity699 factor. Journal of Atmospheric and Oceanic Technology,11, 1656–1663, https://doi.org/10.700 1175/1520-0426(1994)011⟨1656:EOVIPS⟩2.0.CO;2.701 Walker McLinden, M. L. W., A. M. Loftus, L. Li, and G. M. Heymsfield, 2021: Ap-702 plication of nonuniform beam filling (nubf) doppler velocity error correction on airborne703 radar measurements. IEEE Geoscience and Remote Sensing Letters,19, 1–5, https://doi.org/704 10.1109/LGRS.2021.3120647.705 Xu, Q., 1989: Frontal circulations in the presence of small viscous moist symmetric stability and706 weak forcing. Quart. J. Roy. Meteor. Soc.,115, 1325–1353.707 35 Zhang, F., C. Davis, M. Kaplan, and S. Koch, 2001: Wavelet analysis and the governing dynamics708 of a large-amplitude mesoscale gravity-wave event along the east coast of the united states. Q.J.R.709 Meteorol. Soc.,127, 2209–2245, https://doi.org/10.1002/qj.49712757702.710 36