• Allen, D. R., and N. Nakamura, 2003: Tracer equivalent latitude: A diagnostic tool for isentropic transport studies. J. Atmos. Sci., 60, 287304, https://doi.org/10.1175/1520-0469(2003)060<0287:TELADT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Allen, D. R., J. L. Stanford, L. S. Elson, E. F. Fishbein, L. Froidevaux, and J. W. Waters, 1997: The 4-day wave as observed from the Upper Atmosphere Research Satellite Microwave Limb Sounder. J. Atmos. Sci., 54, 420434, https://doi.org/10.1175/1520-0469(1997)054<0420:TDWAOF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Allen, D. R., A. R. Douglass, and S. E. Strahan, 2013: The large-scale frozen-in anticyclone in the 2011 Arctic summer stratosphere. J. Geophys. Res. Atmos., 118, 26562672, https://doi.org/10.1002/JGRD.50256.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Andrews, D., C. Leovy, and J. Holton, 1987: Middle Atmosphere Dynamics Academic Press, 502 pp.

  • Australian Government, 2019: Dangerous bushfire weather in spring 2019. Australian Government Special Climate Statement 72, 28 pp., http://www.bom.gov.au/climate/current/statements/scs72.pdf.

  • Bishop, C. H., and A. J. Thorpe, 1994: Potential vorticity and the electrostatics analogy: Quasi-geostrophic theory. Quart. J. Roy. Meteor. Soc., 120, 713731, https://doi.org/10.1002/qj.49712051710.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bonavita, M., L. Isaksen, and E. Hólm, 2012: On the use of EDA background error variances in the ECMWF 4D-Var. ECMWF Tech. Memo. 664, 33 pp., https://www.ecmwf.int/en/elibrary/8272-use-eda-background-error-variances-ecmwf-4d-var.

    • Crossref
    • Export Citation
  • Boone, C. D., P. F. Bernath, and M. D. Fromm, 2020: Pyrocumulonimbus stratospheric plume injections measured by the ACE-FTS. Geophys. Res. Lett., 47, e2020GL088442, https://doi.org/10.1029/2020GL088442.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Buehner, M., P. L. Houtekamer, C. Charette, H. L. Mitchell, and B. He, 2010: Intercomparison of variational data assimilation and the ensemble Kalman filter for global deterministic NWP. Part II: One-month experiments with real observations. Mon. Wea. Rev., 138, 15671586, https://doi.org/10.1175/2009MWR3158.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Clayton, A. M., A. C. Lorenc, and D. M. Barker, 2013: Operational implementation of a hybrid ensemble/4D-Var global data assimilation system at the Met Office. Quart. J. Roy. Meteor. Soc., 139, 14451461, https://doi.org/10.1002/qj.2054.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Daley, R., 1991: Atmospheric Data Analysis. Cambridge University Press, 457 pp.

  • Daley, R., and E. Barker, 2001: NAVDAS Source Book 2001: NRL Atmospheric Variational Data Assimilation System. NRL Publ. NRL/PU/7530—01-441, 163 pp., http://www.dtic.mil/docs/citations/ADA396883.

  • Eckermann, S. D., and Coauthors, 2009: High-altitude data assimilation system experiments for the northern summer mesosphere season of 2007. J. Atmos. Sol.-Terr. Phys., 71, 531551, https://doi.org/10.1016/j.jastp.2008.09.036.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fairlie, T. D., 1995: Three-dimensional transport simulations of the dispersal of volcanic aerosol from Mount Pinatubo. Quart. J. Roy. Meteor. Soc., 121, 19431980, https://doi.org/10.1002/qj.49712152809.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fairlie, T. D., R. B. Pierce, J. A. Al-Saadi, W. L. Grose, J. M. Russell, M. H. Proffitt, and C. R. Webster, 1999: The contribution of mixing in Lagrangian photochemical predictions of polar ozone loss over the Arctic in summer 1997. J. Geophys. Res., 104, 26 59726 609, https://doi.org/10.1029/1999JD900111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fromm, M., D. T. Lindsey, R. Sevranckx, G. Yue, T. Tricki, R. Sica, P. Doucet, and S. Godin-Beekman, 2010: The untold story of pyrocumulonimbus. Bull. Amer. Meteor. Soc., 91, 11931210, https://doi.org/10.1175/2010BAMS3004.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 54195454, https://doi.org/10.1175/JCLI-D-16-0758.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grise, K. M., D. W. J. Thompson, and T. Birner, 2010: A global survey of static stability in the stratosphere and upper troposphere. J. Climate, 23, 22752292, https://doi.org/10.1175/2009JCLI3369.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harvey, V. L., R. B. Pierce, T. D. Fairlie, and M. H. Hitchman, 2002: A climatology of stratospheric polar vortices and anticyclones. J. Geophys. Res., 107, 4442, https://doi.org/10.1029/2001JD001471.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harvey, V. L., R. B. Pierce, M. H. Hitchman, C. E. Randall, and T. D. Fairlie, 2004: On the distribution of ozone in stratospheric anticyclones. J. Geophys. Res., 109, D24308, https://doi.org/10.1029/2004JD004992.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harvey, V. L., C. E. Randall, G. L. Manney, and C. S. Singleton, 2008: Low-ozone pockets observed by EOS-MLS. J. Geophys. Res., 113, D17112, https://doi.org/10.1029/2007JD009181.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Haynes, P. H., 1990: High-resolution three-dimensional modeling of stratospheric flows: Quasi-two-dimensional turbulence dominated by a single vortex. Topological Fluid Mechanics, H. K. Moffat and A. Tsinober, Eds., Cambridge University Press, 345–354.

  • Hogan, T., and Coauthors, 2014: The Navy Global Environmental Model. Oceanography, 27 (3), 116125, https://doi.org/10.5670/oceanog.2014.73.

  • Holton, J. R., 2004: An Introduction to Dynamic Meteorology. Vol. 1. Elsevier Academic Press, 535 pp.

  • Hooghiem, J. D., M. E. Popa, T. Röckmann, J.-U. Grooß, I. Tritscher, R. Müller, R. Kivi, and H. Chen, 2020: Wildfire smoke in the lower stratosphere identified by in situ CO observations. Atmos. Chem. Phys. Discuss., https://doi.org/10.5194/acp-2020-65, in press.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jones, S. C., 1995: The evolution of vortices in vertical shear. 1. Initially barotropic vortices. Quart. J. Roy. Meteor. Soc., 121, 821851, https://doi.org/10.1002/qj.49712152406.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kablick, G. P., D. R. Allen, M. D. Fromm, and G. E. Nedoluha, 2020: Australian pyroCb smoke generates synoptic-scale stratospheric anticyclones. Geophys. Res. Lett., 47, e2020GL088101, https://doi.org/10.1029/2020GL088101.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khaykin, S., S. Godin-Beekmann, A. Hauchecorne, J. Pelon, F. Ravetta, and P. Keckhut, 2018: Stratospheric smoke with unprecedentedly high backscatter observed by lidars above southern France. Geophys. Res. Lett., 45, 16391646, https://doi.org/10.1002/2017GL076763.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khaykin, S., and Coauthors, 2020: The 2019/20 Australian wildfires generated a persistent smoke-charged vortex rising up to 35 km altitude. Commun. Earth Environ., 1, 22, https://doi.org/10.1038/s43247-020-00022-5.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kuhl, D. D., T. E. Rosmond, C. H. Bishop, J. McLay, and N. L. Baker, 2013: Comparison of hybrid ensemble/4DVar and 4DVar within the NAVDAS-AR data assimilation framework. Mon. Wea. Rev., 141, 27402758, https://doi.org/10.1175/MWR-D-12-00182.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Laprise, R., 1992: The resolution of global spectral models. Bull. Amer. Meteor. Soc., 73, 14531455, https://doi.org/10.1175/1520-0477-73.9.1453.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Manney, G. L., N. J. Livesey, C. J. Jimenez, H. C. Pumphrey, M. L. Santee, I. A. MacKenzie, and J. W. Waters, 2006: EOS Microwave Limb Sounder observations of “frozen-in” anticyclonic air in Arctic summer. Geophys. Res. Lett., 33, L06810, https://doi.org/10.1029/2005GL025418.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCormack, J., and Coauthors, 2017: Comparison of mesospheric winds from a high-altitude meteorological analysis system and meteor radar observations during the boreal winters of 2009–2010 and 2012–2013. J. Atmos. Sol.-Terr. Phys, 154, 132166, https://doi.org/10.1016/j.jastp.2016.12.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McLay, J. G., C. H. Bishop, and C. A. Reynolds, 2008: Evaluation of the ensemble transform analysis perturbation scheme at NRL. Mon. Wea. Rev., 136, 10931108, https://doi.org/10.1175/2007MWR2010.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ohneiser, K., and Coauthors, 2020: Smoke of extreme Australian bushfires observed in the stratosphere over Punta Arenas, Chile, in January 2020: Optical thickness, lidar ratios, and depolarization ratios at 355 and 532 nm. Atmos. Chem. Phys., 20, 80038015, https://doi.org/10.5194/acp-20-8003-2020.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Orsolini, Y. J., 2001: Long-lived tracer patterns in the summer polar stratosphere. Geophys. Res. Lett., 28, 38553858, https://doi.org/10.1029/2001GL013103.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peterson, D., J. Campbell, E. Hyer, M. Fromm, G. Kablick, J. Cossuth, and M. DeLand, 2018: Wildfire-driven thunderstorms cause a volcano-like stratospheric injection of smoke. npj Climate Atmos. Sci., 1, 30, https://doi.org/10.1038/S41612-018-0039-3.

    • Search Google Scholar
    • Export Citation
  • Peterson, D., E. Hyer, J. Campbell, M. Fromm, C. Bennese, M. Berman, and T. Van, 2019: Quantifying the impact of intense pyroconvection on stratospheric aerosol loading. 2019 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, Abstract GC11F-1150, https://agu.confex.com/agu/fm19/meetingapp.cgi/Paper/510480.

  • Pumphrey, H. C., M. L. Santee, N. J. Livesy, M. J. Schwartz, and W. G. Read, 2011: Microwave Limb Sounder observations of biomass-burning products from the Australian bush fires of February 2009. Atmos. Chem. Phys., 11, 62856296, https://doi.org/10.5194/acp-11-6285-2011.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rosmond, T., and L. Xu, 2006: Development of NAVDAS-AR: Non-linear formulation and outer loop tests. Tellus, 58A, 4558, https://doi.org/10.1111/j.1600-0870.2006.00148.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thiéblemont, R., N. Huret, Y. J. Orsolini, A. Hauchecorne, and M.-A. Drouin, 2011: Frozen-in anticyclones occurring in polar Northern Hemisphere during springtime: Characterization, occurrence and link with quasi-biennial oscillation. J. Geophys. Res., 116, D20110, https://doi.org/10.1029/2011JD016042.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Torres, O., 2019: OMPS-NPP L2 NM aerosol index swath orbital, version 2. Goddard Earth Sciences Data and Information Services Center, accessed February 2020, https://doi.org/10.5067/40L92G8144IV.

    • Crossref
    • Export Citation
  • Xu, L., T. Rosmond, and R. Daley, 2005: Development of NAVDAS-AR: Formulation and initial tests of the linear problem. Tellus, 57A, 546559, https://doi.org/10.3402/tellusa.v57i4.14710.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, P., and Coauthors, 2019: Black carbon lofts wildfire smoke high into the stratosphere to form a persistent plume. Science, 365, 587590, https://doi.org/10.1126/science.aax1748.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    CALIPSO 532 nm total attenuated backscatter (km−1 sr−1) on 31 Jan 2020 for a nighttime track (time is ~0547 UTC) that passed through the P1 SWIRL. Green lines are NAVGEM θ (K) and white lines are NAVGEM PV anomaly (20%, 40%, 60%, 80%) at 0600 UTC 31 Jan 2020, for the range θ ≥ 350 K. Bottom of plot indicates latitude positions along the track (which is shown in Fig. 2a).

  • View in gallery

    (a) CALIPSO track for curtain plot shown in Fig. 1. (b) Cross section along CALIPSO track. Gray shaded region indicates PV anomaly >20%, representing the location of the smoke plume as seen by close correspondence in Fig. 1. Blue lines are temperature anomaly (–2, –4 K); red lines are temperature anomaly (2, 4, 6 K); green lines are zonal wind contours: 5, 10, 15 m s−1 (solid), −5, −10 m s−1 (dashed); and black lines are geopotential height anomaly contours: 20, 40, 60, 80 m. All anomalies are relative to the zonal mean values. Also included is a plot of the stratospheric AOD (gray dots, with scale) within the region of the plume, calculated using the method described in K20 (scale included with maximum value ~0.3). (c) NAVGEM T (filled color contours) and θ (black lines) along the CALIPSO track. The gray contour indicates PV anomaly = 20%.

  • View in gallery

    PV anomaly (filled color contours) over the SH at 1200 UTC 20 Jan 2020 at (a) θ = 540 K and (b) θ = 420 K. (c) OMPS UVAI on 20 Jan 2020. Locations of P1, P2, and P3 are indicated by the black dots, and boundaries of search regions are indicated by dashed lines for P1 (red) and for P2 and P3 (black).

  • View in gallery

    The horizontal position of the P1 plume, λP1(t), φP1(t), from 0600 UTC 4 Jan to 0600 UTC 3 Mar 2020. The colors indicate the day since 4 Jan and the lines plot the AC edge.

  • View in gallery

    (a) Potential temperature of the plume P1, θP1(t). (b) Altitude of P1, ZP1(t). The black lines indicate the central location determined from the PV anomaly, while the red dots indicate the minimum and maximum extent using the anticyclone selection criteria. Blue lines are linear least squares fits to the potential temperature (shifted down by 200 K for easier visibility) and altitude (shifted down by 10 km) values of the SWIRL from 4 to 31 Jan 2020 and from 1 Feb to 3 Mar 2020. Slopes are indicated for each fit.

  • View in gallery

    PV anomaly (colored filled contours), streamlines (gray lines), AC edge (thick black lines), location of PV maximum (black dot), location of streamfunction minimum (black asterisk), and AC calculation box (dotted red lines) for 0000 UTC 31 Jan 2020 from 535 to 715 K in 20 K increments. Levels from 555 to 695 K meet the SWIRL threshold criteria described in section 3. The 615 K level includes the CALIPSO track used for Fig. 1, with green (black) indicating inside (outside) the P1 plume. Also plotted is the level 3 OMPS UVAI from 30 Jan 2020 using colored contours. Note that the OMPS data are measured at ~1800 UTC, so the 30 Jan data are ~6 h before of the NAVGEM analyses at 0000 UTC 31 Jan, while the CALIPSO track is at ~0600 UTC 31 Jan, ~6 h after the NAVGEM analyses.

  • View in gallery

    Time series plots of the (a) streamfunction minimum (black) and streamfunction at the AC edge as defined in Eq. (7) (red) and (b) ratio of the edge to minimum streamfunction values.

  • View in gallery

    PV anomaly (filled color contours), streamlines (gray lines), AC edge (thick black lines), location of PV maximum (black dot), location of streamfunction minimum (black asterisk), and AC calculation box (dotted red lines) for 1200 UTC 20 Jan 2020 from 425 to 625 K in 20 K increments, centered on the P2 plume over Antarctica.

  • View in gallery

    Curtain plots in time and potential temperature of (a) PV maximum and (b) flow speed following the P1 plume from 0600 UTC 4 Jan to 0600 UTC 3 Mar 2020. Black dots indicate minimum and maximum levels determined from the anticyclone selection criteria.

  • View in gallery

    PV anomaly (filled color contours), streamlines (gray lines), AC edge (thick black lines), location of PV maximum (black dot), location of streamfunction minimum (black asterisk), and AC calculation box (dotted red lines) for 0000 UTC 10 Jan to 0000 UTC 29 Mar 2020 in 5-day increments. The potential temperature is indicated, which is the central value determined from the PV anomaly calculation.

  • View in gallery

    Cross sections in longitude and θ through the latitude of the P1 plume location in 5-day intervals, from 10 Jan to 29 Feb 2020. Date (latitude) are provided in the title of each panel. PV anomaly (filled color contours); temperature anomaly: 2, 4, 6 K (solid green), −2, −4, −6 K (dashed green); and wind contours: 5, 10, 15 m s−1 (solid black), −5, −10, −15 m s−1 (dashed black).

  • View in gallery

    Curtain plots in time and θ following the P1 plume from 0600 UTC 4 Jan to 0600 UTC 3 Mar 2020. (a) T anomaly (K), (b) large-scale zonal wind (m s−1), (c) small-scale zonal wind (m s−1), (d) small-scale meridional wind (m s−1), (e) O3 anomaly (ppmv). (f) Plume O3 (red) and zonal mean O3 (black) as a function of altitude following the plume, smoothed with a 2-day running mean to remove noise.

  • View in gallery

    Time series plots of (a) PV anomaly maximum, (b) flow speed (V) at the central level, (c) magnitude of the vertical wind shear of the vector wind, (d) thickness (D), (e) size (L), (f) ozone anomaly, (g) minimum (black) and maximum (red) T anomaly, and (h) T difference between maximum and minimum anomalies.

  • View in gallery

    Schematic diagram showing the initial tilting and subsequent evolution of an AC vortex placed in an environmental vertical shear in the SH. The vortex is divided into two parts: a lower-level AC (blue circle) and an upper-level AC (red circle). The environmental wind has westerlies at lower levels and easterlies at upper levels, as indicated by the block arrows. The tilt is defined as a vector connecting the lower- and upper-level centers (black arrows), with magnitude τ and angle α = 0° (180°) pointing east (west). Panels show successive evolution from (a) initial shear causing tilt angle (α) of 180° after which secondary rotation (in direction of colored arrows) begins and continues through tilt angles of (b) 225°, (c) 270°, and (d) 315°.

  • View in gallery

    Time series plots of the (a) difference between the longitude (black) and latitude (red) of ψ˜min at the top and bottom of the SWIRL and (b) angle (α) between the location of the ψ˜min at the top and bottom of the SWIRL. Blue lines are linear least squares fits to the data from day 41–49 and day 50–59, which have slopes of −34° and −30° day−1, respectively. (c) Times series of magnitude of the tilt (τ) between the top and bottom of the SWIRL.

  • View in gallery

    PV anomaly (colored filled contours), streamlines (gray lines), AC edge (thick black lines), location of PV maximum (black dot), location of streamfunction minimum (black asterisk), and AC calculation box (dotted red lines) for 0000 UTC 28 Aug 2017 from 420 to 520 K in 10 K increments, centered on one of the PNE plumes over Europe.

  • View in gallery

    SH PV anomaly following the central θ level (indicated in panel titles) of the P1 plume. NAVGEM analyses (all at 0000 UTC) are provided for (a) 10, (b) 12, (c) 14, and (d) 16 Jan 2020. NAVGEM forecasts (initialized at 0000 UTC 10 Jan) of PV anomaly are provided for (e) 10, (f) 12, (g) 14, and (h) 16 Jan 2020.

All Time Past Year Past 30 Days
Abstract Views 462 41 0
Full Text Views 348 191 14
PDF Downloads 333 159 9

Smoke with Induced Rotation and Lofting (SWIRL) in the Stratosphere

View More View Less
  • 1 Remote Sensing Division, Naval Research Laboratory, Washington, D.C.
Full access

Abstract

The Australian bushfires of 2019/20 produced an unusually large number of pyrocumulonimbus (pyroCb) that injected huge amounts of smoke into the lower stratosphere. The pyroCbs from 29 December 2019 to 4 January 2020 were particularly intense, producing hemispheric-wide aerosol that persisted for months. One plume from this so-called Australian New Year (ANY) event evolved into a stratospheric aerosol mass ~1000 km across and several kilometers thick. This plume initially moved eastward toward South America in January, then reversed course and moved westward passing south of Australia in February and eventually reached South Africa in early March. The peculiar motion was related to the steady rise in plume potential temperature of ~8 K day−1 in January and ~6 K day−1 in February, due to local heating by smoke absorption of solar radiation. This heating resulted in a vertical temperature anomaly dipole, a positive potential vorticity (PV) anomaly, and anticyclonic circulation. We call this dynamical component of the smoke plume “smoke with induced rotation and lofting” (SWIRL). This study uses Navy Global Environmental Model (NAVGEM) analyses to detail the SWIRL structure over 2 months. The main diagnostic tool is an anticyclone edge calculation based on the scalar Q diagnostic. This provides the framework for calculating the time evolution of various SWIRL properties: PV anomaly, streamfunction, horizontal size, vertical thickness, flow speed, and tilt. In addition, we examine the temperature anomaly dipole, the SWIRL interaction with the large-scale wind shear, and the ozone anomaly associated with lofting of air from the lower to the middle stratosphere.

For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Douglas Ray Allen, douglas.allen@nrl.navy.mil

Abstract

The Australian bushfires of 2019/20 produced an unusually large number of pyrocumulonimbus (pyroCb) that injected huge amounts of smoke into the lower stratosphere. The pyroCbs from 29 December 2019 to 4 January 2020 were particularly intense, producing hemispheric-wide aerosol that persisted for months. One plume from this so-called Australian New Year (ANY) event evolved into a stratospheric aerosol mass ~1000 km across and several kilometers thick. This plume initially moved eastward toward South America in January, then reversed course and moved westward passing south of Australia in February and eventually reached South Africa in early March. The peculiar motion was related to the steady rise in plume potential temperature of ~8 K day−1 in January and ~6 K day−1 in February, due to local heating by smoke absorption of solar radiation. This heating resulted in a vertical temperature anomaly dipole, a positive potential vorticity (PV) anomaly, and anticyclonic circulation. We call this dynamical component of the smoke plume “smoke with induced rotation and lofting” (SWIRL). This study uses Navy Global Environmental Model (NAVGEM) analyses to detail the SWIRL structure over 2 months. The main diagnostic tool is an anticyclone edge calculation based on the scalar Q diagnostic. This provides the framework for calculating the time evolution of various SWIRL properties: PV anomaly, streamfunction, horizontal size, vertical thickness, flow speed, and tilt. In addition, we examine the temperature anomaly dipole, the SWIRL interaction with the large-scale wind shear, and the ozone anomaly associated with lofting of air from the lower to the middle stratosphere.

For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Douglas Ray Allen, douglas.allen@nrl.navy.mil

1. Introduction

The Black Summer Australian bushfire season of 2019–20 was fueled by unusual springtime conditions, including record daytime temperatures, very low humidity, and gusty winds that caused record forest fire danger index (FFDI) over 60% of the country (Australian Government 2019). Dangerous fires in New South Wales and Victoria culminated in a series of very large pyrocumulonimbus (pyroCb) from 29 December 2019 to 4 January 2020 that lofted huge smoke clouds well into the lower stratosphere, with aerosol mass estimates between 0.3 and 0.9 Tg (Peterson et al. 2019) and 0.2–0.6 Tg (Khaykin et al. 2020). While much of the smoke dispersed throughout the Southern Hemisphere (SH), several localized smoke plumes were long lived, and evidence from satellite data and meteorological analyses showed that the local heating in these plumes caused unusual meteorological features (Kablick et al. 2020, hereafter K20). For example, one plume (called “P1” in K20) rose in altitude from 15 to 30 km over several weeks and exhibited anomalies in potential vorticity (PV) and temperature along with anticyclonic circulation of ~15 m s−1 around the ~1000 km diameter feature. Some of the smaller plumes (“P2” and “P3”) also showed continued rising in the stratosphere and PV anomalies. While lofting of smoke plumes due to local diabatic heating has been observed and modeled previously (Khaykin et al. 2018; Yu et al. 2019; Hooghiem et al. 2020), the smoke-induced rotation in the stratosphere was heretofore an unknown phenomenon, and this discovery opens up a new area of scientific research.

Further research into the behavior of the Australian New Year (ANY) event plumes was performed by Khaykin et al. (2020). They examined meteorological and composition anomalies associated with three different plumes (P1, P2, and P3) using European Centre for Medium-Range Weather Forecasts (ECMWF) analyses and various satellite data. They provided details of the horizontal position, altitude, and maximum vorticity of the plumes, as well as the aerosol, ozone, water vapor, and carbon monoxide anomalies. They also examined Global Navigation Satellite System (GNSS) radio occultation (RO) temperature profiles and radiosonde winds in the vicinity of vortices, as well as ECMWF forecasts of the P1 plume vorticity. Additional observational analyses have confirmed the presence of unusual anomalies of aerosol, smoke, and trace gases in the stratosphere associated with the ANY event plumes. Ohneiser et al. (2020) showed aerosol optical thickness anomalies associated with the “P1” plume as it passed over Punta Arenas Chile using ground-based lidar depolarization ratios and spaceborne lidar from Cloud–Aerosol Lidar with Orthogonal Polarization (CALIOP) and Aeolus Atmospheric Laser Doppler Instrument (ALADIN). Boone et al. (2020) examined Atmospheric Chemistry Experiment Fourier Transform Spectrometer (ACE-FTS) data for numerous atmospheric constituents, showing aerosol and gaseous enhancements occurring as part of the “P3” plume, in addition to plumes from the Pacific Northwest event and Black Saturday pyroCbs. In this paper we further the work of K20 and others by quantifying the dynamical evolution of P1 using diagnostic tools developed for understanding stratospheric anticyclones.

To simplify the discussion in this paper, we hereafter call the induced meteorological component of large smoke plumes “smoke with induced rotation and lofting” (SWIRL). Previous studies of long-lived stratospheric anticyclones (ACs) may help to provide further insight into analyses of SWIRLs. For example, pioneering work by Harvey et al. (2002, 2004, 2008) provided a framework for isolating ACs from the ambient meteorology and assessing their composition, vertical structure, and occurrence frequency. Diagnostic tools from this work were later used by Allen et al. (2013) to examine frozen-in anticyclones (FrIACs) in the Northern Hemisphere (NH) summer. Originally discovered by Manney et al. (2006), FrIACs occur when tropical air is advected poleward after the springtime polar stratospheric final warming (Thiéblemont et al. 2011). The AC is generated due to PV conservation, and the horizontal air motion causes the FrIAC to have tracer (e.g., nitrous oxide) mixing ratios characteristic of low latitudes. Due to the slow mixing processes in the summer easterlies, these tracer anomalies can remain at high latitudes for months (Orsolini 2001). While FrIACs have some similarities to SWIRLs, there are major differences. Unlike FrIACs, SWIRLs ascend diabatically, have a significant temperature anomaly dipole, and have composition anomalies due to vertical rather than horizontal motion (K20). Diagnosing properties of SWIRLs is challenging since the SWIRL is traveling horizontally and vertically, evolving, and interacting with other meteorological features.

In this paper, we lay out a framework for quantifying SWIRL dynamical properties. We focus our analysis on the ANY event P1 SWIRL identified by K20, employing AC diagnostics developed for FrIACs. This particular work does not detail either the formation stage or final demise of the P1 SWIRL, but rather focuses on the mature stage when the AC is relatively easy to identify and characterize. This work also does not attempt to understand completely the complex physical processes that maintain the SWIRL, but rather provides a comprehensive diagnostic study to lay the groundwork for future studies. Follow-up research is needed to understand how SWIRLs form, are maintained, and decay, as well as their occurrence frequency. This paper is organized as follows. Section 2 details the meteorological analyses used in this study. Section 3 presents the diagnostic approach to characterize SWIRL properties. Section 4 discusses the evolution of the P1 SWIRL over nearly two months, while section 5 examines how the SWIRL is maintained in the presence of vertical shear. Sections 6 and 7 provide a discussion and summary, respectively, and section 8 provides an outlook with suggestions for future research directions.

2. Model description

The Navy Global Environmental Model (NAVGEM) global forecast model uses a semi-Lagrangian/semi-implicit integration of the hydrostatic equation, the first law of thermodynamics, and conservation of moisture and ozone (Hogan et al. 2014). This study employs a high-altitude (HA) research version (NAVGEM-HA) using a hybrid sigma–pressure coordinate with 74 vertical levels (useful levels up to ~100 km) as described in Eckermann et al. (2009) and McCormack et al. (2017). There are 18 levels in the stratosphere (~14–50 km), which is the focus of our work, and vertical spacing in the stratosphere is ~2 km. The model is run at a triangular truncation T119 (Gaussian grid of 360° longitude × 180° latitude, with horizontal resolution of ~1.0°). The predicted variables are vorticity, divergence, virtual potential temperature, specific humidity, ozone mixing ratio (O3), and surface pressure. In addition, the zonal (u) and meridional (υ) wind, temperature (T), geopotential height (Z), potential temperature (θ), potential vorticity (q), and streamfunction (ψ) are derived fields.

The NAVGEM data assimilation (DA) system employs a hybrid 4DVar method (e.g., Buehner et al. 2010; Bonavita et al. 2012; Clayton et al. 2013; Kuhl et al. 2013). NAVGEM minimizes a quadratic cost function using the accelerated representer approach as described in Xu et al. (2005) and Rosmond and Xu (2006). The hybrid 4DVar combines conventional and ensemble background error covariances. The conventional initial background error covariance is calculated using an analytic formulation that employs the hydrostatic relationship in the vertical between geopotential and temperature, and wind–geopotential correlations based on approximate geostrophic balance on an f plane, i.e., constant Coriolis parameter with latitude (Daley 1991; Daley and Barker 2001; Kuhl et al. 2013). The ensemble covariance is created with 80 ensemble members, which are updated each cycle using the ensemble transform (ET) approach described by McLay et al. (2008) and Kuhl et al. (2013). The covariance blend used in this analysis is 25% ensemble and 75% conventional. Analyses are produced every six hours using the background 3–9 h forecast from the previous analysis window and assimilation of the operational suite of meteorological observations including conventional data, infrared and microwave satellite radiances, feature-track winds, GNSS RO measurements, and Solar Backscatter Ultraviolet (SBUV) ozone. The NAVGEM-HA (hereafter referred to as simply NAVGEM) 6-hourly output used in this study is from a near-real-time experiment that produced analyses within ~3 days of the current date, allowing our research team to analyze the dynamics of the ANY event as it unfolded.

While NAVGEM fields are output on hybrid sigma model levels, for this study we interpolate several fields (u, υ, T, Z, O3, and q) to a specified set of θ levels ranging from 320 to 950 K in 5 K increments. While 5 K is much higher resolution than the native model grid (~50 K in the stratosphere), interpolating the data to 5 K increments is helpful for diagnosing features of interest in this study. The time range for this work is 0600 UTC 4 January–0600 UTC 3 March 2020. With 6-hourly data, there are 237 individual analyses.

While NAVGEM provides the primary meteorological data used in this paper, we also use the Modern-Era Retrospective Analysis for Research and Applications, version 2 [MERRA-2; see Gelaro et al. (2017) for details], reanalyses for a brief comparative analysis (in section 6) of a plume associated with the Pacific Northwest event (PNE) pyroCbs at 0000 UTC 28 August 2017.

3. SWIRL diagnostics

a. Connection of smoke to meteorological anomalies

As discussed in K20, an important diagnostic for determining the location and extent of SWIRLs is the anomaly of potential vorticity, defined here as Ertel PV (Andrews et al. 1987),
q=ζa×θρ,
where ζa is absolute vorticity (relative vorticity plus planetary vorticity), ρ is density, and θ is potential temperature. We first calculated q on NAVGEM model levels and then interpolate q to potential temperature surfaces from θ1 = 320 K to θ2 = 950 K in Δθ = 5 K increments, resulting in q = q(λ, φ, θ, t), where λ, φ, and t refer to longitude, latitude, and time, respectively. Next, we calculate the PV anomaly (indicated by a prime) from the zonal mean (indicated by an overbar) and express this as a percentage:
q(λ,φ,θ,t)=q(λ,φ,θ,t)q¯(φ,θ,t)|q¯(φ,θ,t)|×100%.

To illustrate the relationship between PV anomaly and smoke, we plot in Fig. 1 the CALIOP lidar 532 nm attenuated backscatter cross section through the plume on 31 January, along with NAVGEM PV anomaly contours. As shown in Fig. 2a, the Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) track extends southward from the west coast of South America toward Antarctica. On this date, the plume extends vertically from ~550 to 700 K (~22–28 km). PV anomaly contours closely correspond to the location of the smoke plume, with the 20% level approximately marking the plume boundary.

Fig. 1.
Fig. 1.

CALIPSO 532 nm total attenuated backscatter (km−1 sr−1) on 31 Jan 2020 for a nighttime track (time is ~0547 UTC) that passed through the P1 SWIRL. Green lines are NAVGEM θ (K) and white lines are NAVGEM PV anomaly (20%, 40%, 60%, 80%) at 0600 UTC 31 Jan 2020, for the range θ ≥ 350 K. Bottom of plot indicates latitude positions along the track (which is shown in Fig. 2a).

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

Fig. 2.
Fig. 2.

(a) CALIPSO track for curtain plot shown in Fig. 1. (b) Cross section along CALIPSO track. Gray shaded region indicates PV anomaly >20%, representing the location of the smoke plume as seen by close correspondence in Fig. 1. Blue lines are temperature anomaly (–2, –4 K); red lines are temperature anomaly (2, 4, 6 K); green lines are zonal wind contours: 5, 10, 15 m s−1 (solid), −5, −10 m s−1 (dashed); and black lines are geopotential height anomaly contours: 20, 40, 60, 80 m. All anomalies are relative to the zonal mean values. Also included is a plot of the stratospheric AOD (gray dots, with scale) within the region of the plume, calculated using the method described in K20 (scale included with maximum value ~0.3). (c) NAVGEM T (filled color contours) and θ (black lines) along the CALIPSO track. The gray contour indicates PV anomaly = 20%.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

The close correspondence between PV anomaly and smoke (identified previously by K20) is an exciting new discovery. Khaykin et al. (2020) showed similar correspondence using the relative vorticity. While a complete understanding of the mechanics of SWIRL formation and maintenance requires detailed radiative transfer modeling beyond the scope of this paper, the basic SWIRL forcing mechanism appears to be sustained, localized heating from absorption of solar radiation by the black carbon in the plume. This warming provides a buoyancy that lifts the plume, and also a radial, outward pressure gradient that forces expansion of the plume, analogous to an expanding hot-air balloon. A schematic diagram of the SWIRL structure is provided in Fig. 2b, which shows anomalies (relative to the zonal mean) of PV, temperature, wind, and geopotential height associated with the P1 plume. Here the shaded region of q′ > 20% provides a proxy for location of smoke. In addition, for further quantification of the presence of smoke, we plot the along-track stratospheric aerosol optical depth (AOD), which was calculated using the method described in K20. In this along-track/pressure cross section, the pressure gradient anomaly is represented by geopotential height anomalies (black lines), which form concentric circles decreasing in magnitude outward from the center. In the vertical direction, the expansion includes ascent above the plume center, with accompanying adiabatic cooling, resulting in a negative temperature anomaly (relative to the surrounding air). Below the plume center, downward expansion results in a positive temperature anomaly. For reference, we also show in Fig. 2c the temperature and potential temperature (not anomalies) associated with the SWIRL. The vertical temperature dipole can be seen as a local minimum and maximum, and the potential temperature contours “bulge” upward above and downward below the plume center, consistent with vertical expansion.

In the horizontal direction, the outward pressure gradient will tend to balance with the Coriolis force to produce a synoptic-scale vorticity anomaly with anticyclonic circulation. The combined vorticity and temperature anomalies produce a PV anomaly that closely resembles the “PV charge” concept introduced by Bishop and Thorpe (1994), and the details of Fig. 2b are similar to the PV charge schematics presented in that paper and also in Fig. 14a of Allen et al. (1997). Maintaining this PV charge structure over months requires sustained heating, since radiative cooling and wind shear will tend to destroy the SWIRL. Khaykin et al. (2020) showed that 10-day ECMWF forecasts of the P1 plume without aerosol heating do not match the observed ascent and are unable to maintain the vorticity anomaly. We further test the ability of weather models to maintain SWIRL circulation using NAVGEM (results shown in section 6).

In this paper, we have developed a three-step process for identifying and characterizing the ANY event P1 SWIRL. This basic procedure provides a framework for characterizing other plumes as well, although modifications to specific settings may be necessary depending on the characteristics of the particular event. As explained below, the three steps are 1) identifying PV anomalies, 2) determining the anticyclone edge, and 3) calculating SWIRL diagnostics. A summary of the parameters used for each of these steps is provided in Table 1.

Table 1.

Parameters used for the P1 plume analysis.

Table 1.

b. Identifying PV anomaly

We first developed an algorithm to detect and track PV anomalies associated with smoke plumes. The location of the maximum PV anomaly is used to define the position of the SWIRL in longitude, latitude, and θ for each 6-hourly NAVGEM analysis. The horizontal location of the SWIRL is determined by finding the maximum value of q′ within a given geographic region bounded in longitude and latitude: λ1, λ2, φ1, φ2:
qmax(θ,t)=maxλ1,λ2,φ1,φ2[q(λ,φ,θ,t)].

Note that in the NH, the SWIRL signature will have a negative q′, so the minimum should be used in Eq. (3). The longitude and latitude bounds are determined by examining smoke diagnostics from satellite observations such as the Ozone Mapping and Profiler Suite (OMPS) ultraviolet aerosol index (UVAI) (Torres 2019). As an illustration, Fig. 3 shows q′at two θ levels for 1200 UTC 20 January 2020 over the SH along with the OMPS UVAI. Large UVAI is seen off the southern tip of South America, along with large q′ at 540 K, both associated with the P1 plume. For the P1 plume we generally use a search boundary of [40°–90°S, 0°–360°], which is indicated by the red dashed line in Fig. 3a. Due to northward motion of P1, the boundary was altered from 28 February to 3 March to avoid other unrelated PV anomalies (see Table 1 for details). Figure 3a also illustrates that if we shift the search boundary poleward (black dashed line), the detection algorithm captures the P2 plume over Antarctica at θ = 540 K. Figure 3b shows that another boundary (black dashed line) could be used to capture the P3 plume over the Atlantic Ocean at θ = 420 K, which is also accompanied by a strong UVAI anomaly. Note that the P2 plume is too far poleward to be observed in the OMPS UVAI data. In the following analyses we focus on P1, but Fig. 3 shows how the approach can be used to follow other plumes as well.

Fig. 3.
Fig. 3.

PV anomaly (filled color contours) over the SH at 1200 UTC 20 Jan 2020 at (a) θ = 540 K and (b) θ = 420 K. (c) OMPS UVAI on 20 Jan 2020. Locations of P1, P2, and P3 are indicated by the black dots, and boundaries of search regions are indicated by dashed lines for P1 (red) and for P2 and P3 (black).

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

At each time and θ level from 320 to 950 K, we save the qmax(θ,t) and the horizontal position λmax(θ, t), φmax(θ, t). To determine the central θ of the plume, at each time t, we vertically smooth qmax(θ,t) using a five-point (20 K) boxcar smoother (approximately one-half of the resolution the native model level grid) and then find the level with the maximum of the smoothed function:
θP1(t)=where[maxθ1,θ2{smooth[qmax(θ,t)]}]

As a final criterion, to avoid other unrelated PV features, we set a lower bound on θ equal to 420 K + 4 K day−1 × (tt0), where t is time in days and t0 is 0000 UTC 3 January 2020.

The horizontal positions for the plume center are determined from
λP1(t)=λmax(θP1,t),φP1(t)=φmax(θP1,t).

For a large plume like P1, this algorithm works well, tracking the plume for over two months. This approach provides an independent estimate of the plume position and altitude that can be compared with other observations (e.g., from CALIOP). Care must be taken to track smaller plumes, like the ANY event P2 and P3 plumes (K20) or to isolate multiple plumes occurring at different levels. In the latter case, the bounds of θ1 and θ2 may have to be more finely adjusted with time: θ1 = θ1(t), θ2 = θ2(t).

Additional plume diagnostics will focus on other meteorological properties, such as streamfunction, but the PV anomaly positions λP1(t), φP1(t), θP1(t) provide the crucial reference position for these further diagnostics. The horizontal positions λP1(t), φP1(t) over the analysis period are indicated by colored numbers in Fig. 4 (colored lines indicate the anticyclone edge described in section 3c). The horizontal position of P1 moves initially eastward over the Pacific Ocean to South America, then westward past Australia, and continues northwestward, crossing the tip of South Africa on 3 March. Note that for this study, we often reference the time as days since the start of SWIRL detection (day 0 is 4 January 2020, day 28 is 1 February 2020, day 57 is 1 March, and day 59 is 3 March). The smoke plume continued westward after 3 March to South America (K20), and remnants exist until early April (Khaykin et al. 2020), but the anticyclone edge calculation that we use here loses the plume before it reaches the Atlantic Ocean. The vertical position θP1(t) is provided in Fig. 5a (also discussed in K20). In addition, in Fig. 5b we plot the vertical position in altitude using ZP1(t) = Z[λP1(t), φP1(t), θP1(t)]. The plume altitude rose from ~400 to 800 K (~17–30 km) over the course of 2 months. Also plotted in Fig. 5 are linear least squares fits of θP1 and ZP1 versus time, which provide independent estimates of the heating and ascent rates that can be compared with observations of aerosol or trace gases. We obtained heating (ascent) rates of 7.8 K day−1 (0.28 km day−1) over 4–31 January 2020 and 5.5 K day−1 (0.16 km day−1) for 1–28 February 2020. In comparison, Khaykin et al. (2020) calculated an initial ascent rate of ~10 K day−1 (~0.45 km day−1) and a 3-month average (4 January–1 April) ascent rate of 5.94 ± 0.07 K day−1 (~0.2 km day−1).

Fig. 4.
Fig. 4.

The horizontal position of the P1 plume, λP1(t), φP1(t), from 0600 UTC 4 Jan to 0600 UTC 3 Mar 2020. The colors indicate the day since 4 Jan and the lines plot the AC edge.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

Fig. 5.
Fig. 5.

(a) Potential temperature of the plume P1, θP1(t). (b) Altitude of P1, ZP1(t). The black lines indicate the central location determined from the PV anomaly, while the red dots indicate the minimum and maximum extent using the anticyclone selection criteria. Blue lines are linear least squares fits to the potential temperature (shifted down by 200 K for easier visibility) and altitude (shifted down by 10 km) values of the SWIRL from 4 to 31 Jan 2020 and from 1 Feb to 3 Mar 2020. Slopes are indicated for each fit.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

c. Determining the AC edge

To diagnose SWIRL properties, we incorporate an AC-edge algorithm based on the “Q diagnostic” approach discussed in Harvey et al. (2002) and used by Allen et al. (2013) for characterizing a FrIAC in the NH summer stratosphere. The approach is based on the scalar quantity Q, which is a measure of the relative contribution of strain and rotation on the wind field (Haynes 1990; Fairlie 1995; Fairlie et al. 1999):
Q=12(1acosφdudλυatanφ)2+12(1adυdφ)2+(1a2cosφdudφdυdλ+utanφa2dudφ),
where a is Earth’s radius, and other quantities are as previously defined. The sign of Q is positive (negative) where strain (rotation) dominates the flow; integrating Q around streamlines can therefore provide a determination of the AC edge as the streamline where the Q integral changes sign (Harvey et al. 2002):
ψ˜EDGE(θ,t)where[ψ˜Q(θ,t)=0].

The first step in AC-edge identification is to calculate ψ on the previously specified isentropic surfaces. We start by interpolating winds to θ surfaces, u = u(λ, φ, θ, t) and υ = υ(λ, φ, θ, t), and then for each θ and t we invert the winds using spherical harmonic transformations (at the native resolution of T119) to calculate ψ = ψ(λ, φ, θ, t) [see Holton (2004) for a discussion of this inversion]. While this raw ψ works well for FrIACs (Allen et al. 2013), we found that for the P1 SWIRL there were times in its evolution where other large-scale AC signatures were present, making it difficult to exclusively isolate the P1 AC. This was partly due to the smaller horizontal size of the SWIRL (~1000 vs ~3000 km for the FrIAC), partly due to the weaker strength of the circulation (~20 vs ~40 m s−1 for the FrIAC), and partly due to the ambient meteorology. We therefore decided to filter ψ to isolate smaller-scale features. We found that a spectral filter (SF) using triangular truncation with total wavenumber (n) of greater than or equal to N = 10 worked well for the duration of the P1 lifetime. The triangular truncation offers a uniform resolution filter on the globe, with N ≥ 10 representing spatial scales of less than ~2000 km based on the L2 estimate of Laprise (1992). This filtered streamfunction, ψ˜(λ,φ,θ,t)=SFn=Nn=119[ψ(λ,φ,θ,t)], was calculated at each θ and t. The shape of ψ˜ can be seen in the maps in Fig. 6, which are for 0000 UTC 31 January 2020 at 10 isentropic levels from 535 to 715 K, in 20 K increments. The gray and black lines are select streamlines at each level, with most of them showing closed circulation loops with AC (counterclockwise in the SH) rotation.

Fig. 6.
Fig. 6.

PV anomaly (colored filled contours), streamlines (gray lines), AC edge (thick black lines), location of PV maximum (black dot), location of streamfunction minimum (black asterisk), and AC calculation box (dotted red lines) for 0000 UTC 31 Jan 2020 from 535 to 715 K in 20 K increments. Levels from 555 to 695 K meet the SWIRL threshold criteria described in section 3. The 615 K level includes the CALIPSO track used for Fig. 1, with green (black) indicating inside (outside) the P1 plume. Also plotted is the level 3 OMPS UVAI from 30 Jan 2020 using colored contours. Note that the OMPS data are measured at ~1800 UTC, so the 30 Jan data are ~6 h before of the NAVGEM analyses at 0000 UTC 31 Jan, while the CALIPSO track is at ~0600 UTC 31 Jan, ~6 h after the NAVGEM analyses.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

To verify the correspondence of the AC with smoke, we also plot in Fig. 6 the OMPS UVAI (similar to Fig. 3b in K20). The UVAI shows a peak coincident with the PV maximum and the closed streamlines. There is an interesting secondary peak on this day that occurs over South America. The cause of this is uncertain, but may be related to orographic gravity waves modulating the smoke plume. The CALIOP track used in Fig. 1 is also included on the 615 K level in Fig. 6, with green (black) indicating inside (outside) the P1 plume. The position of the plume in longitude and latitude corresponds closely to the closed AC lines in Fig. 6.

Once ψ˜ has been calculated, the second step in AC-edge identification is isolate individual contours of ψ˜ on which to calculate the average Q. Allen et al. (2013) used a fixed interval of 0.5 × 107 m2 s−1 for calculations on a single isentropic level (θ = 850 K) for FrIAC identification. For the SWIRL, which ascends in altitude (K20), we decided to adjust the values relative to the minimum streamfunction ψ˜min at each level and time. We use K = 101 levels from the minimum value to zero, in equal intervals of ψ˜min/(K1). As with the PV anomaly calculation, this analysis was performed within a chosen region defined by longitude and latitude bounds. Since the PV anomaly gives the local position, we center this region on λP1(t), φP1(t) and choose dimensions of the longitude–latitude box that encompasses the SWIRL, but is not so large that unrelated dynamical features are incorporated into the analysis. We call this a “box” to distinguish from the “region” criteria used for finding the PV anomaly maximum. While in principle both could be defined identically, we found it convenient to use separated conditions. We found that a box of width 40° in longitude and 20° in latitude (see red dotted lines in Fig. 6) worked for most of the P1 lifetime. Some adjustments were made to the box near the beginning and end of the period considered to avoid other features (set Table 1 for details). We then used a contouring procedure to discretize longitude–latitude positions of points along each streamline within these boxes. Both ψ˜min within this box and its location λψ˜min(t),φψ˜min(t) will be used in later diagnostics.

The third step for AC-edge identification is to calculate Q using Eq. (6) with small-scale horizontal winds obtained from ψ˜:
u˜=1adψ˜dφ,υ˜=1adψ˜dλ.

We then interpolate Q to the discrete points along each streamline and integrate these Q values along K streamlines, giving Q(ψ˜1),Q(ψ˜2),,Q(ψ˜K). The ψ˜ value at which the integrated Q changes sign is determined as the “AC edge” or ψ˜EDGE(θ,t). The AC-edge streamline is plotted in Fig. 6 using thick black contours and is plotted in Fig. 4 for each day of the time period analyzed.

A time series plot of ψ˜min(θP1,t), scaled by 106 m2 s−1, is provided in Fig. 7a (black line). As a reference, if the “environmental” circulation has a value of zero for ψ˜, and if the AC is circular and centered on the pole with a constant zonal wind speed of 10 m s−1 over a latitude range of 10°, then the minimum streamfunction can be obtained by integrating Eq. (8): ψ = a × u × [φ2φ1] = −1.11 × 107 m2 s−1. So, a scaled minimum streamfunction of about −11 would be associated with wind speed 10 m s−1, a value of −22 would be associated with 20 m s−1, etc. This is consistent with the calculated flow values for the P1 SWIRL to be discussed later. The ψ˜EDGE(θP1,t) values are plotted with a red line in Fig. 7a. These vary with time similarly to ψ˜min(θP1,t), and when the ratio is taken, it is approximately 0.5 throughout the lifetime, with some variations (Fig. 7b). This suggests that a simple edge calculation algorithm could select 0.5ψ˜min(θP1,t), without having to go through the calculation of Q. However, we retain the full calculation here for a closer theoretical tie to the Q diagnostic application to FrIACs. The streamfunction shows short time-scale variations (order of days) as well as an overall increase up to day 26 (30 January) followed by a slow decline from day 30–60, indicating a weakening of the SWIRL. Further insight into these variations will be provided in section 5.

Fig. 7.
Fig. 7.

Time series plots of the (a) streamfunction minimum (black) and streamfunction at the AC edge as defined in Eq. (7) (red) and (b) ratio of the edge to minimum streamfunction values.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

As a verification that this approach can be applied to other potential SWIRLs, we calculated the AC edge for the P2 SWIRL at 1200 UTC 20 January 2020, the same date used for Fig. 3. The bounding box dimensions used are 90° in longitude and 20° in latitude, as indicated by the red dotted lines in Fig. 8. The algorithm captures closed streamlines from 485 to 625 K, coincident with positive PV anomalies. While we focus this paper on the P1 plume, Fig. 8 indicates the AC-edge calculation can be applied to other SWIRLs as well.

Fig. 8.
Fig. 8.

PV anomaly (filled color contours), streamlines (gray lines), AC edge (thick black lines), location of PV maximum (black dot), location of streamfunction minimum (black asterisk), and AC calculation box (dotted red lines) for 1200 UTC 20 Jan 2020 from 425 to 625 K in 20 K increments, centered on the P2 plume over Antarctica.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

d. Calculating SWIRL diagnostics

To describe the structure and behavior of SWIRLs, we calculate several other diagnostics over a θ range that encompasses the dynamical properties of the SWIRL. These diagnostics are calculated on 11 different θ surfaces centered on θP1(t), with a range extending from θbot(t) = θP1(t) − 100 K to θtop(t) = θP1(t) + 100 K in 20 K increments. Note that these vary with time due to changing position of the plume (Fig. 5a). On each of these levels we first calculate the T and O3 anomalies (relative to the zonal mean and referenced as T′ and O3) in order to examine the vertical temperature anomaly dipole structure and O3 signatures discussed in K20 and Khaykin et al. (2020). We also save the small-scale winds (u˜, υ˜) on these levels to examine how the circulation varies with height. The size of the SWIRL is next calculated by integrating the area A encompassed by ψ˜EDGE(θ,t) at each level. This area is converted to a length by first determining the equivalent latitude (e.g., Allen and Nakamura 2003), which represents the circle centered on the pole that encompasses the area A:
φeq=sin1(A2πa21).
The great circle distance across the circle is specified as the size (L) of the SWIRL. The flow speed (V) around the SWIRL is then determined by interpolating the small-scale wind speed u˜2+υ˜2 to the AC-edge contours and averaging the wind speed along the AC edge:
V(θ,t)ψ˜EDGEu˜2+υ˜2dsψ˜EDGEds,
where ds is an arc length along the contour of ψ˜EDGE(θ,t). We next calculate a vertical range for the SWIRL AC using three selection criteria: 1) a minimum threshold on qmax(θ,t) of 50%; 2) a minimum threshold on V of 5 m s−1; 3) a lower bound on θ equal to 380 K + 5.6 K day−1 × (tt0), where t is time in days and t0 is the initial time: 0600 UTC 4 January 2020. The first two criteria eliminate smaller circulation features that may contaminate the SWIRL identification, while the third criterion avoids the possibility of encountering other nonrelated features. These criteria provide the minimum and maximum theta levels θP1min(t) and θP1max(t) shown in Fig. 5a as well as the altitude bounds ZP1min(t) and ZP1max(t) in Fig. 5b. From the altitude bounds we also calculate the vertical thickness of the SWIRL as D(t)=ZP1max(t)ZP1min(t).

Figures 9a and 9b show curtain plots in time and θ for qmax and V calculated over the selected range. These plots illustrate why the lower bound criteria on θ are useful, particularly early on where PV and V anomalies not associated with the SWIRL are evident in the lower levels. Applying the selection criteria results in the black dots in Figs. 9a and 9b, which indicate the minimum and maximum levels for qmax and V that are identified with the SWIRL.

Fig. 9.
Fig. 9.

Curtain plots in time and potential temperature of (a) PV maximum and (b) flow speed following the P1 plume from 0600 UTC 4 Jan to 0600 UTC 3 Mar 2020. Black dots indicate minimum and maximum levels determined from the anticyclone selection criteria.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

We also calculate the large-scale (LS) winds using uLS=uu˜ and υLS=υυ˜ to determine the vertical shear in the ambient wind. The components of this shear, duLS/dZ and LS/dZ, are determined from the slope of a linear fit over the Z values associated with the selected SWIRL range. The magnitude of the shear of the large-scale vector wind is then determined from
|dVLSdZ|=[(duLSdZ)2+(dυLSdZ)2]1/2.

To diagnose the internal shape of the SWIRL, we define the tilt (τ) as the displacement vector pointing from the location of ψ˜min at θP1min(t) to the location of ψ˜min at θP1max(t). The magnitude of this vector τ is the great-circle distance between the two minima, while the orientation of the vector is the angle (α, in polar coordinates on a local Cartesian grid) with respect to the east–west direction using the horizontal line connecting the two positions. More details on these diagnostics are provided in section 5.

4. SWIRL evolution

a. Overview of horizontal and vertical structure

To provide an overview of the evolution of the P1 SWIRL, maps of PV anomaly and ψ˜ are provided in Fig. 10 at 5-day intervals from 0000 UTC 10 January to 0000 UTC 29 February 2020. The level θP1(t) is provided in the title of each plot, and the plots are centered on the position λP1(t), φP1(t), which is indicated by the black circle. On 10 January, the SWIRL is gaining strength, and closed streamlines surrounding the PV maximum are evident. The shape of the AC edge is somewhat elongated zonally, and the location of ψ˜min (indicated by an asterisk) is displaced slightly east of the PV minimum. Note that the bounding boxes for the AC-edge calculation are shown with red dotted lines on each plot of Fig. 10. The shape of the edge varies from nearly circular (e.g., 9 and 19 February) to more elliptical (e.g., 10 and 20 January), suggesting modification by the large-scale circulation. The PV anomaly shows a similar shape to the AC edge, and there is evidence of negative PV anomaly surrounding the edge of the SWIRL.

Fig. 10.
Fig. 10.

PV anomaly (filled color contours), streamlines (gray lines), AC edge (thick black lines), location of PV maximum (black dot), location of streamfunction minimum (black asterisk), and AC calculation box (dotted red lines) for 0000 UTC 10 Jan to 0000 UTC 29 Mar 2020 in 5-day increments. The potential temperature is indicated, which is the central value determined from the PV anomaly calculation.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

Vertical cross sections of q′, T′, and υ˜ in longitude and θ, through the latitude φP1(t) are provided in Fig. 11 for the same dates as Fig. 10. On 10–15 January, as the SWIRL center rises from 465 to 500 K, T′ and υ˜ become more coherent. These anomalies strengthen in magnitude and expand in depth over the second half of January. As shown in K20 and Khaykin et al. (2020), the evolution of dynamical anomalies is consistent with the vertical range of smoke and trace gas anomalies. The temperature anomaly is negative (positive) in the region above (below) the central point. During late February, υ˜ and T′ gradually weaken. On 29 February, T′ is mostly weaker than ±2 K, and υ˜<10m s1. The PV anomaly is still easy to identify, and this appears to be the last signature of the SWIRL to disappear. In K20, the PV anomaly was tracked out to 10 March 2020.

Fig. 11.
Fig. 11.

Cross sections in longitude and θ through the latitude of the P1 plume location in 5-day intervals, from 10 Jan to 29 Feb 2020. Date (latitude) are provided in the title of each panel. PV anomaly (filled color contours); temperature anomaly: 2, 4, 6 K (solid green), −2, −4, −6 K (dashed green); and wind contours: 5, 10, 15 m s−1 (solid black), −5, −10, −15 m s−1 (dashed black).

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

b. SWIRL diagnostics

We now analyze several of the diagnostics discussed in section 3d. First, T′ is shown using a time–θ curtain plot in Fig. 12a. For most of the time series, there is a strong vertical dipole with warmer (cooler) air below (above) the central location. The strength of the dipole is consistent with the T anomalies calculated from Microwave Limb Sounder (MLS) data by K20, who showed the maximum range from −9 to +9 K above and below the P1 anomaly during the second week of January, as well as GNSS RO measurements analyzed by Khaykin et al. (2020). The temperature dipole weakens after day 40, suggesting that the smoke heating diminishes with time. The large-scale zonal wind uLS is plotted in Fig. 12b. There is strong westward shear with altitude over the first several weeks, followed by weaker shear around day 30 when the P1 plume stalled over South America. The wind becomes consistently westward after day 40. Figures 12c and 12d plot the small-scale winds u˜ and υ˜, respectively. These tend to show a vertical dipole structure with alternating polarity. During the last 20 days, u˜ and υ˜ appear to oscillate back and forth suggesting a secondary rotational motion of the SWIRL, which will be examined in more detail in section 5.

Fig. 12.
Fig. 12.

Curtain plots in time and θ following the P1 plume from 0600 UTC 4 Jan to 0600 UTC 3 Mar 2020. (a) T anomaly (K), (b) large-scale zonal wind (m s−1), (c) small-scale zonal wind (m s−1), (d) small-scale meridional wind (m s−1), (e) O3 anomaly (ppmv). (f) Plume O3 (red) and zonal mean O3 (black) as a function of altitude following the plume, smoothed with a 2-day running mean to remove noise.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

Figure 12e shows low O3 values associated with the SWIRL. These highly unusual anomalies indicate that air from lower altitudes has been lofted into the middle stratosphere, as detailed in analysis of MLS O3 by K20. Further insight into the process is provided in Fig. 12f, which shows the O3 value in the center of the plume (red line) as a function of altitude, along with the zonal mean value O3¯ (black line) at the plume altitude and latitude. A 2-day running mean is applied to these curves to reduce noise. The plume O3 matches O3¯ up to 19 km, where the curves begin to separate. The plume maintains fairly constant O3 from 19 to 25 km, suggesting very little mixing is occurring as it rises over these altitudes. From 25 to 29 km, the plume O3 steadily increases with altitude until it is near the value of O3¯. While photochemical processes will become more important as the plume rises, and will need to be accounted for, it is likely that increased mixing is occurring at the upper levels. A complete analysis of mixing within the plume is beyond the scope of this paper, but the evidence shows a rather remarkable containment of air as it is lofted.

Figure 13 summarizes several SWIRL properties as line plots over the 2-month period. First, the maximum PV anomaly is provided in Fig. 13a. This generally increases over the first 30 days from ~50% to near 130%, and then declines after day 40. There is also an anomalous peak on day 45 of ~150%, with currently unknown cause. V at the central level is plotted in Fig. 13b. V increases gradually from ~5 m s−1 to a peak of 20 m s−1 around day 26 (30 January). The strengthening of V may be related to magnitude of the large-scale wind shear (Fig. 13c), which starts strong, ~2–3 m s−1 km−1, but weakens with time as P1 increases in altitude. Studies of the evolution of vortices in the presence of vertical shear (e.g., Jones 1995) show that vortices tend to weaken in the presence of vertical shear; therefore, as the shear decreases V increases. From day 26 to 60, V decreases gradually, likely due to the gradual weakening of diabatic heating within the smoke plume, but the exact mechanisms of the flow evolution remain to be explored. There is interesting variation on smaller time scales in both V and the shear, suggesting the possibility of wave interaction with the SWIRL dynamics. The SWIRL vertical thickness (D) and horizontal size (L) are provided in Figs. 13d and 13e, respectively. As with V, D increases over the first half of the time series and decreases over the second half. The peak value of D is ~6 km in early February, consistent with the CALIOP observations (Fig. 1), and overall D varies from ~2 to 6 km. L is initially ~1500 km, but decreases to ~900 km over the first 2 weeks. Then L is relatively steady over the rest of the time series, remaining larger than ~800 km throughout. The O3 is plotted in Fig. 13f, showing decline to about day 30, with values reaching −3 ppmv (consistent with Fig. 12f).

Fig. 13.
Fig. 13.

Time series plots of (a) PV anomaly maximum, (b) flow speed (V) at the central level, (c) magnitude of the vertical wind shear of the vector wind, (d) thickness (D), (e) size (L), (f) ozone anomaly, (g) minimum (black) and maximum (red) T anomaly, and (h) T difference between maximum and minimum anomalies.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

The temperature dipole can be inferred from line plots of the maximum and minimum T′ in Fig. 13g. The first few days show wide swings in maximum and minimum T′ as the SWIRL begins to take strength. The cause of these swings is uncertain, but may be related to strong tilting of the vortex early on as the SWIRL starts to spin up. The magnitude of the dipole (Fig. 13h) increases up to day 20, maximizing at ~15 K, before gradually declining from day 20–60. The decreasing temperature anomaly dipole is likely due to the slow dispersion of the smoke associated with the plume. Eventually, the aerosol concentration in the plume will become too small to support further rising via solar absorption. According to Fig. 5, there is rising motion at least through day 50, which is associated with temperature differences of <5 K. Detailed radiative transfer modeling is necessary to fully understand the plume behavior.

5. SWIRL maintenance in the presence of vertical shear

The overall SWIRL evolution appears to have a relatively simple description in terms of a “spinup” phase (day 1–26, or 4–30 January) in which V and D increase, and a “spindown” phase (day 27–50, 31 January–3 March) when these quantities decrease gradually. Superposed on this structure are variations on smaller time scales. We hesitate to speculate too much into the cause of these variations, particularly without detailed modeling that includes diabatic smoke heating, but here we discuss briefly one interesting suggestion related to the vertical tilt of the AC vortex that may help provide a framework for understanding SWIRL dynamics. In particular, we discuss a process that could help prevent the SWIRL from being sheared by the vertically varying wind. Note that a 2.5 m s−1 km−1 shear (approximate average over the first 15 days of the SWIRL, as seen in Fig. 13c) would cause a wind difference of 10 m s−1 over a vertical cylinder with thickness D = 4 km. A passively advected cylinder with L = 1000 km would therefore have a tilt (τ) equal to L after 100 000 s, or ~1.16 days. The longevity of the SWIRL suggests other processes are involved in its maintenance. While diabatic heating undoubtedly plays a major role, here we explore internal dynamical mechanisms in a vortex that can act to resist the tilting effects of vertical shear.

Jones (1995) modeled tropical cyclone behavior using a primitive equation numerical simulation on an f plane without diabatic heating, and found that as vortices tilt due to vertically sheared environmental (or background) winds the upper and lower centers begin to rotate about the midlevel center. This is caused by a secondary circulation resulting from the horizontal displacement of upper and lower level PV anomalies. As illustrated later, this rotation tends to oppose the tendency of the vertical shear to destroy the vortex. A key part of the argument is that PV anomalies can exert a vertical influence on the air above and below. This influence reaches a vertical penetration depth, which is approximated by fL/N, where f is the Coriolis parameter, N is the square root of the static stability, and L is the horizontal scale of the vortex. With typical values for a SWIRL: L ~ 1000 km, φ = 40°S, and N ~ 2 × 10−2 s−1 in the stratosphere (e.g., Grise et al. 2010) the penetration depth is ~5 km, similar to the thickness of the SWIRL. This suggests the Jones (1995) mechanism could apply to the SWIRL.

As is shown in Fig. 12b, the vortex initially experiences an environmental wind with an easterly vertical shear. In Fig. 14, we illustrate a simplified two-level model of the vortex under these conditions. The environmental shear will cause the vortex to initially start tilting in the east–west direction as shown in Fig. 14a, with the lower level displaced east of the upper level. Due to vertical penetration of the anticyclonic anomalies, the upper level vortex will induce a northward advection on the lower level anomaly, while the lower level vortex will induce a southward advection on the upper level anomaly. This will cause secondary displacement as shown in Fig. 14b. Further mutual action of the vortices on each other will result in a north–south orientation as seen in Fig. 14c. This illustrates the first stabilizing mechanism in which the environmental shear is opposed to the mutual advection of the vortices. Further advection causes a NW–SE tilt as shown in Fig. 14d. Now the vortex tilt in the E–W direction has a zonal component that is opposite to the tilting tendency of the environmental wind, a second mechanism that stabilizes the vortex. These two mechanisms (further explained in Jones 1995) may at least partly explain the stability of the SWIRL. However, detailed numerical modeling studies, which include the effects of diabatic heating, are needed for a more complete understanding.

Fig. 14.
Fig. 14.

Schematic diagram showing the initial tilting and subsequent evolution of an AC vortex placed in an environmental vertical shear in the SH. The vortex is divided into two parts: a lower-level AC (blue circle) and an upper-level AC (red circle). The environmental wind has westerlies at lower levels and easterlies at upper levels, as indicated by the block arrows. The tilt is defined as a vector connecting the lower- and upper-level centers (black arrows), with magnitude τ and angle α = 0° (180°) pointing east (west). Panels show successive evolution from (a) initial shear causing tilt angle (α) of 180° after which secondary rotation (in direction of colored arrows) begins and continues through tilt angles of (b) 225°, (c) 270°, and (d) 315°.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

We examine whether this type of motion may be occurring in the P1 SWIRL by looking at the tilt in terms of the position of the minimum streamfunction λψ˜min(t),φψ˜min(t) at the top and bottom of the identified SWIRL region. In Fig. 6, we see that the center of the bottom of the SWIRL (555 K) is coincident with the PV maximum, while the top of the SWIRL (695 K) is displaced slightly to the south of the PV maximum. This causes a negative latitude difference with height, as illustrated in Fig. 15a. The tilt angle is southward and increasing over most of the first 30 days, but during the middle and latter half of the time series, there appears to be a sinusoidal variation in tilt orientation, which is likely related to the alternating polarity of the small-scale wind anomalies seen in Figs. 12c and 12d. The angle (α) made between the top and bottom positions, plotted in Fig. 15b, is in the range of 180°–360° over most of the first 30 days, again indicating southward tilt with height (cf. with Fig. 14). This orientation may be helping to stabilize the SWIRL over the first several weeks.

Fig. 15.
Fig. 15.

Time series plots of the (a) difference between the longitude (black) and latitude (red) of ψ˜min at the top and bottom of the SWIRL and (b) angle (α) between the location of the ψ˜min at the top and bottom of the SWIRL. Blue lines are linear least squares fits to the data from day 41–49 and day 50–59, which have slopes of −34° and −30° day−1, respectively. (c) Times series of magnitude of the tilt (τ) between the top and bottom of the SWIRL.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

Later in the evolution, the behavior becomes more complicated; α shows a fluctuating behavior from day 25–40, followed by a more coherent approximately linear change with time orientation from day 40–60. The latter period has a decreasing α with time, which indicates counterclockwise (anticyclonic) rotation. Jones (1995) showed that the rotation rate is sensitive to various parameters such as the vertical shear, strength and size of the vortex, static stability, and Coriolis parameter. Several of these factors are clearly changing as the SWIRL moves vertically and horizontally throughout the stratosphere. Figure 15b shows that the SWIRL orientation makes ~1.5 anticyclonic rotations from days 40–60. Blue lines in Fig. 15b are linear least squares fits to the data from day 41–49 and day 50–59, which have slopes of −34° and −30° day−1, respectively.

The tilt magnitude τ is shown in Fig. 15c. The time evolution of τ is complicated, but it appears that there is a delicate interplay between τ and the large-scale wind shear (Fig. 13c), which both show oscillations with period of ~10 days. Beyond day 40, τ increases as the top and bottom parts of the AC start to rotate around each other. Detailed numerical modeling studies would be needed to completely understand the complex dynamics of the SWIRL, including the diabatic heating due to the absorbing smoke aerosol, but the idea of vortex interactions with the environmental conditions provides a useful framework for examining the complicated second-order behavior.

6. Discussion

The ANY event P1 plume was so large that the dynamical signature was easy to detect. Smaller-scale plumes associated with this event (e.g., P2 and P3 identified by K20) also showed PV anomalies, and their complete dynamical properties need to be further studied. This work also suggests that SWIRLs may have been present in past pyroCb events as well, such as the Black Saturday plumes in February 2009 (Pumphrey et al. 2011), the PNE in August 2017 (Peterson et al. 2018), and other smaller events (Fromm et al. 2010). Dynamical features in some events may be obscured due to the transient nature of the events and the presence of other synoptic-scale weather patterns. The P1 plume extended high into the stratosphere, where wind shears were relatively weak, and the SWIRL lasted for at least 2 months. Plumes at lower altitudes, or in more complicated environmental circulations, may require more sophisticated filtering algorithms to determine their dynamical signatures.

As an example, we applied the SWIRL detection and analysis algorithm to one of the plumes associated with the PNE as it passed over Europe on 28 August 2017. For this plume we use MERRA-2 interpolated to the NAVGEM T119 grid, but otherwise follow the same algorithms discussed in section 3. Due to the smaller size of this plume, we used filtering of total wavenumber 20 (rather than 10 used for P1) and a box size of 10° longitude × 5° latitude (rather than 20° × 10°). While the streamlines are not circular, the diagnostics still apply as long as the streamlines are closed at the AC edge. The resulting PV anomaly and streamlines are provided in Fig. 16, and these indicate that PNE did indeed produce at least one SWIRL. Flow speeds for this event were calculated at ~2–3 m s−1, and the horizontal size was ~600 km. Further analysis of this and other PNE plumes is ongoing.

Fig. 16.
Fig. 16.

PV anomaly (colored filled contours), streamlines (gray lines), AC edge (thick black lines), location of PV maximum (black dot), location of streamfunction minimum (black asterisk), and AC calculation box (dotted red lines) for 0000 UTC 28 Aug 2017 from 420 to 520 K in 10 K increments, centered on one of the PNE plumes over Europe.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

We showed that NAVGEM analyses were able to resolve synoptic-scale rotation as well as lofting consistent with independent observational analyses of constituents (K20). These analyses benefit from assimilation of observed temperatures and radiances that help to maintain the dynamics (see also discussion in Khaykin et al. 2020). However, it is likely that forecasts with numerical weather prediction models, which do not have smoke-related diabatic heating, will fail to maintain SWIRLs. As an initial test, we ran a 10-day NAVGEM forecast initialized on 10 January 2020 and found that the PV anomaly associated with the P1 SWIRL diminished significantly within ~4–6 days (see Fig. 17), and there was no apparent lofting over the forecast period. More detailed forecast comparisons are planned for future work, but the initial test suggests that large stratospheric forecast errors may occur in regions of strong smoke concentrations. These NAVGEM forecast results are consistent with a series of 10-day ECMWF forecasts examined by Khaykin et al. (2020). For nearly all forecasts of the P1 SWIRL between 19 January and 28 February, the vortex did not show the observed ascent, and the forecast predicted a rapid decay of vortex intensity as identified by relative vorticity. These discrepancies between forecasts and analyses are almost certainly due to the lack of aerosol heating in the forecast model.

Fig. 17.
Fig. 17.

SH PV anomaly following the central θ level (indicated in panel titles) of the P1 plume. NAVGEM analyses (all at 0000 UTC) are provided for (a) 10, (b) 12, (c) 14, and (d) 16 Jan 2020. NAVGEM forecasts (initialized at 0000 UTC 10 Jan) of PV anomaly are provided for (e) 10, (f) 12, (g) 14, and (h) 16 Jan 2020.

Citation: Journal of the Atmospheric Sciences 77, 12; 10.1175/JAS-D-20-0131.1

7. Summary

Smoke-induced stratospheric dynamics is a recently discovered phenomenon, which opens up a brand new area of scientific research. In this paper, we quantified the anomalies in PV, wind, temperature, and ozone following the P1 SWIRL over the main part of its life cycle using NAVGEM analyses from 4 January to 3 March 2020. The plume rose rapidly over this 2-month period, from 15 to 30 km, and nearly circumnavigated the globe. Using diagnostic tools originally developed for understanding stratospheric anticyclones, we found the P1 SWIRL to have PV anomalies of up to ~130%, rotation rates up to ~15 m s−1, and dimensions of ~1000 km in the horizontal by ~6 km in the vertical. Vertical temperature differences of up to ~15 K were found as well as ozone anomalies of up to ~3 ppmv. In addition to overall growth (decay) of PV anomalies and flow over the first (second) month of its lifetime, significant variations on shorter time scales were observed, which we attempted to explain with dynamics associated with tilting vortices. Evidence of a SWIRL was also provided for the PNE in August 2017, opening to door for further research into plumes in past pyroCb events. Finally, we examined a NAVGEM forecast of the P1 plume, which showed no lofting and significant decay of potential vorticity over ~5 days, suggesting the lack of smoke-induced heating can cause large forecast errors into the stratosphere.

8. Outlook

Although work done by K20, Khaykin et al. (2020), and this paper has revealed many properties of SWIRLs, there are still numerous unanswered questions. While future SWIRL research may take many different directions, here we outline a few areas that could be explored.

One research direction is to understand the radiative processes that produce the combined vertical temperature dipole, lofting, and rotation. It is obvious that smoke-induced heating is the ultimate cause, but many details remain to be discovered. Radiative calculations combined with observational data will likely provide fruitful insight into SWIRL dynamics and thermodynamics. Also, dynamical modeling studies are necessary to ascertain the link between the rotation and the vertical dipole and how the rotation is maintained as the SWIRL ascends.

Another question is how the SWIRL acts as a containment vessel of large aerosol and trace gas anomalies as it moves laterally and vertically. While it is well known that PV gradients occur in the presence of mixing barriers for large-scale features such as the winter stratospheric polar vortex and the Asian summer monsoon, it is not as obvious that PV gradients associated with a small-scale feature such as a SWIRL can significantly inhibit mixing. Kinematic studies could shed significant light on this phenomenon, including how the smoke and trace gas anomalies eventually mix with the surrounding air. Photochemical modeling would also be useful for understanding chemically active species such as ozone as they are transported vertically and horizontally over the SWIRL life cycle.

A third area of research involves the secondary dynamical effects. We provided some possibilities for further work related to the behavior of vortices in vertical shear. Other possibilities for causing shorter-time-scale oscillations may include Rossby wave and/or gravity wave interactions with the vortex. In this paper we found evidence of possible orographic gravity wave modulation of the P1 plume as it encountered the Andes. This work focused on the middle part of the life cycle while the SWIRL is relatively stable. Work is also needed to understand the spinup and breakdown of the SWIRL and its relation to the peculiar radiative qualities of the smoke plume.

Finally, more detailed studies are necessary to discern whether forecast errors due to neglect of stratospheric smoke heating may also impact tropospheric forecasts. This would likely occur via radiative impacts, which can extend down to the surface (Khaykin et al. 2020), but there may also be downward penetrating dynamical effects. It may therefore be important to develop parameterizations of smoke-heating using radiative transfer models in order for weather models to maintain the SWIRL circulation over the nominal 2-week operational forecast. We have focused on SWIRLs that are well above the tropopause. It is possible that SWIRLs will develop in the upper troposphere as well. However, the details of tropospheric SWIRLs will be likely more difficult to discern than stratospheric SWIRLs due to more complicated background meteorology.

Acknowledgments

This work was funded by the U.S. Office of Naval Research (ONR) and by the NASA Aura Science Team (NNH19ZDA001N-AURAST) project “Studying UTLS Impacts of PyroCb with Aura MLS, the A-Train, and Radiative Transfer Models” (19-AURAST19-0057). Near-real-time NAVGEM-HA results were provided by the support of the ONR project “Middle Atmosphere Numerical Weather Prediction.” NAVGEM-HA analyses and forecasts were produced under a grant of computer time from the Department of Defense High Performance Computing Modernization Program. This work also benefitted from the helpful comments and suggestions by three anonymous reviewers.

Data availability statement

CALIOP data are available at https://www-calipso.larc.nasa.gov/products and OMPS level 3 UVAI data are available at https://ozoneaq.gsfc.nasa.gov/data/omps/. MERRA-2 data are available at the NASA Goddard Earth Sciences (GES) Data and Information Services Center (DISC), https://disc.gsfc.nasa.gov/. The NAVGEM data used in this study have not yet been released by the Department of Defense and are therefore not publicly available at this time.

REFERENCES

  • Allen, D. R., and N. Nakamura, 2003: Tracer equivalent latitude: A diagnostic tool for isentropic transport studies. J. Atmos. Sci., 60, 287304, https://doi.org/10.1175/1520-0469(2003)060<0287:TELADT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Allen, D. R., J. L. Stanford, L. S. Elson, E. F. Fishbein, L. Froidevaux, and J. W. Waters, 1997: The 4-day wave as observed from the Upper Atmosphere Research Satellite Microwave Limb Sounder. J. Atmos. Sci., 54, 420434, https://doi.org/10.1175/1520-0469(1997)054<0420:TDWAOF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Allen, D. R., A. R. Douglass, and S. E. Strahan, 2013: The large-scale frozen-in anticyclone in the 2011 Arctic summer stratosphere. J. Geophys. Res. Atmos., 118, 26562672, https://doi.org/10.1002/JGRD.50256.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Andrews, D., C. Leovy, and J. Holton, 1987: Middle Atmosphere Dynamics Academic Press, 502 pp.

  • Australian Government, 2019: Dangerous bushfire weather in spring 2019. Australian Government Special Climate Statement 72, 28 pp., http://www.bom.gov.au/climate/current/statements/scs72.pdf.

  • Bishop, C. H., and A. J. Thorpe, 1994: Potential vorticity and the electrostatics analogy: Quasi-geostrophic theory. Quart. J. Roy. Meteor. Soc., 120, 713731, https://doi.org/10.1002/qj.49712051710.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bonavita, M., L. Isaksen, and E. Hólm, 2012: On the use of EDA background error variances in the ECMWF 4D-Var. ECMWF Tech. Memo. 664, 33 pp., https://www.ecmwf.int/en/elibrary/8272-use-eda-background-error-variances-ecmwf-4d-var.

    • Crossref
    • Export Citation
  • Boone, C. D., P. F. Bernath, and M. D. Fromm, 2020: Pyrocumulonimbus stratospheric plume injections measured by the ACE-FTS. Geophys. Res. Lett., 47, e2020GL088442, https://doi.org/10.1029/2020GL088442.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Buehner, M., P. L. Houtekamer, C. Charette, H. L. Mitchell, and B. He, 2010: Intercomparison of variational data assimilation and the ensemble Kalman filter for global deterministic NWP. Part II: One-month experiments with real observations. Mon. Wea. Rev., 138, 15671586, https://doi.org/10.1175/2009MWR3158.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Clayton, A. M., A. C. Lorenc, and D. M. Barker, 2013: Operational implementation of a hybrid ensemble/4D-Var global data assimilation system at the Met Office. Quart. J. Roy. Meteor. Soc., 139, 14451461, https://doi.org/10.1002/qj.2054.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Daley, R., 1991: Atmospheric Data Analysis. Cambridge University Press, 457 pp.

  • Daley, R., and E. Barker, 2001: NAVDAS Source Book 2001: NRL Atmospheric Variational Data Assimilation System. NRL Publ. NRL/PU/7530—01-441, 163 pp., http://www.dtic.mil/docs/citations/ADA396883.

  • Eckermann, S. D., and Coauthors, 2009: High-altitude data assimilation system experiments for the northern summer mesosphere season of 2007. J. Atmos. Sol.-Terr. Phys., 71, 531551, https://doi.org/10.1016/j.jastp.2008.09.036.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fairlie, T. D., 1995: Three-dimensional transport simulations of the dispersal of volcanic aerosol from Mount Pinatubo. Quart. J. Roy. Meteor. Soc., 121, 19431980, https://doi.org/10.1002/qj.49712152809.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fairlie, T. D., R. B. Pierce, J. A. Al-Saadi, W. L. Grose, J. M. Russell, M. H. Proffitt, and C. R. Webster, 1999: The contribution of mixing in Lagrangian photochemical predictions of polar ozone loss over the Arctic in summer 1997. J. Geophys. Res., 104, 26 59726 609, https://doi.org/10.1029/1999JD900111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fromm, M., D. T. Lindsey, R. Sevranckx, G. Yue, T. Tricki, R. Sica, P. Doucet, and S. Godin-Beekman, 2010: The untold story of pyrocumulonimbus. Bull. Amer. Meteor. Soc., 91, 11931210, https://doi.org/10.1175/2010BAMS3004.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 54195454, https://doi.org/10.1175/JCLI-D-16-0758.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grise, K. M., D. W. J. Thompson, and T. Birner, 2010: A global survey of static stability in the stratosphere and upper troposphere. J. Climate, 23, 22752292, https://doi.org/10.1175/2009JCLI3369.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harvey, V. L., R. B. Pierce, T. D. Fairlie, and M. H. Hitchman, 2002: A climatology of stratospheric polar vortices and anticyclones. J. Geophys. Res., 107, 4442, https://doi.org/10.1029/2001JD001471.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harvey, V. L., R. B. Pierce, M. H. Hitchman, C. E. Randall, and T. D. Fairlie, 2004: On the distribution of ozone in stratospheric anticyclones. J. Geophys. Res., 109, D24308, https://doi.org/10.1029/2004JD004992.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harvey, V. L., C. E. Randall, G. L. Manney, and C. S. Singleton, 2008: Low-ozone pockets observed by EOS-MLS. J. Geophys. Res., 113, D17112, https://doi.org/10.1029/2007JD009181.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Haynes, P. H., 1990: High-resolution three-dimensional modeling of stratospheric flows: Quasi-two-dimensional turbulence dominated by a single vortex. Topological Fluid Mechanics, H. K. Moffat and A. Tsinober, Eds., Cambridge University Press, 345–354.

  • Hogan, T., and Coauthors, 2014: The Navy Global Environmental Model. Oceanography, 27 (3), 116125, https://doi.org/10.5670/oceanog.2014.73.

  • Holton, J. R., 2004: An Introduction to Dynamic Meteorology. Vol. 1. Elsevier Academic Press, 535 pp.

  • Hooghiem, J. D., M. E. Popa, T. Röckmann, J.-U. Grooß, I. Tritscher, R. Müller, R. Kivi, and H. Chen, 2020: Wildfire smoke in the lower stratosphere identified by in situ CO observations. Atmos. Chem. Phys. Discuss., https://doi.org/10.5194/acp-2020-65, in press.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jones, S. C., 1995: The evolution of vortices in vertical shear. 1. Initially barotropic vortices. Quart. J. Roy. Meteor. Soc., 121, 821851, https://doi.org/10.1002/qj.49712152406.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kablick, G. P., D. R. Allen, M. D. Fromm, and G. E. Nedoluha, 2020: Australian pyroCb smoke generates synoptic-scale stratospheric anticyclones. Geophys. Res. Lett., 47, e2020GL088101, https://doi.org/10.1029/2020GL088101.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khaykin, S., S. Godin-Beekmann, A. Hauchecorne, J. Pelon, F. Ravetta, and P. Keckhut, 2018: Stratospheric smoke with unprecedentedly high backscatter observed by lidars above southern France. Geophys. Res. Lett., 45, 16391646, https://doi.org/10.1002/2017GL076763.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khaykin, S., and Coauthors, 2020: The 2019/20 Australian wildfires generated a persistent smoke-charged vortex rising up to 35 km altitude. Commun. Earth Environ., 1, 22, https://doi.org/10.1038/s43247-020-00022-5.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kuhl, D. D., T. E. Rosmond, C. H. Bishop, J. McLay, and N. L. Baker, 2013: Comparison of hybrid ensemble/4DVar and 4DVar within the NAVDAS-AR data assimilation framework. Mon. Wea. Rev., 141, 27402758, https://doi.org/10.1175/MWR-D-12-00182.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Laprise, R., 1992: The resolution of global spectral models. Bull. Amer. Meteor. Soc., 73, 14531455, https://doi.org/10.1175/1520-0477-73.9.1453.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Manney, G. L., N. J. Livesey, C. J. Jimenez, H. C. Pumphrey, M. L. Santee, I. A. MacKenzie, and J. W. Waters, 2006: EOS Microwave Limb Sounder observations of “frozen-in” anticyclonic air in Arctic summer. Geophys. Res. Lett., 33, L06810, https://doi.org/10.1029/2005GL025418.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCormack, J., and Coauthors, 2017: Comparison of mesospheric winds from a high-altitude meteorological analysis system and meteor radar observations during the boreal winters of 2009–2010 and 2012–2013. J. Atmos. Sol.-Terr. Phys, 154, 132166, https://doi.org/10.1016/j.jastp.2016.12.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McLay, J. G., C. H. Bishop, and C. A. Reynolds, 2008: Evaluation of the ensemble transform analysis perturbation scheme at NRL. Mon. Wea. Rev., 136, 10931108, https://doi.org/10.1175/2007MWR2010.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ohneiser, K., and Coauthors, 2020: Smoke of extreme Australian bushfires observed in the stratosphere over Punta Arenas, Chile, in January 2020: Optical thickness, lidar ratios, and depolarization ratios at 355 and 532 nm. Atmos. Chem. Phys., 20, 80038015, https://doi.org/10.5194/acp-20-8003-2020.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Orsolini, Y. J., 2001: Long-lived tracer patterns in the summer polar stratosphere. Geophys. Res. Lett., 28, 38553858, https://doi.org/10.1029/2001GL013103.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peterson, D., J. Campbell, E. Hyer, M. Fromm, G. Kablick, J. Cossuth, and M. DeLand, 2018: Wildfire-driven thunderstorms cause a volcano-like stratospheric injection of smoke. npj Climate Atmos. Sci., 1, 30, https://doi.org/10.1038/S41612-018-0039-3.

    • Search Google Scholar
    • Export Citation
  • Peterson, D., E. Hyer, J. Campbell, M. Fromm, C. Bennese, M. Berman, and T. Van, 2019: Quantifying the impact of intense pyroconvection on stratospheric aerosol loading. 2019 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, Abstract GC11F-1150, https://agu.confex.com/agu/fm19/meetingapp.cgi/Paper/510480.

  • Pumphrey, H. C., M. L. Santee, N. J. Livesy, M. J. Schwartz, and W. G. Read, 2011: Microwave Limb Sounder observations of biomass-burning products from the Australian bush fires of February 2009. Atmos. Chem. Phys., 11, 62856296, https://doi.org/10.5194/acp-11-6285-2011.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rosmond, T., and L. Xu, 2006: Development of NAVDAS-AR: Non-linear formulation and outer loop tests. Tellus, 58A, 4558, https://doi.org/10.1111/j.1600-0870.2006.00148.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thiéblemont, R., N. Huret, Y. J. Orsolini, A. Hauchecorne, and M.-A. Drouin, 2011: Frozen-in anticyclones occurring in polar Northern Hemisphere during springtime: Characterization, occurrence and link with quasi-biennial oscillation. J. Geophys. Res., 116, D20110, https://doi.org/10.1029/2011JD016042.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Torres, O., 2019: OMPS-NPP L2 NM aerosol index swath orbital, version 2. Goddard Earth Sciences Data and Information Services Center, accessed February 2020, https://doi.org/10.5067/40L92G8144IV.

    • Crossref
    • Export Citation
  • Xu, L., T. Rosmond, and R. Daley, 2005: Development of NAVDAS-AR: Formulation and initial tests of the linear problem. Tellus, 57A, 546559, https://doi.org/10.3402/tellusa.v57i4.14710.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, P., and Coauthors, 2019: Black carbon lofts wildfire smoke high into the stratosphere to form a persistent plume. Science, 365, 587590, https://doi.org/10.1126/science.aax1748.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save