Gravity wave in uences on mesoscale divergence:...
Transcript of Gravity wave in uences on mesoscale divergence:...
manuscript submitted to Geophysical Research Letters
Gravity wave influences on mesoscale divergence: An1
observational case study2
C. C. Stephan1, T. P. Lane2,4, C. Jakob3,43
1Max Planck Institute for Meteorology, Hamburg, Germany42School of Earth Sciences, The University of Melbourne, Melbourne, Australia5
3School of Earth, Atmosphere and Environment, Monash University, Melbourne, Australia64ARC Centre of Excellence for Climate Extremes, Sydney, Australia7
Key Points:8
• 1. Mesoscale divergence profiles near Darwin show similar long-lived fine-scale ver-9
tical structure to that reported over the tropical Atlantic.10
• 2. High-resolution soundings are used to test if gravity waves can serve as a plau-11
sible explanation for the observed divergence variability.12
• 3. Results imply that mesoscale patterns of horizontal divergence may follow scal-13
ing laws that are determined by gravity wave properties.14
Corresponding author: Claudia C. Stephan, [email protected]
–1–
manuscript submitted to Geophysical Research Letters
Abstract15
Characteristics of tropospheric low-frequency gravity waves are diagnosed in radiosonde16
soundings from the Tropical Warm Pool-International Cloud Experiment (TWP-ICE)17
near Darwin, Australia. The waves have typical vertical wavelengths of 3 - 4 km, hor-18
izontal wavelengths of 200 - 600 km and intrinsic periods of 8 - 12 hours. These scales19
match those of the vertical, horizontal and temporal variability found in area-averaged20
horizontal wind divergence over the same domain. Vertical profiles of divergence show21
wave-like structures with variability of the order of 2×10−5s−1 in the free troposphere.22
The results for Darwin help explain previously reported observed mesoscale patterns of23
divergence/convergence over the tropical Atlantic. The findings imply that tropical di-24
vergence on spatial scales of a few hundred kilometers, which are typical scales of orga-25
nized convection, may follow universal scaling laws that are determined by a background26
gravity wave forcing.27
Plain Language Summary28
On horizontal scales of several tens to hundreds of kilometers, which we call ”mesoscale”,29
mean vertical motion is very small compared to mean horizontal motion. Yet, the ver-30
tical motion exerts a critical influence on the formation of clouds: large-scale descent is31
associated with fair weather, ascent is associated with cloudiness. Weather and climate32
modeling often assumes that mesoscale vertical motion varies slowly in the three spa-33
tial dimensions and with time. Recent observations over the tropical Atlantic, however,34
showed strong variability in mesoscale vertical motion, implying that clouds do not only35
respond to vertical motion but may themselves trigger vertical motion in their vicinity.36
This study reports similar variability also near Darwin, Australia. One way in which clouds37
can trigger remote vertical motion is by emitting waves, similar to stones that are thrown38
into a pond. This study examines vertical profiles of horizontal wind speed that were mea-39
sured by instruments on ascending balloons near Darwin, Australia. These observations40
do indeed show waves that can provide a plausible explanation for the patterns of note-41
worthy variability in mesoscale motions. These findings suggest a two-way coupling of42
clouds to their environment with potentially important consequences for our understand-43
ing of weather and climate phenomena.44
1 Introduction45
Although gravity waves (GWs) are generated mainly in the troposphere, most re-46
search has been concerned with their effects on the middle and upper atmosphere, where47
they are key drivers of the large-scale circulation (e.g. , J. R. Holton, 1982; J. Holton,48
1983; Alexander et al., 2010). Yet, it has been known for many decades that GWs also49
constitute an important component of tropospheric dynamics across a range of spatial50
and temporal scales. For instance, GWs can promote the growth of shallow convection51
into long-lived convective bands (Shige & Satomura, 2001), initiate severe convection (Balaji52
& Clark, 1988; Su & Zhai, 2017), induce quasi-regularly spaced cloud-systems (Lane &53
Zhang, 2011), or provide a mechanism for the longevity of convective systems (Stechmann54
& Majda, 2009; Lane & Zhang, 2011). Diabatic heating from condensation is an efficient55
source of GWs, which may in turn suppress, favor or trigger remote convection by in-56
ducing large scale rising or sinking motion, respectively (e.g., Mapes, 1993; Stephan et57
al., 2016). In this sense, clouds are sources as well as sinks of wave energy and able to58
remotely condition the surrounding environment for subsequent convection.59
Here, we diagnose the characteristics of low-frequency GWs from radiosonde ob-60
servations in the troposphere, taken during the Tropical Warm Pool-International Cloud61
Experiment (TWP-ICE) that took place in Darwin, Australia (130◦E, 12◦S) from Jan62
17 to Feb 13, 2006 (May et al., 2008). Our study is motivated by more recent observa-63
tions over the tropical Atlantic near Barbados (Bony & Stevens, 2019). They diagnosed64
–2–
manuscript submitted to Geophysical Research Letters
vertical profiles of divergence from dropsonde measurements of horizontal wind veloc-65
ities. The sondes were released along circular flight paths of 180 km diameter. The re-66
sulting vertical profiles of area-averaged divergence showed a noteworthy wave-like struc-67
ture (i.e., divergence and convergence) with magnitudes of variability of around 2×10−5s−168
in the free troposphere. This is four times the climatological divergence required to bal-69
ance radiative cooling. Furthermore, Bony and Stevens (2019) analyzed high-resolution70
nested simulations that were nudged to the observed conditions. The simulations pro-71
duced autocorrelation scales of the divergence in space and time of the order of 200 km72
and 4 h, respectively. Of the potential explanations for the observed wave-like vertical73
structure of the mesoscale divergence profiles and the simulated scaling laws, GWs are74
one potential candidate.75
The TWP-ICE data set provides an excellent opportunity to investigate the influ-76
ences of GWs on vertical profiles of mesoscale divergence. During the campaign 3-hourly77
radiosondes were launched from five stations which were situated around an area of ap-78
proximately 200 km diameter (Fig. 1). Thus, the experimental domain is of similar size79
as the circular regions studied by Bony and Stevens (2019), and the soundings are of a80
sufficiently high temporal resolution to sample the relevant time scales. We will provide81
evidence that i) vertical profiles of divergence near Darwin exhibit similar structures as82
those observed near Barbados, ii) the previously reported autocorrelation timescales in-83
ferred from modeling can also be found in observations, and iii) low-frequency gravity84
waves can serve as a potential explanation for the vertical, horizontal and temporal struc-85
tures of mesoscale divergence patterns.86
Section 2 introduces the data from TWP-ICE that are analyzed for the properties87
of divergence and GWs in Section 3. The results are discussed and summarized in Sec-88
tion 4.89
2 Data90
The following is a brief summary of the synoptic and convective environment dur-91
ing the TWP-ICE campaign. More information can be found in May et al. (2008), where92
the TWP-ICE field experiment is described in detail. The campaign began with an ac-93
tive monsoon period that was characterized by rainfall rates of 17 mm d−1 averaged over94
the 300-km-diameter area sampled with a polarimetric radar at Darwin. The degree of95
convective organization differed widely and a major mesoscale convective system devel-96
oped on Jan 23. A suppressed monsoon period followed from Jan 28 to Feb 6 before deep97
convection featuring intense afternoon thunderstorms and some squall lines commenced98
again. The suppressed period was characterized by surface westerly winds and mid-level99
advection of dry continental air, which limited convection to isolated cells or short lines100
with only few deep single cells reaching up to 10 km. The average rain rates were 5 - 7101
mm d−1 and originated from relatively shallow convection. The three days following Feb102
3 were completely clear with no rain. We will focus on this 10-day suppressed period Jan103
28 to Feb 6, as it presents the most undisturbed conditions in the troposphere, allow-104
ing the extraction of wave signatures from tropospheric wind perturbations that at other105
times would be obfuscated by convection or turbulence. Moreover, the atmospheric con-106
ditions during the suppressed period are similar to the environment sampled by the cir-107
cle flights analyzed in Bony and Stevens (2019), where intermittent shallow cumulus con-108
vection was observed.109
Area-averaged divergence is taken from variational analysis data. By following the110
objective method of Zhang and Lin (1997), Xie et al. (2010) created a variational anal-111
ysis for the TWP-ICE domain, which we will refer to as the large domain, or ”L-Domain”,112
as well as separate variational analyses for subdomains (Fig. 1a). The variational anal-113
ysis technique combines a background from the European Centre for Medium-Range Weather114
Forecasts (ECMWF) analysis with constraints from satellite, surface, and radiosonde mea-115
–3–
manuscript submitted to Geophysical Research Letters
surements taken during TWP-ICE. It includes as boundary forcing the observations from116
radiosondes launched inside the respective domains. The technique applies small per-117
turbations to the observed profiles of temperature, wind, and humidity, while conserv-118
ing column-integrated mass, moisture, energy, and momentum. In this way the area-averaged119
time-evolving vertical profiles of thermodynamic variables are obtained. In addition, Davies120
et al. (2013) repeated the variational analysis for L-Domain based on ECMWF data only121
(”L-Domain-EC”). Instead of ingesting radiosonde measurements, they replaced them122
with vertical profiles from ECMWF grid points near the respective radiosonde launch123
locations.124
Measurements from 3-hourly radiosonde soundings at five stations (Ship, Cape Don,125
Garden Point, Mount Bundy, and Point Stuart; Fig. 1a,b) are analyzed for GW signa-126
tures. Winds, air temperature, mixing ratio, pressure and the altitude of the balloons127
were recorded every 2 s. As in Hankinson et al. (2014b), we use area-averaged vertical128
profiles of U and V from the variational analysis on L-Domain to derive perturbations129
u′ and v′ from radiosonde measurements, which are then analyzed for GWs (Fig. 1c).130
Furthermore, we intercompare divergence profiles from variational analyses on different131
domains to test how their characteristics vary in space and with domain size and bound-132
ary forcing (i.e., L-Domain versus L-Domain-EC).133
Several studies have shown that a rich spectrum of stratospheric GWs existed dur-134
ing the suppressed monsoon period of TWP-ICE. For example, Evan and Alexander (2008)135
identified southeastward-propagating inertia-GWs in radiosonde observations with ver-136
tical wavelengths λz ∼ 6 km, horizontal wavelengths λh ∼ 5500 - 7000 km and peri-137
ods τ ∼ 2 days. A subsequent modeling study confirmed that convection near Indone-138
sia likely generated these stratospheric waves (Evan et al., 2012). Satellite images showed139
waves with λh =200 - 400 km near an altitude of 40 km (Hecht et al., 2009). Air glow140
images revealed both waves with τ =1 - 2 h and τ =15 - 25 min, λh =30-40 km at 80141
km altitude (Hecht et al., 2009). Hankinson et al. (2014b) and Hankinson et al. (2014a)142
analyzed radiosonde measurements for low and high-frequency GWs in the stratosphere,143
respectively. During the suppressed monsoon period they detected inertia-GWs with ori-144
gins near Indonesia in the middle stratosphere (λz =5 - 6 km, λh =3000 - 6000 km, τ =145
35 h). During the monsoon break period they reported inertia-GWs with origins near146
New Guinea in the lower stratosphere (λz ∼2 km, λh =2000 - 4000 km, τ = 45 h). High-147
frequency GWs (τ = 10 - 50 min) appeared mostly below 22 km in the stratosphere and148
were generated by local convection in the afternoon during the suppressed period. Our149
study is the first to investigate GWs during TWP-ICE with a specific focus on the tro-150
posphere.151
3 Results152
3.1 Vertical profiles of divergence153
Fig. 2a shows the temporal evolution of vertical profiles of divergence from the vari-154
ational analysis on L-Domain. Typical magnitudes of divergence are ±2×10−5s−1, as155
can also be seen from Fig. 2f. The profiles show a rich spectrum of rather persistent or156
oscillating wave-like signals. At a 3-hour lag, temporal autocorrelations for L-Domain157
are ∼0.5 for divergence averaged between the surface and 5 km (Fig. 2b), ∼0.7 for av-158
erages between 5 - 10 km (Fig. 2c), and close to 0.8 for 10 - 15 km (Fig. 2d). At a 9-hour159
lag time, they drop to zero below 5 km and to 0.2 at higher levels.160
It is insightful to intercompare the divergence profiles of the different subdomains,161
as they are spatially separated and ingest data from independent soundings. The ver-162
tical profiles of all subdomains are strongly correlated with that of L-Domain at all times163
(Fig. 2e). In contrast, L-Domain-EC does not capture the vertical structure of divergence.164
This suggests that all soundings are subject to a similar large-scale forcing of divergence,165
–4–
manuscript submitted to Geophysical Research Letters
which is not included in the ECMWF analysis. The ECMWF analysis used for L-Domain-166
EC assimilated only the 6-hourly soundings from Darwin, and was available for inges-167
tion into the variational analysis algorithm at a longitude-latitude resolution of 0.5◦×168
0.5◦ and 15 vertical levels between 1000 hPa and 10 hPa.169
Unsurprisingly, all subdomains have similar autocorrelation timescales (Fig. 2b-d).170
In addition, their magnitudes of temporal variability are very similar (Fig. 2f). This sug-171
gests that any mechanism responsible for the observed variability of divergence must be172
sufficiently large in scale to simultaneously affect all TWP-ICE domains used here.173
3.2 Gravity waves174
We next investigate if the presence of GWs may explain the observed character-175
istics of the vertical profiles of large-scale divergence. To first order, the intrinsic frequen-176
cies of GWs are bound by the inertial frequency f , which at Darwin would correspond177
to a period of 57.6 h, and the buoyancy frequency N , which would correspond to peri-178
ods of less than 10 min in the mid-troposphere during TWP-ICE. We focus on low-frequency179
waves, as the typical periods and horizontal wavelengths of high-frequency GWs are too180
short to simultaneously affect an area of the size of L-Domain. Consider the wind per-181
turbations associated with a plane wave with wave vector (k, l,m), frequency ω and phase182
φ,183
(u, v, w) = (Ax, Ay, Az)ei(kx+ly+mz−ωt+φ). (1)
Mass conservation implies that the amplitudes of the wind components (Ax, Ay, Az) are184
related through185
Az = −Axk +Ayl
m. (2)
Therefore, Az grows with decreasing vertical and increasing horizontal wavenumber rel-186
ative to the perturbations of the horizontal winds. For this reason, vertical motion can187
be used to diagnose high-frequency, typically short horizontal-wavelength waves, while188
horizontal motion can be used to diagnose low-frequency waves (Lane et al., 2003). As189
described earlier, we obtain wave perturbations of the zonal and meridional velocities,190
u′ and v′, by subtracting the respective vertical profiles from the variational analysis on191
L-domain. Examples of the resulting profiles of u′ for Point Stuart are shown in Fig. 1c.192
For illustrative purposes Fig. 1d also shows vertical velocity perturbations, defined as193
deviations from the mean ascent rate of the balloon. The comparison between the ver-194
tical and horizontal wind perturbations confirms that the latter exhibit more consistency195
between neighboring profiles, i.e., they are associated with lower-frequency, longer-wavelength,196
waves. That these wind perturbations are not mere noise but contain coherent physi-197
cal features is also clear from a visual inspection that reveals vertical phase propagation198
between subsequent soundings.199
The main spectral peaks of the low-frequency GW spectrum are identified by com-200
puting the quadrature spectrum of u′ and v′ as a function of the ground-based frequency201
ωgb and the vertical wavenumber m. Since this requires evenly spaced data points in time202
and height, u′ and v′ are first interpolated to a 10-m resolution in the vertical and to a203
10,000 s resolution in time, which roughly corresponds to the average radiosonde launch204
interval. The TWP-ICE soundings were not evenly spaced in time and the number of205
soundings differ between the stations. The spectral analysis is carried out between 2 -206
14 km altitude to avoid the boundary layer and the tropopause. Not all soundings reach207
14 km and there are intervals of missing data, as can be seen from Fig. 1c. We treat miss-208
ing observations by linearly interpolating in time from values observed at the same al-209
titude.210
Fig. 3a shows the quadrature power spectrum of u′ and v′ averaged over all sta-211
tions. We separately analyze the peaks at λz = 3 km and λz = 4 km. To isolate per-212
turbations in the relevant spectral regions, u′ and v′ are first filtered in Fourier space.213
–5–
manuscript submitted to Geophysical Research Letters
For λz = 3 km we retain ground-based periods τgb = 1.0 - 2.0 days and λz = 2.5 - 3.5214
km, and for λz = 4 km we retain τgb = 0.5 - 1.5 days, λz = 3.5 - 4.5 km. The filtered215
wind perturbations are subjected to the Stokes parameter method developed by Eckermann216
and Vincent (1989). The horizontal wind perturbation vector of low-frequency GWs ro-217
tates with height, such that the wave’s hodograph follows a polarization ellipse. The Stokes218
algorithm diagnoses key parameters of these ellipses. The angle θ of the counter-clockwise219
rotation of the major axis from the eastward direction corresponds to the propagation220
direction of the wave (with a 180◦ ambiguity). Here, the average θ is -2◦ for λz = 3 km221
and -12 ◦ for λz = 4 km (Table 1). Thus, the waves are propagating either nearly due222
eastward or westward. Studies investigating stratospheric waves showed eastward prop-223
agation (see Section 2) and, as we will argue below, the tropospheric waves are likely also224
propagating eastward. The ratio R of the major to minor axis of the polarization ellipse225
is used to estimate the typical intrinsic frequency of the waves as226
ω∗ =f
R. (3)
Background wind shear transverse to propagation direction of a wave can create an ad-227
ditional eccentricity (e.g., Cho, 1995; Hines, 1989). Here, we can neglect the correspond-228
ing correction term, as shear in the transverse direction, i.e., the meridional direction is229
close to zero (Fig. 3b).230
The horizontal wavenumber k is calculated from the GW dispersion relation as231
k2 =
(ω2∗ − f2
) (m2 + 1
4H2s
)N2
, (4)
where Hs is the density scale height and N the mean buoyancy frequency. To evaluate232
Eq. 3 we insert for m wavenumbers corresponding to λz = 3 km or 4 km, respectively.233
Intrinsic periods and horizontal wavelengths vary substantially from station to station234
(Table 1). This may in part be due to noise and the effects of missing data. But it is also235
physically plausible, as we expect the tropospheric GW background to consist of a wide236
spectrum of waves and different waves may dominate at a given location. The average237
intrinsic period is 7.6 h for λz = 3 km and 12.5 h for λz = 4 km. The intrinsic period238
for λz = 3 km is shorter than the lower bound of the filtering interval for the ground-239
based period (1 day). The zonal wind averaged between 2 - 14 km of altitude is close240
to zero, whereas the Doppler shift required to explain the difference between the intrin-241
sic and ground-based periods would imply propagation against a wind of 6 ms−1. Nev-242
ertheless, the results are not necessarily inconsistent, because we neglect the temporal243
variability of the background wind. The horizontal wavelengths are on average λh = 200244
km and 580 km, respectively. Thus, areas spanning roughly 12λh ∼ 100 km - 300 km would245
experience similar conditions in terms of GW-driven divergence. This is consistent with246
the high correlations between the vertical divergence profiles from the variational anal-247
yses on different subdomains (Fig. 2e). It is also consistent with a lack of sensitivity of248
the magnitude of temporal variability to the domain area, as all domains fit well within249
half a horizontal wavelength.250
In calm wind conditions rising or sinking motion at any given location would pre-251
vail for the duration of half the intrinsic period, i.e., four to six hours. This timescale252
is consistent with the temporal autocorrelations of divergence (Fig. 2b-d). A wave crest253
would take approximately 7.7 h (λz = 3 km) or 5.6 h (λz = 4 km) to traverse a region254
of 200 km diameter. This timescale is rather insensitive to the station (Table 1), because255
it depends only on the phase speed ch = khω∗
, which is similar for all stations (8 ms−1256
for λz = 3 km, 10 ms−1 for λz = 4 km). In addition to the period of a wave, the lat-257
ter timescale may also be important for explaining the autocorrelation times of diver-258
gence, as it implies that a crest of one wave can dominate the divergence field averaged259
over areas of 200 km diameter for roughly five to eight hours.260
–6–
manuscript submitted to Geophysical Research Letters
4 Conclusion261
We demonstrated that vertical profiles of divergence near Darwin obtained during262
the TWP-ICE campaign exhibit similar wave-like structures as those observed over the263
tropical Atlantic. Bony and Stevens (2019) reported values of 2×10−5s−1 for the mag-264
nitudes of divergence variability in layers of a few km thickness. This magnitude of vari-265
ability is in excellent agreement with TWP-ICE data (Fig. 2f). The observations dis-266
cussed by Bony and Stevens (2019) were taken over the ocean while the TWP-ICE ob-267
servations were mainly taken over land. Both studies examined convectively suppressed268
conditions during local summer at mirroring latitudes. The similar absolute value of the269
Coriolis parameter may partly explain the similarities between the different locations,270
as it enters the gravity wave (GW) dispersion relation and acts as a lower bound on the271
range of wave intrinsic frequencies. The autocorrelations reported by (Bony & Stevens,272
2019) were based on high-resolution simulations, which were initialized and nudged at273
their boundary using data from the Integrated Forecast System (IFS) of the ECMWF.274
They inferred temporal autocorrelation scales of 4 h for circular regions of 100 km ra-275
dius, which is confirmed by the observations from TWP-ICE. The autocorrelation timescale276
of divergence agrees very well with the characteristic GW timescales derived here. TWP-277
ICE data do not allow us to test spatial autcorrelation scales, which were 2.2◦ of lati-278
tude over the tropical Atlantic (Bony & Stevens, 2019). This distance corresponds to the279
typical GW horizontal wavelengths found for Darwin, such that GW dynamics would280
be able to explain the spatial correlations as well. Thus, low-frequency GWs can serve281
as an explanation for the vertical, horizontal and temporal structures of mesoscale di-282
vergence patterns.283
The existence of strong convergence or divergence layers implies that the mesoscale284
environment is potentially strongly influenced by physical processes other than those usu-285
ally included in climate models. Interactions between waves and convection are not cap-286
tured in climate models of coarse resolution. The typical effective horizontal and ver-287
tical resolutions are hardly sufficient for resolving the waves, let alone their sources or288
interaction with subsequent convection. The scales discussed here are large compared289
to single updrafts but correspond well to scales of organized convection. Thus, low-frequency290
GWs might be important for determining the characteristics of cloudiness on the mesoscale.291
Clouds and their coupling to circulation still remain one of the major uncertainties in292
climate predictions (Bony et al., 2015). Further efforts are needed to quantify the wave293
background and its interaction with convection and clouds. The temporal and spatial294
scales reported here can serve as guidance for measurements during future field campaigns,295
such as the field campaign Elucidating the Role of CloudCirculation Coupling in Climate296
(EUREC4A Bony et al., 2017), which will take place in winter 2020, and could be used297
to collect additional data for analyzing the response of convection to GW dynamics in298
relation to other cloud controlling factors. The relatively short inertial periods of the waves299
(Table 1) favor soundings at short intervals, ideally every 2 hours. Given that the hodo-300
graph method can be independently applied to data at single locations, one station with301
high-frequency soundings would already be an improvement over, e.g., an array of lo-302
cations with 6-hourly soundings. Moreover, if soundings at different locations within sev-303
eral hundreds of kilometers were launched simultaneously, this may allow a direct com-304
putation of the horizontal wavelengths by inferring the phase differences between differ-305
ent locations.306
Inertia GWs are able to propagate large distances inside the troposphere. Based307
on their large spatial scales and the fact that convective sources always exist at some lo-308
cation in the tropics, we expect such waves to be present most of the time. Our results309
suggest that GWs are a plausible explanation for the scaling laws of mesoscale divergence310
at two different locations. Thus, it is possible that tropical divergence on spatial scales311
of a few hundred kilometers follows general scaling laws that are determined by the char-312
acteristics of low-frequency gravity waves. Additional measurements are necessary to de-313
–7–
manuscript submitted to Geophysical Research Letters
Table 1. Gravity wave properties derived from radiosonde soundings. θ is the propagation
direction measured counter-clockwise from eastward, τ∗ is the intrinsic period, λh the horizon-
tal wavelength, and τ∗200 the time it takes a wave to travel 200 km when there is no background
wind.
(a) Results for λz = 3 kmStation θ (◦) τ∗ (hours) λh (km) τ∗200 (hours)
Garden Point 20 12.4 330 7.6Mount Bundy -18 6.6 170 7.7Point Stuart 37 5.4 140 7.7Ship -38 6.7 180 7.7Cape Don -12 7.1 180 7.7
Average -2 7.6 200 7.7
(b) Results for λz = 4 kmStation θ (◦) τ∗ (hours) λh (km) τ∗200 (hours)
Garden Point 20 20.9 770 5.4Mount Bundy 21 2.3 750 5.4Point Stuart -37 12.4 440 5.7Ship -42 11.3 400 5.7Cape Don -24 15.5 550 5.6
Average -12 12.5 580 5.6
termine to what degree the tropospheric wave background is saturated or consists of in-314
dividual spectral peaks or individual waves, as this would affect the universality of po-315
tential scaling laws.316
Acknowledgments317
C. C. Stephan was supported by the Minerva Fast Track Programme of the Max Planck318
Society. C. C. Stephan is grateful for the joint funding from the ARC Centre of Excel-319
lence for Climate Extremes (CE170100023) and the Max Planck Institute for Meteorol-320
ogy to enable the research visit to Australia that has led to the research reported here.321
The authors also acknowledge the support of the US Department of Energy Atmospheric322
Radiation Measurement (ARM) Program for the TWP-ICE campaign. The data used323
in this study are available from the ARM data archive (https://www.arm.gov/data).324
References325
Alexander, M. J., Geller, M., McLandress, C., Polavarapu, S., Preusse, P., Sassi, F.,326
. . . Watanabe, S. (2010). Recent developments in gravity-wave effects in cli-327
mate models and the global distribution of gravity-wave momentum flux from328
observations and models. Quart. J. Roy. Meteor. Soc., 136 (650), 1103–1124.329
doi: 10.1002/qj.637330
Balaji, V., & Clark, T. L. (1988). Scale selection in locally forced convective fields331
and the initiation of deep cumulus. J. Atmos. Sci., 45 , 3189–3211. doi: 10332
.1175/1520-0469(1988)045〈3188%3ASSILFC〉2.0.CO%3B2333
Bony, S., & Stevens, B. (2019). Measuring area-averaged vertical motions with drop-334
sondes. J. Atmos. Sci., 76 , 767–783. doi: 10.1175/JAS-D-18-0141.1335
Bony, S., Stevens, B., Ament, F., Bigorre, S., Chazette, P., Crewell, S., . . . Wirth,336
M. (2017). EUREC4A: A field campaign to elucidate the couplings between337
–8–
manuscript submitted to Geophysical Research Letters
clouds, convection and circulation. Surveys in Geophysics, 38 (6), 1529–1568.338
doi: 10.1007/s10712-017-9428-0339
Bony, S., Stevens, B., Frierson, D. M. W., Jakob, C., Kageyama, M., Pincus, R.,340
. . . Webb, M. J. (2015). Clouds, circulation and climate sensitivity. Nature341
Geoscience, 8 , 261. doi: 10.1038/ngeo2398342
Cho, J. Y. N. (1995). Inertio-gravity wave parameter estimation from cross-343
spectral analysis. J. Geophy. Res. Atm., 100 (D9), 18727–18737. doi:344
10.1029/95JD01752345
Davies, L., Jakob, C., May, P., Kumar, V. V., & Xie, S. (2013). Relationships346
between the large-scale atmosphere and the small-scale convective state347
for Darwin, Australia. J. Geophys. Res. Atmos., 118 , 1153411545. doi:348
10.1002/jgrd.50645349
Eckermann, S. D., & Vincent, R. A. (1989). Falling sphere observations of350
anisotropic gravity wave motions in the upper stratosphere over Australia.351
Pure Appl. Geophys., 130 , 509–532. doi: 10.1007/BF00874472352
Evan, S., & Alexander, M. J. (2008). Intermediate-scale tropical inertia gravity353
waves observed during the TWP-ICE campaign. J. Geophys. Res., 113 . doi: 10354
.1029/2007JD009289355
Evan, S., Alexander, M. J., & Dudhia, J. (2012). Model study of intermediate-scale356
tropical inertia-gravity waves in comparison to TWP-ICE campaign observa-357
tions. J. Atmos. Sci., 69 , 591–610. doi: 10.1175/JAS-D-11-051.1358
Hankinson, M. C. N., Reeder, M. J., & Lane, T. P. (2014a). Gravity waves gener-359
ated by convection during TWP-ICE: 2. high-frequency gravity waves. J. Geo-360
phys. Res. Atmos., 119 , 5257–5268. doi: 10.1002/2013JD020726361
Hankinson, M. C. N., Reeder, M. J., & Lane, T. P. (2014b). Gravity waves gen-362
erated by convection during TWP-ICE: I. Inertia-gravity waves. J. Geophys.363
Res. Atmos., 119 , 5269–5282. doi: 10.1002/2013JD020724364
Hecht, J. H., Alexander, M. J., Walterscheid, R. L., Gelinas, L. J., Vincent, R. A.,365
MacKinnon, A. D., . . . Russell III, J. M. (2009). Imaging of atmospheric366
gravity waves in the stratosphere and upper mesosphere using satellite and367
ground-based observations over Australia during the twpice campaign. J.368
Geophys. Res., 114 . doi: 10.1029/2008JD011259369
Hines, C. O. (1989). Tropopausal mountain waves over Arecibo: A case study. J. At-370
mos. Sci., 46 , 476–488. doi: 10.1175/1520-0469(1989)046〈0476%3ATMWOAA〉371
2.0.CO%3B2372
Holton, J. (1983). The influence of gravity wave breaking on the general circulation373
of the middle atmosphere. J. Atmos. Sci., 40 , 2497–2507. doi: 10.1175/1520374
-0469(1983)040〈2497:TIOGWB〉2.0.CO;2375
Holton, J. R. (1982). The role of gravity wave induced drag and diffusion in the mo-376
mentum budget of the mesosphere. J. Atmos. Sci., 39 , 791–799. doi: 10.1175/377
1520-0469(1982)039〈0791:TROGWI〉2.0.CO;2378
Lane, T. P., Reeder, M. J., & Guest, F. M. (2003). Convectively generated grav-379
ity waves observed from radiosonde data taken during MCTEX. Quart. J. Roy.380
Meteor. Soc., 129 , 1731–1740. doi: 10.1256/qj.02.196381
Lane, T. P., & Zhang, F. (2011). Coupling between gravity waves and trop-382
ical convection at mesoscales. J. Atmos. Sci., 68 , 2582–2598. doi:383
10.1175/2011JAS3577.1384
Mapes, B. E. (1993). Gregarious tropical convection. J. Atmos. Sci., 50 , 2026–2037.385
doi: 10.1175/1520-0469(1993)0502026:GTC2.0.CO;2386
May, P. T., Mather, J. H., Vaughan, G., Jakob, C., McFarquhar, G. M., Bower,387
K. N., & Mace, G. G. (2008). The Tropical Warm Pool-International388
Cloud Experiment. Bull. Am. Meteorol. Soc., 89 , 629–645. doi: 10.1175/389
BAMS-89-5-629390
Shige, S., & Satomura, T. (2001). Westward generation of eastward-moving tropi-391
cal convective bands in TOGA COARE. J. Atmos. Sci., 58 , 37243740. doi: 10392
–9–
manuscript submitted to Geophysical Research Letters
.1175/1520-0469(2001)058〈3724%3AWGOEMT〉2.0.CO%3B2393
Stechmann, S. N., & Majda, A. J. (2009). Gravity waves in shear and implica-394
tions for organized convection. J. Atmos. Sci., 66 , 2579–2599. doi: 10.1175/395
2009JAS2976.1396
Stephan, C. C., Alexander, M. J., & Richter, J. H. (2016). Characteristics397
of gravity waves from convection and implications for their parameteri-398
zation in global circulation models. J. Atmos. Sci., 73 , 2729–2742. doi:399
10.1175/JAS-D-15-0303.1400
Su, T., & Zhai, G. (2017). The role of convectively generated gravity waves on con-401
vective initiation: A case study. Mon. Wea. Rev., 145 , 335359. doi: 10.1175/402
MWR-D-16-0196.1403
Xie, S., Hume, T., Jakob, C., Klein, S. A., McCoy, R. B., & Zhang, M. (2010). Ob-404
served large-scale structures and diabatic heating and drying profiles during405
TWP-ICE. J. Clim., 23 , 57–79. doi: 10.1175/2009JCLI3071.1406
Zhang, M. H., & Lin, J. L. (1997). Constrained variational analysis of sounding data407
based on column-integrated budgets of mass, heat, moisture, and momentum:408
Approach and application to ARM measurements. J. Atmos. Sci., 54 , 1503–409
1524. doi: 10.1175/1520-0469(1997)054〈1503%3ACVAOSD〉2.0.CO%3B2410
–10–
manuscript submitted to Geophysical Research Letters
Jan 28 Jan 30 Feb 1 Feb 3 Feb 5 Feb 7time
0
5
10
15
he
igh
t (k
m)
(c) u’ Point Stuart soundings
Jan 28 Jan 30 Feb 1 Feb 3 Feb 5 Feb 7time
0
5
10
15
he
igh
t (k
m)
(d) w’ Point Stuart soundings
Jan 28 Jan 30 Feb 1 Feb 3 Feb 5 Feb 7time (days)
0
1
2
3
4
5
height (km)
(b) Data coverage
Gard
en P
oin
t
2 -
14 k
m
Mount B
undy
2 -
14 k
m
Poin
t S
tuart
2 -
14 k
m
Ship
2 -
14 k
m
Cape D
on
2 -
14 k
m
129 130 131 132 133Longitude (deg E)
-14.0
-13.5
-13.0
-12.5
-12.0
-11.5
-11.0
La
titu
de
(d
eg
N)
(a) TWP-ICE domain
Darwin
Ocean
Mainland
Tiwi
L-Domain
Garden Point
Mount Bundy
Point Stuart
Ship
Cape Don
Figure 1. (a) Locations of Darwin and the five radiosonde launching sites during TWP-
ICE. L-Domain (black) encloses the five boundary forcing stations. Subdomains are shown with
colored lines. (b) For each launch location colored areas show times and heights that are not cov-
ered by soundings. Data were first interpolated to a temporal resolution of 10,000 s. (c) Zonal,
and (d) vertical wind perturbations from radiosonde soundings interpolated to a 10 m vertical
grid but not interpolated in time from Point Stuart. w′ are deviations from a linear fit to the
mean ascent rate of the balloon, scaled such that 10 ms−1 corresponds to 0.85 days on the x-
axis. u′ are deviations from the area-mean horizontal wind speed of the variational analysis on
L-Domain, scaled such that 10 ms−1 corresponds to 1.7 days on the x-axis.
–11–
manuscript submitted to Geophysical Research Letters
Jan 28 Jan 30 Feb 1 Feb 3 Feb 5 Feb 7time
2
4
6
8
10
12
14h
eig
ht
(km
)
(a) Div profiles of L-Domain
2 4 6 8 10lag (hours)
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
au
toco
rre
latio
n
(b) Sfc-5 km
L-DomainMainlandOceanTiwi
2 4 6 8 10lag (hours)
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
autocorrelation
(c) 5-10 km
2 4 6 8 10lag (hours)
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
autocorrelation
(d) 10-15 km
Jan 28 Jan 30 Feb 1 Feb 3 Feb 5 Feb 7time
-1.0
-0.5
0.0
0.5
1.0
co
rre
latio
n
L-Domain-ECMainlandOceanTiwi
(e) Correl. with Div profiles of L-Domain
1.0 1.2 1.4 1.6 1.8 2.0div (10-5s-1)
0
1
2
3
layer
L-DomainMainlandOceanTiwi
Sfc
-5 k
m5
-10
km
10
-15
km
(f) Div magnitude
Figure 2. Characteristics of large-scale divergence from the variational analyses for the sup-
pressed period Jan 28 00:00 UTC - Feb 7 00:00 UTC. (a) 3-hourly vertical profiles of divergence
on L-Domain. Divergence is scaled such that one day on the x-axis corresponds to 104s−1. (b-d)
Autocorrelations of divergence vertically averaged over (a) surface - 5 km, (b) 5 - 10 km and (c)
10 - 15 km for all domains. (e) Correlations of the vertical divergence profiles between 2 - 14 km
with that of L-Domain. L-Domain-EC (dashed) denotes the variational analysis on L-Domain
that is based on virtual soundings from the ECMWF analysis. (f) Time-averaged divergence
amplitude in the vertical layers, which are labeled on the y-axis. Data points for the different
variational analyses are vertically off-set for better visibility.–12–
manuscript submitted to Geophysical Research Letters
0.5 1.0 1.5 2.0 2.5period (days)
1
2
3
4
5
6
ve
rtic
al w
ave
len
gth
(km
)
(a) Normalized FFT power
0 10 20 30 40 50 60 70 80 90 100
0
1
-20 -10 0 10 20speed (ms-1)
2
4
6
8
10
12
14
16
he
igh
t (k
m) UV
N
0.005 0.010 0.015 0.020 0.025 0.030N (s-1)
(b) Wind and stability profiles
Figure 3. (a) The quadrature spectrum of u′ and v′ for the suppressed period Jan 28 00:00
UTC - Feb 7 00:00 UTC averaged over all five radiosonde stations. u′ and v′ are the deviations
from the variational analysis on L-Domain and were first interpolated to vertical and temporal
resolutions of 10 m and 10,000 s, respectively. Boxes mark the spectral regions that are sepa-
rately subjected to a Stokes analysis. (b) Average background profiles of U , V and N from the
variational analysis on L-Domain for the suppressed period.
–13–