ACPAtmospheric Chemistry and PhysicsACPAtmos. Chem. Phys.1680-7324Copernicus PublicationsGöttingen, Germany10.5194/acp-16-541-2016The tropopause inversion layer in baroclinic life-cycle
experiments: the role of diabatic processesKunkelD.dkunkel@uni-mainz.dehttps://orcid.org/0000-0002-9652-0099HoorP.WirthV.Institute for Atmospheric Physics, Johannes Gutenberg University
Mainz, Mainz, GermanyD. Kunkel (dkunkel@uni-mainz.de)19January201616254156021July201510August201524November20153December2015This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://acp.copernicus.org/articles/16/541/2016/acp-16-541-2016.htmlThe full text article is available as a PDF file from https://acp.copernicus.org/articles/16/541/2016/acp-16-541-2016.pdf
Recent studies on the formation of a quasi-permanent layer of
enhanced static stability above the thermal tropopause revealed the
contributions of dynamical and radiative processes. Dry dynamics leads
to the evolution of a tropopause inversion layer (TIL), which is,
however, too weak compared to observations and thus diabatic
contributions are required. In this study we aim to assess
the importance of diabatic processes in the
understanding of TIL formation at midlatitudes.
The non-hydrostatic model COSMO (COnsortium for Small-scale
MOdelling) is applied in an idealized midlatitude
channel configuration to simulate baroclinic life cycles.
The effect of individual diabatic processes related to humidity,
radiation, and turbulence is studied first to estimate the
contribution of each of these processes to the TIL formation in addition to
dry dynamics. In a second step these processes are stepwise included
in the model to increase the complexity and finally estimate the
relative importance of each process.
The results suggest that including turbulence leads to a weaker TIL
than in a dry reference simulation. In contrast, the TIL evolves
stronger when radiation is included but the temporal evolution is
still comparable to the reference. Using various cloud schemes in
the model shows that latent heat release and consecutive increased vertical
motions foster an earlier and stronger appearance of the TIL than in
all other life cycles.
Furthermore, updrafts moisten the upper troposphere and as such
increase the radiative effect from water vapor. Particularly, this
process becomes more relevant for maintaining the TIL during later
stages of the life cycles. Increased convergence of the vertical
wind induced by updrafts and by propagating inertia-gravity waves,
which potentially dissipate, further contributes to the enhanced
stability of the lower stratosphere. Finally, radiative
feedback of ice clouds reaching up to the tropopause is identified to
potentially further affect the strength of the TIL in the region of
the clouds.
Introduction
The sharpness of the tropopause in the extratropics has gained increased
attention in recent years e.g.,. Local maxima of
static stability, usually measured by the squared Brunt–Vaisala frequency
N2=g/Θ⋅∂Θ/∂z, where g is the gravitational
acceleration, Θ the potential temperature, and z the geometric
altitude, inferred from radiosonde measurements e.g., and Global Positioning System (GPS) radio occultation
measurements , revealed the existence of a quasi-permanent
inversion layer above the thermal tropopause. This tropopause inversion layer
(TIL) is a distinct feature of the region of the upper troposphere and lower
stratosphere (UTLS), from tropical to polar regions
e.g., and is also evident in general circulation models
and climate analysis data sets e.g.,.
Global studies of GPS temperature profiles and re-analysis data sets showed
that the TIL is present at all latitudes . In
the tropical lower stratosphere two maxima of enhanced static stability are
found at about 17 and 19 km altitude. The upper peak shows a seasonal
cycle with a winter maximum, while the lower peak has relatively large values
all year round . In polar regions a distinct summer maximum
occurs , while the TIL is evident in midlatitudes
throughout the entire year with a slightly deeper appearance during winter
. Generally, the smallest values of static stability above
the thermal tropopause are evident in the region of the subtropical jet
.
In several studies it was shown that a TIL can form from balanced, adiabatic,
and frictionless dynamics without explicit contributions from radiation in
the extratropics. These idealized model simulations span the range from local
to global scales, with studies of the dynamics of upper-level anomalies of
potential vorticity (further abbreviated with PV) , of baroclinic life cycles , and of the dynamical
response to a forcing of a Held–Suarez test in a dry
general circulation model . In the latter case, the TIL forms
spontaneously under a wide variety of model parameters, such as horizontal
and vertical model resolution. From the analysis of positive and negative
PV anomalies it was found that the sharpening of the tropopause was linked to
the convergence of the vertical wind. Particularly, this was related to
a cross-frontal circulation . Furthermore, the TIL evolved
stronger above anticyclonic than over cyclonic flow . This
result was confirmed in studies of adiabatic baroclinic life cycles, in which
the TIL became evident after breaking of baroclinic waves .
Recently, the impact of dissipating inertia-gravity waves was suggested to
persistently contribute to the formation and maintenance of the TIL. These
waves result from imbalances along the jet and the dissipation may alter the
thermal structure through energy dissipation, local heating, and turbulent
motions . Moreover, showed that the
vertical structure of the residual circulation in the stratosphere
contributes to the sharpening of the tropopause by inducing a dipole forcing
of static stability around the tropopause. This process was identified to
significantly add to the tropopause sharpening during winter in the
midlatitudes.
Balanced dynamics alone, however, cannot explain all features
related to the TIL and as has been shown by
radiative processes contribute significantly to the
TIL. From fixed dynamical radiative transfer calculations it
was concluded that water vapor cooling around the tropopause and heating by
ozone in the lower and middle stratosphere contribute to a layer of enhanced
static stability above the thermal tropopause. Particularly, the water vapor
cooling has been identified to be a major process for the summer TIL in polar
regions .
Thus, several mechanisms have been identified so far to explain the strength
and occurrence of the TIL at all latitudes. Since dry dynamics is not
sufficient to fully explain all features of the TIL, processes beyond
adiabatic and frictionless dynamics are required to close this gap.
Especially in the midlatitude tropopause region, all processes,
synoptic-scale and stratospheric dynamics as well as the radiative forcings,
need to be considered. With this knowledge we can ask the question which of
the before mentioned processes is most important to form and maintain the
TIL. In this study we aim to address this question in the framework of
idealized baroclinic life cycles with a limited area, non-hydrostatic model.
We extend the work of and include diabatic processes, i.e.,
related to humidity, radiation, or turbulence. These processes can violate
material conservation of potential vorticity Q and are further referred to
as non-conservative processes in this study. Since we focus on a rather short
timescale, we assume that the effect of the stratospheric circulation is
rather small and exclude this effect in the interpretation of our results.
Thus, we focus mainly on the following questions: (1) how do non-conservative
processes, i.e., diabatic processes, alter the TIL evolution in baroclinic
life cycles compared to the well-known evolution in the adiabatic and
frictionless case? (2) What is the relative importance of individual
processes that contribute to the formation the TIL during different stages of
the life cycles?
To answer these questions we structured our analysis as follows. In
Sect. we introduce the model setup along with the physical
parameterizations and a summary of the conducted simulations. We then present
results from two sets of simulations of so-called anticyclonic life cycles.
In Sect. we show results from baroclinic life cycles in which
only one individual non-conservative process is turned on separately to
address question 1. In a second set of simulations we show results of
simulations with a successively increasing number of physical processes to
address question 2 (Sect. ). Before we summarize our results
and give further conclusions in Sect. , we discuss the evolution
of the tropopause inversion layer in experiments of the cyclonic life cycle
in Sect. .
Model formulation and baroclinic life-cycle experimentsAdiabatic model configuration and initial state
We conducted baroclinic life-cycle experiments in an idealized, spherical,
midlatitude channel configuration of the non-hydrostatic regional model
COSMO COnsortium for Small-scale
MOdelling;. For the adiabatic model we
only use the dynamical core of the model, which solves the
hydro-thermodynamical equations. Only a fourth-order horizontal
hyper-diffusion has to be applied to guarantee numerical stability. Physical
processes, such as microphysics, convection, turbulence, and radiation are
introduced in more detail further below (see Sect. ). Time
integration is performed with a third-order, two-time-level Runge–Kutta
scheme, in which fast terms, i.e., sound and gravity waves, are stepped
forward in time with a smaller time step. We use a fifth-order centered
finite difference approximation in the horizontal and a third-order scheme in
the vertical. Passive tracer advection is done with a fourth-order Bott scheme with Strang splitting .
We study baroclinic waves with wave number six with a model setup similar to
and . Our model domain spans over
60∘ longitude and 70∘ latitude, from the surface up to
a height of 25.0 km and with a grid spacing of 0.4∘ (∼44km) in the horizontal and 110 m in the vertical in the
region of the tropopause. Consequently, we obtain an aspect ratio (Δz/Δx) of about 1/400, which is considered favorable to study the
TIL . In the uppermost 7 km of the
model domain, Rayleigh damping is applied to avoid reflection of upward
propagating signals and there is no orography at the bottom. In the meridional
direction the boundary conditions are relaxed towards the initial values to
avoid reflection of outgoing signals, while periodic boundary conditions are
specified in the zonal direction.
Summary of experiment acronyms, description, and water treatment
ExperimentShort descriptionWater speciesREFadiabatic reference simulationno water speciesBMPstandard cloud microphysicsinteractive waterRADstandard radiation schemepassive water vaporTURBstandard turbulence schemeno water speciesBMP R30BMP sensitivity, reduced specific water vaporinteractive waterBMP R80BMP sensitivity, increased specific waterinteractive waterBMP NOICEBMP sensitivity, only warm cloudsinteractive water, no ice phaseBMP SATADBMP sensitivity, saturation adjustmentwater vapor and cloud waterRAD woSWRAD sensitivity, no stratospheric waterpassive water vaporRAD R30RAD sensitivity, reduced specific water vaporpassive water vaporRAD rO3RAD sensitivity, reduced ozone concentrationpassive water vaporBMP RADcloud microphysics and radiationinteractive waterBMP RADcloud microphysics and radiation,interactive water NOCRF no cloud radiative feedbackBMP TURBcloud microphysics and turbulenceinteractive waterBRTcloud microphysics, radiation, and turbulenceinteractive waterBRTCcloud microphysics, radiation, turbulence,interactive waterand convectionBRTCScloud microphysics, radiation, turbulence,interactive waterconvection, and surface fluxes for momentum and heatREF LC2adiabatic reference simulation for LC2no water speciesBMP LC2standard cloud microphysics for LC2interactive waterRAD LC2standard radiation scheme for LC2passive water vaporTURB LC2standard turbulence scheme for LC2no water speciesBRTC LC2cloud microphysics, radiation, turbulence,interactive waterand convection for LC2
For the initial conditions, we follow and
with slight adaptations to account for the spherical geometry of our
approach. A background state is obtained for three-dimensional fields of
temperature, T, and pressure, p, from which a thermally balanced wind is
calculated as in . The initial vertical wind, w, is zero
and the background state is baroclinically unstable by construction. However,
to allow a fast evolution of the baroclinic wave, this state is superimposed
by perturbation fields for p,T,u, and v, which result from an inversion
of a specified PV anomaly. This circular anomaly is introduced in the middle
of the domain at the altitude of the tropopause. Slight changes in the
initial state allow us to study various types of baroclinic life cycles
for details we refer to. To obtain a solution to our
experiments that is known as LC2 (cyclonic life cycles) , an additional
cyclonic barotropic shear is added to the background state described above.
However, the main focus of this study is on the classical LC1 (anticyclonic life cycles) wave type
, since it produces a stronger TIL in the adiabatic
case . In Sect. we will present differences in
the evolution of the TIL in LC2 experiments. The LC1 type is characterized by
a thinning trough, which then forms a streamer and later a cut-off cyclone,
while the baroclinic wave breaks anticyclonically. Thus, the LC1 is also
known as the anticyclonic case. In contrast, in the LC2 a large cyclonic
trough dominates the evolution of the wave with no streamer and no cut-off
cyclone being evident. This case is known as the cyclonic case, since the
wave breaks cyclonically. More details on the development of these waves and
the corresponding evolution of the tropopause inversion layer are generally
given in and for the LC1 setup specifically in
, where the authors used a higher-resolution version of
this model. It is noted here that the lower-resolution model well reproduces
the results of . For this reason and because of the vast
number of conducted model simulations (see Table ), we decided to
use a coarser grid spacing in our simulations.
Meridional cross section of the initial state at the center
of the model domain: the zonal wind U is color-coded for values of
5, 15, 25, 35, and 45 m s-1; the potential temperature
Θ is shown by the black dotted lines for 280, 320, and
360 K (from bottom to top); the water vapor mixing ratio
is shown by the blue lines for values of 2.0, 0.2,
and 0.02 gkg-1 (from bottom to top); the
location of thermal tropopause is indicated by the solid thick black
line and also separates the region of tropospheric values (N2<1.5×10-4s-2) from stratospheric values (N2∼4.0×10-4s-2) of static stability; the
location of the dynamical tropopause, defined as the isosurface of
potential vorticity Q=2.0 pvu, is shown by the
dashed thick line.
Figure shows the initial state in the center of our model domain.
The zonal wind u has its maximum velocity between the thermal and dynamical
tropopause (here defined as the Q=2.0pvu contour line, where pvu
is potential vorticity units, and 1.0pvu=1.0×10-6Km2kg-1s-1). For the thermal tropopause we
follow the definition given in , where the tropopause is
defined as the lowest level where the temperature lapse rate falls below
2.0Kkm-1 and its average between this level and all higher
levels within 2 km above this level remains below this value. The thermal
tropopause further separates tropospheric (N2<1.5×10-4s-2) from the stratospheric (N2>4.0×10-4s-2) background values of static stability. The initial
zonally symmetric specific humidity field, depicted with the blue lines, has
been constructed such that it is comparable in magnitude and distribution to
moisture profiles from re-analysis data. For this it is constructed as
follows: a constant surface relative humidity (RHs) is
given, which decreases linearly with height everywhere. If not specified
otherwise, RHs is 60% and decreases with
a gradient of 10%/2km. Thus, above 12 km
altitude the relative humidity (RH) is zero. The model, however,
requires specific humidity qv as input variable. This quantity
is obtained by multiplication of the relative humidity with the saturation
specific humidity qvs:qv=RH/100⋅qvs. The latter quantity is computed
from the saturation water vapor, which is computed with the parameterization
of Magnus . A final constraint is given for the initial
distribution of qv, i.e., that
minqv=2.0×10-6kgkg-1.
Note that this leads to a constant initial value of qv=2.0×10-6kgkg-1 in the stratosphere in our simulations.
We further use passive tracers to diagnose particular features of our
baroclinic life cycles. These tracers are purely advected and not explicitly
mixed vertically or horizontally by a parameterization scheme. However,
mixing due to numerical reasons does still affect the tracer distribution. In
particular, we use three tracers, which carry information of the initial state
of the baroclinic life cycles: (1) the initial height of each grid box
z0, (2) the initial static stability N02, and (3) the initial
potential vorticity Q0. With these tracers it is possible to calculate
the differences between the current and the initial distribution of these
quantities and as such obtain information about whether an air parcel has
gained or lost (1) altitude, measured by Δz=z-z0, (2) static
stability, measured by ΔN2=N2-N02, and (3) changed their
potential vorticity because of non-conservative processes, measured by
ΔQ=Q-Q0, with Q=ϱ-1η⋅∇Θ and ϱ air density, η absolute vorticity,
and Θ potential temperature. We want to note here that we will use
positive values of Δz as a predictor of vertically ascending air
masses. In this study we will also use the term updraft to describe these air
masses, independently of the cause of the ascent, e.g., frontal or
convective, and of any time period in which the ascent has occurred. The only
criteria are that Δz is larger than 2.5 km and that these air
masses reach the altitude of the tropopause.
Formulation of non-conservative processes in COSMOTurbulence
Turbulence is calculated for the three-dimensional wind (u, v, and
w), the liquid water potential temperature (Θl),
and the total water (qw), which is the sum of specific
water vapor qv and specific cloud water
qc. Budget equations for the second-order moments are reduced under application of a closure of level 2.5
(in the notation of ), i.e., local equilibrium is
assumed for all moments except for turbulent kinetic energy (TKE), for
which advection and turbulent transport is
retained. Three-dimensional turbulent effects are neglected,
which is a valid approximation for simulations on the mesoscale, which
means that horizontal homogeneity is assumed. Hence, only vertical
turbulent fluxes are parameterized under consideration of the Boussinesq
approximation. Moreover, the TKE budget equation depends significantly
on the vertical shear of the horizontal wind components and the vertical
change in Θl and qw. More details are
given in .
Cloud microphysics
Cloud microphysics follow a bulk approach using a single moment scheme with
five types of water categories being treated prognostically: specific
humidity qv for the gas phase, two non-precipitating cloud
types, i.e., cloud water qc and cloud ice qi, as
well as two precipitating types, i.e., rain qr and snow
qs. These five water types can interact within various
processes such as cloud condensation and evaporation, depositional growth and
sublimation of snow, evaporation of snow and rain, melting of snow and cloud
ice, homogeneous and heterogeneous nucleation of cloud ice, autoconversion,
collection, and freezing. More details are given in and
.
Radiation
Radiation is parameterized by the δ-2 stream approximation, i.e.,
separate treatment of solar and terrestrial wavelengths. In total,
eight spectral bands are considered, five in the solar range and three
infrared bands. Absorbing and scattering gases are water vapor
(H2O) with a variable content as well as CO2,
O3, CH4, N2O, and O2
with fixed amounts. Aerosols have been totally neglected whereas
a cloud radiative feedback can be calculated in all spectral
bands. Further details about the general scheme are given in
and about the implementation in .
Convection
The scheme of is used to parameterize sub-grid-scale
convective clouds and their effects on the large-scale environment. This
approach uses moisture convergence in the boundary layer to estimate the
cloud base mass flux. The convection scheme then affects the large-scale
budgets of the environmental dry static energy, the specific humidity, and
the potential energy.
Surface fluxes
Instead of using a bottom free-slip boundary condition surface fluxes of
momentum and heat are calculated explicitly in one experiment. This results
in non-zero turbulent transfer coefficients of momentum and heat and thus
affects the roughness length and the fluxes of latent and sensible heat. As
we will show later, this has some significant effects on the initiation of
convection.
Simulations of baroclinic life cycles
In total we present the results of 17 different simulations of the
anticyclonic and of five different simulations of the cyclonic baroclinic
life cycle (see Table ). Variations between the individual
simulations are introduced by either the kind or the number of
non-conservative processes. Moreover, additional variability is created by
changing the initial humidity as well as by the complexity of treating cloud
related processes.
In a first set of simulations, we conducted four different baroclinic life
cycles. Using the adiabatic and frictionless life cycle as conservative
reference simulation (REF), we obtain further results from life cycles
additionally including either turbulence, further denoted as TURB, or
radiation, RAD, or bulk microphysics, BMP. For these
simulations we apply the standard physical parameterizations of COSMO, which
were briefly described in the previous section.
We performed further sensitivity simulations for BMP and
RAD to test for the impact of initial conditions as well as
the model formulation of a diabatic process. For microphysics we conducted
in total four additional life-cycle experiments. We first tested for the initial
specific humidity qv. In one case we reduced the
initial qv by setting the surface relative humidity to
30% and the gradient to 5.0%/2km (BMP R30), while we increased the initial qv by using
RHs=80% and a gradient of
13.33%/2km in another case (BMP R80).
Furthermore, we conducted simulations in which we used different schemes to
represent cloud processes. In one simulation only warm phase clouds are
considered, excluding cloud ice (BMP NOICE). In another simulation
condensation and evaporation between water vapor and cloud water is realized
by a saturation adjustment process (BMP SATAD). Since this simulation
includes only large-scale diabatic effects from latent heating, it has the
least additional effects compared to the dry reference .
Dynamical and thermodynamical state of the baroclinic life
cycles after 120 h of model integration. In the upper row
the distribution of potential temperature Θ (in K) on the dynamical
tropopause is depicted, while the lower row shows the distribution
of static stability N2 (in 10-4s-2) averaged over
the first kilometer above the thermal tropopause. The four columns
show from left to right the following simulations: (a) REF, (b) TURB,
(c) RAD, and (d) BMP.
In the case of radiation, we performed sensitivity simulations with respect to the
initial distribution of specific humidity and ozone. These two trace gases
are thought to have the largest impact on the thermal structure around the
tropopause e.g.,. We conducted one
simulation with reduced initial specific humidity (RAD R30), similar
to BMP R30, while we explicitly set the specific humidity to zero
above the tropopause in another simulation (RAD woSW). In another case
we reduced the amount of ozone (RAD rO3). However, we explicitly note
here that ozone is poorly represented in the model. Instead of a three-dimensional distribution, only a simple vertical distribution is
assumed,
which has a maximum concentration at altitudes that are close to our model
top at a pressure of 42 hPa and a total vertically integrated ozone
partial pressure of 0.06 Pa. These two parameters are used in the
radiation code to calculate the feedback of the solar and thermal extinction
by ozone. We reduced the total amount of ozone by one-third to estimate
whether this has an impact on the strength of the TIL.
In a next step we use a set of simulations with combinations of
non-conservative processes to study potential additive effects as well
as to assess the relative contribution of individual processes on the
TIL formation and maintenance during different stages of the life
cycles. For this we compare results from BMP (here as
a reference) to results from simulations where we first add radiation
(BMP RAD) and turbulence (BMP TURB) individually and
then together (abbreviated with BRT for BMP RAD TURB). In
further simulations we include convective clouds (BRTC) and surface
fluxes (BRTCS). The convective activity is
much stronger in the simulation with surface fluxes than in the
simulation with the free-slip boundary condition. Hence, BRTCS
can be regarded as simulation with strong convection,
while BRTC can rather be seen as a life cycle with
weak to moderate convective activity. A final sensitivity study was
conducted in which the cloud radiative forcing has been neglected to
study the effect of this feedback in the region of the tropopause
(BMP RAD NOCRF).
Non-conservative processes and the formation of a TIL
in baroclinic life cycles
In a first step we aim to answer the question which non-conservative process,
i.e., related to clouds, radiation, or turbulent mixing, has the largest
impact on the formation of the TIL in baroclinic life cycles. For this we
compare first the results of four anticyclonic life cycles (REF,
TURB, RAD, and BMP), before we discuss the effects of
initial conditions and process formulations on the model results.
Impact of non-conservative processes on the TIL evolution
The baroclinic life cycle 1, also known as LC1, has been discussed under
various aspects e.g., and also in light of the
evolution of the tropopause inversion layer . Our REF
simulation features the same general characteristics of this life cycle and
is described in more detail in . One dominant feature of
the LC1 is the thinning trough, the so-called stratospheric streamer
often also referred to as Θ or PV streamer; e.g.,.
In the mature stage of the baroclinic wave this feature is evident for
instance in the distribution of potential temperature Θ on an
isosurface of potential vorticity, e.g., Q=2.0pvu. The
distribution of potential temperature for our four cases is shown in the
upper row of Fig. . After 120 h of model integration, we see
similar structures for REF, TURB, and RAD with minor
differences in the exact location of the streamer and the absolute values of
Θ in the warm sector (red colors). The most complex distribution
occurs in BMP with warmer temperatures than in the other three
simulations at the southern tip of the streamer. These warmer temperatures
are associated with cloud processes and the release of latent heat during
rapid ascent. Moreover, the entire Θ field shows a more in-homogeneous
appearance compared to the other three simulations.
Temporal evolution over the entire simulated life-cycles of
(a) the minimum surface pressure ps (in
hPa), (b) the maximum static stability Nmax2 (in
10-4s-2) above the thermal tropopause, and (c) the
area A5.5 (in 106km2) of N2
threshold exceedance above the thermal tropopause (with a threshold of
N2=5.5× 10-4s-2). The colored lines
indicate the following simulations: REF (blue), BMP (red),
RAD (orange), and TURB (gray).
Our main focus is, however, on the static stability N2 in the lowermost
stratosphere. In particular, we are interested in the regions where the
stability increases significantly during the life cycle. This is typically
the case within the first kilometer above the thermal tropopause. However,
the spatial appearance is not homogeneous, as is evident from the lower
panels in Fig. . These panels depict the vertical mean of N2 over
the first kilometer above the thermal tropopause. In all four cases large
values of N2 appear in the warm sector west of the streamer, which is in
the region of anticyclonic flow. This region has been shown to exhibit
a stronger TIL in models and in observations
. The life cycle with turbulence shows the lowest values of
N2, while the static stability has generally larger values in the case of
radiation than in the reference simulation. In the life cycle with cloud
processes we additionally see enhanced values of N2 on smaller scales than
in the other cases. As we will show later these enhancements are related to
moist dynamics and vertical motions.
The moist life cycle shows the strongest development in terms of minimum
surface pressure, ps, evolution, in contrast to the life cycle
with radiation (Fig. a). While all other life cycles still show
a deepening of ps, the absolute minimum pressure has already
been reached in BMP after 140 h of model integration.
Moreover, by considering two metrics to trace the evolution of the TIL in our
life cycles, we infer that the TIL formation differs most significantly from
the dry reference case in the moist life cycle. The maximum static stability
Nmax2 increases rather suddenly in BMP instead of more
gradually as in the other three simulations (Fig. b). After reaching
its absolute maximum value, Nmax2 keeps values above 7.0×10-4s-2 at consecutive times. Only after about 130 h
after model start Nmax2 in RAD, and a little bit later in
REF and TURB, has reached the same magnitude as in the moist
simulation. Furthermore, an earlier increase of Nmax2 is evident
in RAD than in REF and TURB, while in the latter case
Nmax2 is smaller than in the reference case at all times.
A similar picture is obtained from the metric that is used as a proxy for the
spatial extent of the TIL in the life cycles, i.e., the area in which N2>5.5×10-4s-2, denoted as A5.5 (Fig. c).
The earliest appearance is evident in BMP, the latest in TURB.
Moreover, the temporal evolution of A5.5 clearly shows that the TIL
covers a larger area when moist or radiative processes are included in the
life cycles. We also tested other thresholds for N2 for this metric with
no significant changes with respect to the qualitative interpretation of our
results.
So far, we provided a rather descriptive view on the TIL evolution in our
life cycles without giving details about the underlying processes. For the
case with turbulence the TIL appears weaker due to the tendency of turbulence
to reduce strong vertical gradients. Turbulence acts against the effects of
dry dynamics, which enhance the lower stratospheric stability during the life
cycle. Consequently, only a weak TIL forms in this case.
Including radiation results in a stronger TIL than in the reference case.
This is related to the radiative feedback of water vapor, which increases
over time in the region of the tropopause (Fig. a). Since no
microphysics is included in RAD, water vapor is transported as
a passive tracer in this simulation. Upward motions in the troposphere and
tropopause dynamics lead to more water vapor at the altitude of the
tropopause, finally changing the water vapor gradient significantly
(Fig. b). This causes differential cooling by water vapor in the
UTLS, which then results in a non-uniform change of the thermal structure
e.g.,. Additionally, recently lifted, moist air is then
partly located also in the lower stratosphere, where its residence time is
longer and thus can potentially affect the thermal structure over longer timescales. This process further enhances the static stability directly above the
tropopause and thus strengthens the TIL, which also forms by the dynamics of
the baroclinic wave. Thus, a process directly changing the thermal structure
alters the appearance of the TIL in the case with radiation.
Instantaneous thermal tropopause-based domain mean values of
(a) specific humidity
qv (in 10-6kgkg-1) and (b) the
vertical gradient of specific humidity ∂qv/∂z (in 10-6kgkg-1m-1)
for RAD. The domain mean is
calculated within 25–65∘ latitude and the entire
zonal domain. dztp is the distance to the height of
the thermal tropopause. The intensity of the gray colors indicates
the time since model start in 24 h intervals.
Temporal evolution between 30 and 80 h after simulation
start of (a) the maximum static stability Nmax2 (in
10-4s-2) above the thermal tropopause and (b) the
maximum temperature increment due to latent heating TLH (in
K) in the model domain for REF (blue lines), BMP
(red lines), and BMP SATAD (dark red
lines).
Instantaneous thermal tropopause-based domain mean values of ΔN2 (in 10-4 s-2) in the left panels and ΔQ (in pvu) in
the right panels for (a) REF, (b) TURB,
(c) RAD, and (d) BMP. The domain mean is
calculated within 25–65∘ latitude and the entire zonal domain. The
intensity of the gray colors indicates the time since model start in
24 h intervals. ΔN2 is the difference between the current
static stability N2 and the advected initial static stability N02,
ΔQ is the difference between the current potential vorticity Q and
the advected initial potential vorticity Q0. dztp
is the distance to the height of the thermal tropopause.
In the moist case we present evidence that a process at lower tropospheric
levels is responsible for the different appearance of the TIL. The
spontaneous increase in Nmax2 is well correlated with the
earliest release of latent heat in the model (Fig. a and b). Since
the same effect is evident from the simulation with the saturation adjustment
scheme (BMP SATAD), we can conclude that it is the release of latent
heat rather than a microphysical process being responsible for the observed
effect. Latent heat release is, however, a sign of not only condensation but
also fosters vertical motions in the model. These vertical motions reach in
many cases the tropopause and often lift this vertical transport barrier.
Consequently, also the air above is slightly lifted, thereby increasing the
vertical gradient of potential temperature, resulting in enhanced static
stability above the tropopause. This process differs, however, fundamentally
from the process related to dry dynamics on spatial and temporal scales.
While the latter is rather slow and occurs predominantly in an anticyclonic
flow region with on average descending air motion, this lifting process is
fast, occurs on small scales, and is related to upward motions. Thus taken
together, the incorporation of water in the model fosters a stronger TIL
development as consequence of enhanced upward motions within the life cycle
due to the release of latent heat. Our results agree with those obtained by
. They compared dry and moist baroclinic life cycles and
showed that including moisture leads to stronger updrafts as well as to a
faster evolution of the life cycle.
Although the temporal and spatial appearance of the TIL is rather
heterogeneous in all four simulations, the TIL becomes also evident in the
domain mean vertical profiles of N2. These averages are obtained between
25 and 65∘ N in the meridional direction and in the entire zonal
direction. ΔN2 represents the difference between the current N2
and the passively advected tracer N02 (Fig. , left panels) and
ΔQ the difference between the current potential vorticity Q and the
passively advected initial potential vorticity Q0 (Fig. , right
panels), respectively. The vertical profiles of ΔN2 and ΔQ
are given in a tropopause-based coordinate system for every 24 h of
the model integration and the thin solid line shows the location of the
tropopause. In all four simulations an increase in static stability forms
sooner or later during the life cycles just above the tropopause. While the
domain mean TIL appears only during the late stages in REF and
TURB, it is much earlier obvious in RAD and BMP.
However, PV at the tropopause shows significant positive changes only in the
simulation with radiation. The location of the maximum diabatic change in PV
correlates temporally and spatially (relative to the thermal tropopause) well
with the changing gradient of water vapor (see Fig. ). Moreover, this
change in PV occurs over large areas in the model domain (not explicitly
shown) and is thus clearly evident in the mean vertical profile of ΔQ. In simulations of real extratropical cyclones over the North Atlantic,
the evolution of a dipole structure with a positive PV anomaly above the
tropopause and a negative anomaly below have been reported by
. They could also show that these anomalies are largely
related the radiation scheme in their model. In contrast, only minor changes
of PV are found in the simulations with turbulence and cloud processes. In
the latter case the largest changes of PV occur rather at low- and
mid-tropospheric altitudes where the major release of latent heat occurs.
These changes occur, however, on smaller spatial areas, and more specifically
not always at the same altitude relative to the tropopause. Thus, compared to
RAD ΔQ has no pronounced tendency in the domain mean in the case
of BMP. In the reference case the minor changes of potential vorticity
are solely related to the numerics, especially to the tracer advection scheme
. Thus, in the case of radiation the formation of the TIL is
directly related to a diabatic process in the tropopause region, while the
diabatic processes related to clouds have an indirect impact on the TIL,
i.e., the diabatic processes and the response of the static stability above
the tropopause occur at a different places. Mixing, like radiation, also
directly affects the TIL but to a much lesser extent.
Sensitivity of individual diabatic processes
In the next paragraphs we briefly discuss the impact of initial conditions on
the model results, focusing especially on experiments with cloud microphysics
and radiation.
For microphysics we tested for the amount of initial specific humidity,
comparing BMP to BMP R30, and BMP R80, as well as for
the representation of the cloud processes, comparing BMP to BMP NOICE, and BMP SATAD. From the temporal evolution of
Nmax2 (Fig. a), we infer that the amount of specific
humidity is more important than the model formulation of cloud processes. If
more water is initially present, then the TIL appears earlier. In contrast,
with less initial water the TIL appears later and the entire appearance
approximates towards the adiabatic case. Moreover, the occurrence of the TIL
is relatively insensitive to the representation of the cloud processes as
long as the initial amount of specific humidity is the same as it is the case
in BMP, BMP NOICE, and BMP SATAD.
In the case of radiation we tested for the initial amount and distribution of
water, comparing RAD to RAD R30, and RAD woSW, as well
as for the amount of ozone, comparing RAD to RAD rO3. We find only minor differences in the evolution of
Nmax2 for the various sensitivity simulations (Fig. b).
Reducing the amount of water leads to a reduced radiative feedback and thus
to a less strong TIL. Changing the amount of ozone has, in our case, no
significant effect at all, however, with the caveat of the simple
representation of ozone in our model. The largest difference is found if we
completely remove the water in the stratosphere. This results in an
artificially large water vapor gradient between the troposphere and the
stratosphere. As we have seen before (Fig. ), a strong water vapor
gradient results in a sharp tropopause. A similar result has been discussed
by , who studied amongst many other features the response of
the static stability to the sharpness of a gradient between saturated and
unsaturated air.
Temporal evolution of the maximum static stability
Nmax2 (in 10-4s-2)
above the thermal tropopause for sensitivity simulations of
(a) BMP and (b) RAD. In (a)Nmax2
is shown for REF (blue), BMP (red),
BMP R30 (light red), BMP R80 (purple), BMP NOICE (magenta), and BMP SATAD (dark red). In (b)Nmax2 is shown for REF (blue),
RAD (orange), RAD woSW (coral), RAD R30 (dark orange), and RAD rO3 (brown).
Dynamical and thermodynamical state of baroclinic life cycles
after 120 h of model integration. In the upper rows of
the six panels, the distribution of potential temperature Θ
(in K) on the dynamical
tropopause is depicted, while the lower rows show the distribution
of static stability N2 (in 10-4s-2) averaged over
the first kilometer above the thermal tropopause for (a) BMP, (b) BMP RAD, (c) BMP TURB, (d) BRT,
(e) BRTC, and (f) BRTCS.
Static stability N2 (color-coded, in
10-4s-2) above the thermal tropopause, Δz
(black lines, in 2.5 km), column integrated cloud-ice
content tqi (blue lines, in 0.01 kgm-2),
and tropopause close column integrated cloud-ice content
tqi,tp (cyan lines, in
0.001 kgm-2). Tropopause close means the region
between the thermal tropopause and 500 m below. The
distribution is shown for BMP between 72 and
138 h after simulation start in a 6-hourly interval.
Relative importance of dynamical and diabatic processes on the
TIL formation
Until here we provided new insights of the isolated effect of individual
physical processes on the formation of the tropopause inversion layer in
baroclinic life cycles. Now we turn our discussion to the relative importance
of these processes, and especially whether the dynamical or the radiative
forcing is more important for the TIL formation and maintenance. For this
purpose we use our second set of baroclinic life-cycle experiments where we
successively increase the number of processes and as such increase
complexity. The simulation with cloud processes (BMP) serves as
reference while we first add radiation (BMP RAD) and turbulence
(BMP TURB) separately and then combine all three processes
(BRT). We further add convection (BRTC) and then also surface
fluxes of momentum and heat (BRTCS).
The six life cycles evolve similarly, all forming a Θ-streamer and
anticyclonic wave breaking. Again the temperature distribution at the
southern tip of the streamer varies most between the individual life cycles
(Fig. ). Moreover, in some cases a smooth Θ-distribution is
evident, e.g., BMP TURB, BRT, or BRTC, while the
distribution is more variable and shows more small-scale features in other
life cycles, especially in BRTCS. In all six cases the static
stability above the tropopause is larger in the anticyclonic part of the
wave than in the cyclonic part (not explicitly shown). After 120 h at
least two regions with enhanced values of N2 are evident. One is further
to the north along the cold front ahead of the cyclonic center. The other is
more located at the southwestern edge of the streamer. As evident from the
time series in Fig. both maxima are related to the outflow of the
warm conveyor belt (WCB). This airstream originates in the lower troposphere
in the region ahead of the trough axis e.g.,. In the
WCB, moist air masses rapidly ascend within 1–2 days into the upper
troposphere, associated with cloud formation, precipitation, and release of
latent heat e.g.,. The existence of a
relation between WCB and TIL has been proposed by , who used
HIRDLS satellite and ECMWF model data to obtain their results. Moreover,
Fig. shows that enhanced values of static stability above the
tropopause are closely related to the location of strong updrafts and cirrus
clouds at the time of the first TIL appearance. The cirrus clouds are
identified by the cloud-ice content below the tropopause. Note again that we
refer to updrafts here, when an air mass has been lifted by at least
2.5 km since model start. This change in altitude of an air parcel is
calculated from the difference of the current altitude z of this air parcel
and its initial altitude z0, which is carried by a passive
tracer. We further denote this difference as Δz, which is positive if
an air parcel raised and negative if an air parcel descended since model
start. The static stability is enhanced almost at all times in the center of
the WCB outflow, where the ice cloud branches towards the northwest and
southeast. From 102 h onward, a second maximum is evident in the
southeastern branch of the ice cloud, which moves further to the south in
subsequent hours. This maximum is located more in the region where
inertia-gravity waves are generated and influence the thermal structure of
the tropopause . This influence is such that the static
stability maximum keeps its large values almost entirely constant at
subsequent hours of the simulation. In the case of BRTCS, a larger area
exhibits enhanced static stability values above the tropopause, which is the
result of convective activity as we will see later in more detail.
Temporal evolution over the first 80 h of the life cycles
of (a)Nmax2 (in 10-4s-2) above the
thermal tropopause, (b) the
maximum of the Δz tracer (in km) in a 500 m
thick layer below the thermal tropopause, (c) the
maximum specific humidity qv in a 500 m thick layer below
the thermal tropopause (in 10-6kgkg-1), (d) the
maximum specific cloud-ice content qi in a 500 m thick layer below the thermal tropopause (in
10-6kgkg-1), and (e) the maximum cloud base
mass-flux ρCONV (in kgm-2s-1). The
time of TIL occurrence is split into three time sectors. Without
radiation and convection, the TIL appears after 65 h, with
radiation between 50 and 65 h, and with strong
convection before 50 h (more information is given in
the text). The colored lines indicate the following simulations: BMP (red),
BMP RAD (orange), BMP TURB (cyan), BRT (dark
cyan), BRTC (dark blue), and BRTCS (purple).
Zonal cross sections along 45∘ N of static
stability N2 (in 10-4s-2) after 120 h of model
integration. Red lines show specific cloud-ice content
qi (for 5.0×10-6kgkg-1), solid
blue lines show regions with positive values of
∂w/∂z (for 10.0×10-5s-1), dashed
blue lines show negative values (for -10.0×10-5s-1),
and solid gray lines show regions with Δz tracer larger
than 2.5km. The thick black line is the thermal
tropopause. The six panels show (a) BMP, (b) BMP RAD, (c) BMP TURB, (d) BRT, (e) BRTC,
and (f) BRTCS.
(a) Instantaneous thermal tropopause-based vertical
profiles of difference between the mean of static stability in regions with
Δz>2.5kmNdz2 and the domain mean N2 (in
10-4s-2) for each 24 h of the model integration.
(b) Differences for regions with ∂w/∂z≤-5.0×10-5s-1. The values in the top left corner of each
panel show the number of individual profiles used for calculating the
respective mean profile of Ndz2 and
Nwz2. dztp is the distance to the height of
the thermal tropopause.
In the following we aim to answer the question why the TIL appears earlier in
some life cycles and how the TIL is maintained after it has been generated.
We first compare the time of first appearance of the TIL between the six life
cycles. Figure a–e show the first 80 h of model
integration for various variables. The initial increase of Nmax2
can be divided into three sections, which are related to the physical
processes considered in the respective life cycle (Fig. a). The
latest TIL appearance after about 65 h is found when considering only
cloud processes and turbulence. Including radiation to the model simulations
shifts the time of appearance 10 h ahead, while the earliest TIL
formation starts already after about 35 h in the case of considering
convection and surface fluxes. This division into three time sectors
correlates well with the proxy for strong updrafts Δz.
Figure b depicts the maximum Δz in the layer between the
thermal tropopause and 500 m below this level, from which we infer
that there is strong temporal coincidence between the first appearance of
Nmax2 and updrafts originating at low levels. The earlier
appearance of vertically ascending air masses in the case with radiation and
convection is related to the these processes, since they foster an earlier
emerging of updrafts in the model. This finding supports our results from the
previous section that moist dynamics including stronger updrafts than in the
dry case has a strong impact on the first appearance of the TIL. These lifted
air masses further enhance the local convergence of the vertical wind just
above the tropopause as we will see later. Moreover, we find good agreement
between the temporal increase of Nmax2 and two tracers for
moisture, specific humidity qv (Fig. c) and specific
cloud-ice content qi (Fig. d). Thus, the ascending air
masses moisten the upper troposphere below the tropopause, which, as shown
before, supports the TIL formation by differential radiative cooling. The
gradual increase of Nmax2 in the case of BRTCS can further be
related to another tracer for updrafts, which is the cloud base mass flux
that is available for the two simulations in which the convective cloud
parameterization is switched on (Fig. e). This quantity serves as
proxy for convective activity and starts to increase gradually in the case
with surface fluxes early during the simulation. Thus, these findings further
support our suggestion from Sect. that vertical motions are
the essential key parameter for the initial TIL appearance in baroclinic life
cycles with moist diabatic processes.
We further provide evidence that there is not only a temporal but also
a spatial coincidence between updrafts and TIL occurrence. Figure
shows zonal cross sections of N2 for the six simulations along
45∘ N after 120h of model integration. Indications of
increased static stability are found in all cases above the lifted air
masses,
which reach the tropopause. Clouds often form in the regions of the updrafts
and in the lowermost stratosphere we find regions of convergence of the
vertical wind. This convergence results from emerging gravity waves from the
updrafts, but is also present in regions of propagating inertia gravity in
the eastern most region of the cross sections. Gravity waves can alter the
TIL temporarily during propagation and possibly
permanently by breaking or wave capture . In addition to
the effects of dry dynamics, i.e., distribution of cyclonic and anticyclonic
flow and breaking of the baroclinic wave see, the
effects from updrafts, small-scale convergence, and radiation, contribute
most strongly to the TIL formation. Furthermore, note that low- and
mid-tropospheric diabatic heating causes a negative change in PV above the
region of maximum heating, thus enhancing the anticyclonic flow in the
tropopause region above e.g.,, which further
has a positive feedback on the TIL evolution.
To this point we demonstrated that lifted air masses reaching the tropopause
level are initially important to form the TIL. However, this could be a
transient effect on the static stability in the stratosphere and as such its
contribution could decrease over time with other effects becoming more
important. One other potential process might be related to the convergence of
the vertical wind ∂w/∂z. If this term becomes negative at
or just above the tropopause, the static stability is increased in this
region . Convergence can occur on small scales when gravity
waves are present or on large scales in anticyclonic flow. We introduce here
another metric to measure the impact of updrafts and convergent regions on
enhanced static stability. For this we calculate the domain mean vertical
profile of static stability N2 as well as the mean vertical profile of
static stability in regions with strong updrafts Ndz2, i.e.,
Δz≥ 2.5km below the tropopause, and in regions with
strong convergence of the vertical wind Nwz2, i.e., ∂w/∂z≤-5.0×10-5s-1. We subtract the domain
mean from these values to obtain quantitative measures how strong the TIL is
enhanced in the respective regions compared to the TIL in the entire domain.
Figure shows the tropopause-based vertical profiles of
Ndz2-N2 (upper panel a) and Nwz2-N2 (lower
panel b) for every 24 h. In Ndz2-N2 a TIL like vertical
profile (i.e., with maximum values just above the tropopause) is evident in
all six cases, especially in the first days of the simulations. However, the
difference becomes smaller with time, which is partly related to the fact
that the TIL becomes more evident in the domain mean N2. Moreover, the
number of grid cells contributing to Ndz2 stagnates at later
times, indicating the decreasing number of new updrafts over time, which
reach the tropopause (compare the numbers in the top left corners in each
panel of Fig. ). The differences Nwz2-N2 also become
smaller above the tropopause with time, i.e., the TIL like shape is less
evident. However, compared to the relative decreases of the differences
Ndz2-N2, the decreases of Nwz2-N2 over time are
relatively smaller. Moreover, the number of grid cells contributing to
Nwz2 becomes significantly larger over time and is in most cases
also larger than the number for Ndz2. From this we follow that
updrafts might be potentially more important during the initial formation of
the TIL. In contrast, the convergence of the vertical wind might become
relatively more important in maintaining the TIL during later times of the
life cycles.
We already saw that moistening the upper troposphere fosters the evolution of
the TIL. Since ice clouds also reach the level of the tropopause, we briefly
discuss their potential impact on the thermal structure above the tropopause.
We only use cloud processes and radiation in this analysis here and exclude
the effects of mixing and convection. We conducted a further simulation in
which we turned off the cloud radiative feedback (BMP RAD NOCRF) and
compare the results to those from a simulation with feedback (BMP RAD)
to assess the impact of ice clouds on TIL in the model. From instantaneous
vertical profiles of meteorological and tracer quantities within a region,
which exhibits a TIL and ice clouds up to the tropopause, we infer the
following points (Fig. ): (1) the net heating rate is much more
negative in the upper troposphere when the forcing is turned on, with the
cooling being the strongest just below the thermal tropopause (black solid
lines); (2) the temperature profile in the UTLS differs significantly between
both cases – while there is a clear minimum in the case with cloud radiative
forcing, an almost neutral temperature profile is evident in the first two
kilometers above the tropopause in BMP RAD NOCRF (black dashed lines);
(3) the upper edge of the ice cloud is located slightly above the tropopause
in BMP RAD and slightly below in the other case (blue solid lines);
(4) the specific humidity has a local maximum at the top of the ice cloud,
which is stronger in the case with feedback (blue dashed lines); (5) the
static stability is increased in both cases with a slightly higher located
and stronger maximum in the case of feedback (red solid lines); and (6) the height
tracer indicates lifted air mass in the troposphere below the maximum of
static stability, however, with stronger updrafts in the case with feedback
(red dashed lines). From points (1), (2), and (5) we conclude that the
tropopause can be sharper due to strong differential cooling in the UTLS, if
ice clouds are present. Moreover, from (3), (4), and (6) it follows that the
potential to moisten the lower stratosphere is also increased, which might in
turn enhance the radiative formation process of the TIL. Thus, the results
from this sensitivity suggest that there is a larger potential to obtain
a stronger TIL when clouds reach up to the level of the tropopause. Moreover,
this might be of further interest, since ice clouds, or ice super-saturated
regions, have been shown to occur frequently in the lower stratosphere
e.g.,.
Tropopause-based vertical profiles through an ice cloud along
the central latitude at 120 h for (a) a simulation with
cloud radiative forcing (BMP RAD) and (b) a simulation
without cloud radiative forcing (BMP RAD NOCRF). Solid
lines show net radiative heating (in K d-1, scaled for better
comparability, black), cloud-ice content
(in 10-6kgkg-1, blue), and ΔN2
(in 10-4s-2, red). Dashed lines show temperature
(in K – 230 K, black),
specific humidity (in 10-5kgkg-1, blue), and
Δz (in km, red). dztp is the distance to the height of
the thermal tropopause.
So far we mainly focused on radiative and moist effects. In the last
paragraph we turn to the effect of mixing and analyze where turbulent mixing
occurs at the tropopause and whether this spatially and temporally coincides
with the appearance of the TIL. Turbulent mixing contributes to the process
of small-scale stratosphere–troposphere exchange (STE). It has been
speculated in several studies that TIL and STE are causally related beyond
a pure spatial coincidence (e.g., ).
used airborne measurements and ECMWF analysis data
from which they concluded that mixing at the tropopause is a synoptic-scale
process on rather short timescales, which, however, enhances the
concentration of radiatively active trace gases in the mixing layer. This
then leads to an increase in static stability further downwind of the region
of the STE event. Thus, they focused on the long-term relation between mixing
and N2. On the other hand we see that values of TKE are often increased in regions where a TIL is present (Fig. ).
These values are smaller than in the boundary layer, but nevertheless
increased compared to the background values in the tropopause region at other
locations and times in our model simulations. Such exchange events may only have
a spatial extension of a few tenths of kilometers or even less.
recently reported a comparable event based on airborne
in situ measurements of nitrous oxide, ozone, and ice cloud particles.
However, since our model is not capable of resolving this process with
sufficient accuracy to conduct a quantitative estimate of STE, we will leave
a more detailed analysis open to further studies.
The TIL in cyclonic life-cycle experiments
So far, the discussion of the results focused on the LC1
. We will now extend the analysis and show
results for five selected LC2. We obtain this life
cycle by adding a cyclonic shear to the background state of the LC1 (see
Sect. ). We briefly compare the results of the LC1 and LC2
and discuss the main difference in the following paragraphs. For this we
analyze the results from a dry reference experiment (REF LC2), from three
simulations with one additional diabatic process, i.e., with clouds (BMP
LC2), with radiation (RAD LC2), and with turbulence (TURB LC2), and from one
simulation with a more complex setup including clouds and convection,
radiation, and turbulence (BRTC LC2).
Generally, LC2 experiments show a less strong deepening of the minimum
surface pressure compared to their LC1 counterparts (Fig. a).
Similarly to the LC1 waves, the deepening of the surface cyclone is less
strong, when radiation is included in the simulations (RAD LC2, BRTC LC2).
Nmax2 above the thermal tropopause shows several differences
between LC1 and LC2. In the cases without moisture (REF LC2, RAD LC2, and
TURB LC2) the maximum values are always below
7.0× 10-4s-2. Moreover, in contrast to the sudden
increase of Nmax2 in all moist LC1 cases, Nmax2
increases rather stepwise, in particular in the BMP LC2 case. The absolute
maximum is reached only after 110 h after simulation start and thus
much later than in the LC1 BMP case (compare Fig. b). Furthermore, at
the end of the simulated period Nmax2 is almost equal in all
LC1 cases, which is, however, not the case in the LC2 cases. The TIL area
(A5.5, see Fig. c) is largest for BMP LC2 and shows even
comparable numbers to its LC1 counterpart. However, in the other cases the
A5.5 is much smaller in the LC2 cases than in the LC1 cases.
Thus, the TIL evolves less strong in amplitude and spatial extent in the LC2
compared to the LC1. Generally, this is in agreement with the results from
for dry adiabatic life cycles.
The processes relevant for the TIL formation are rather similar between LC1
and LC2. In the moist cases BMP LC2 and BRTC LC2 Nmax2 shows a
strong correlation to Δz (see Fig. d) and thus updrafts may
be as important in the LC2 as they are in the LC1 to initially form the TIL
in the life cycles. This relation is further obvious when the spatial
co-occurrence between lifted air masses and enhanced static stability is
studied (Fig. ). The first enhancement of N2 in the lower
stratosphere are again present just above regions that exhibit strong
updrafts and also ice clouds just below the tropopause. Thus, except for the
difference in the timing of the first vertical ascent patterns, there is no
major difference to the LC1 baroclinic life cycle. However, the temporal
variability of Nmax2 in BMP LC2 and BRTC LC2 is slightly
larger than in their LC1 counterparts. This might be related to the less
strong evolving gravity waves in the LC2 simulations. In particular, gravity
waves from the jet-front system are much more evident in LC1 than in LC2,
which has been discussed in . Thus, the effect of gravity
waves on the TIL maintenance might be less strong in the case of LC2. Taken
together the LC2 cases generally show a less strong developed TIL compared
with their LC1 counterparts. Nevertheless, the physical processes leading the
TIL formation seem to be similar in LC1 and LC2.
Conclusions and summary
By conducting various simulations of baroclinic life cycles we aimed to
improve the understanding of whether dynamical or diabatic processes are more
relevant to form a tropopause inversion layer (TIL). For this we used the
non-hydrostatic, limited area model COSMO in a midlatitude channel
configuration along with a varying number of physical parameterizations. We
first analyzed the effect of individual diabatic processes, i.e., related to
clouds, radiation, and mixing processes before we estimated the relative
importance of each process.
Zonal cross sections along 45∘ N of static
stability N2 (in 10-4s-2) after 144 h of model
integration. Solid blue lines show regions with positive values of
the vertical divergence ∂w/∂z (for 5.0 × 10-5 s-1, 50.0 × 10-5 s-1), dashed blue lines show negative
values (for -5.0 × 10-5 s-1, 50.0 × 10-5 s-1). Red lines show specific cloud-ice content
qi (for 5.0 × 10-6 kg kg-1, 10 × 10-6 kg kg-1). Gray lines show turbulent kinetic energy (TKE) (in
0.5, 1.0, 5.0 m2s-2). The four
panels show (a) BMP TURB, (b) BRT,
(c) BRTC, and (d) BRTCS.
In a first set of simulations the evolution of the TIL has been compared in
baroclinic life cycles. A life-cycle experiment with only dry dynamics served
as reference case, while three additional life-cycle experiments have been
performed with individual non-conservative processes added. We further assessed
the impact of initial conditions and process formulation in the diabatic
cases. In a second step, we successively increased the number of processes to
assess the relative importance of the various dynamical and diabatic
processes to the TIL evolution. We further conducted sensitivity experiments
to study differences between life cycles of type 1 (LC1) and 2 (LC2).
Temporal evolution over the entire simulated life cycles of
(a) the minimum surface pressure ps (in
hPa), (b) the maximum static stability Nmax2 (in
10-4s-2) above the thermal tropopause, (c) the
area A5.5 (in 106km2) of N2
threshold exceedance above the thermal tropopause (with a threshold of
N2=5.5×10-4s-2), and (d) the
maximum of the Δz tracer (in km) in a 500 m
thick layer below the thermal tropopause.
The colored lines indicate the following simulations: REF LC2
(blue), BMP LC2 (red), RAD LC2 (orange), TURB LC2
(gray), and BRTC LC2 (dark blue).
As Fig. , but for BMP LC2.
Most importantly, our experiments highlighted the role of different moisture
related processes for the formation and evolution of the TIL with varying
relevance and strength in different phases of the baroclinic life cycles. In
detail, we derived the following results:
A TIL forms in baroclinic life cycles with only dry dynamics as
well as in life cycles with additionally either vertical turbulence,
cloud processes, or radiation. Compared to the dry reference case the TIL
appears weaker with respect to its maximum value as well as to the
spatial appearance in the case of turbulence. The opposite is
evident in the case of radiation with a larger maximum static stability
and larger spatial appearance. The temporal evolution is,
however, still similar to the reference case. This is different with
cloud processes. The TIL emerges much earlier and shows generally the
largest maximum values and spatial extension.
The processes forming the TIL in the cases with diabatic processes
are as follows. Turbulence acts against the forming
process from dynamics, and as such a weaker TIL is the final
result. With only radiative processes,
the (passive) transport of moisture from low to high levels leads to
an increase in the moisture burden in the UTLS and to a change in the
moisture gradient in this region. The UTLS is then cooled
non-uniformly, which finally further enhances the static stability above
the tropopause. The important process with clouds is the release of latent heat
during condensation. This increases the frequency and strength of
vertical motions, which locally increase the static stability above the
regions of the updrafts. Especially, the TIL forms in the
region of the warm conveyor belt. In contrast to the direct diabatic
forcing (occurring in the region of the tropopause) in the case with
radiation, the enhancement of static stability results from a diabatic
forcing at lower levels in the case with clouds.
Analysis of initial conditions and process formulations
showed that the TIL formation in the model is relatively insensitive to
the formulation of the cloud forming process itself and more dependent
on the initial amount of specific humidity. For radiation no significant dependency
on the initial water or ozone amount is evident. Here, the change of
the gradient of specific humidity is the more important process.
Further simulations of baroclinic life cycles with varying
complexity with respect to the number of incorporated physical
processes showed that there is a correlation between the first
appearance of the TIL and of updrafts reaching the
tropopause. However, the exact timing of this first
occurrence further depends on the included physical processes. The TIL
emerges latest when only cloud processes and turbulence are considered while
it appears earlier when radiation is incorporated and even more
with convection. From this result it is concluded that vertically
ascending air masses, which reach the tropopause altitude, are the
key process in the initial formation of the TIL in moist baroclinic
life cycles; however, noting that their effect is probably
fading with time.
The updrafts that reach the tropopause lead to the emission of gravity
waves in the lower stratosphere. Such small-scale waves have a further
source in the jet-front system (inertia-gravity waves). In recent studies
e.g., it has been shown that
these small-scale disturbances can alter the thermal structure above
the tropopause temporarily as well as permanently and as such affect
the TIL during the entire life cycle after their first
appearance. At least in parts, the appearance and strength of such
gravity waves might explain the weaker appearance of the TIL in the
cyclonic life cycles compared to their anticyclonic counterparts.
Finally, air masses lifted from moist, low-tropospheric regions
enhance the moisture content of the upper troposphere, not only by
transporting water vapor to this altitude. Clouds also form within the
updrafts and locally alter the thermal structure of the upper
troposphere. Especially, at the top of the clouds a strong cooling can
occur, which further contributes to the formation and maintenance of a
strong TIL. In general, radiative impacts become more relevant during
later stages of the life cycle.
Thus, the various dynamical and diabatic processes lead to a highly variable
temporal and spatial appearance of the TIL on the timescale of a week. While
updrafts are important for the first appearance of the TIL when moisture is
included in the baroclinic life cycles, the radiative effects as well as the
convergence of the vertical wind are more important in maintaining the TIL
during later phases of the life cycles. In reality the TIL in the
midlatitudes may be restrengthened by each passing baroclinic wave and the
lifted water vapor serves as a cooling agent in the upper troposphere and
even in the lower stratosphere over a longer timescale than a week. Taking
into account that baroclinic waves occur relatively frequent at midlatitudes,
especially from autumn to spring, might further help to explain the
quasi-permanent appearance of a layer of enhanced static stability.
Acknowledgements
D. Kunkel acknowledges funding from the German Science Foundation under
grant HO 4225/2-1. The authors thank A. Roches, U. Blahak, and S. Schemm
for model support and the HPC team of the university of Mainz for
computing time. We further thank P. Spichtinger for valuable comments on
an earlier version of the manuscript. The comments on the discussion
paper from H. Wernli, S. Schemm, G. Craig, and an anonymous referee helped to
significantly improve the final manuscript. Further information on
data (model code and output)
relevant to this paper can be obtained upon request via email to the
authors (dkunkel@uni-mainz.de). Edited by: H. Wernli
ReferencesBell, S. W. and Geller, M. A.: Tropopause inversion layer: seasonal and
latitudinal variations and representation in standard radiosonde data and
global models, J. Geophys. Res.-Atmos., 113, D05109,
doi:10.1029/2007JD009022,
2008.Birner, T.: Fine-scale structure of the extratropical tropopause region, J.
Geophys. Res., 111, D04104,
doi:10.1029/2005JD006301,
2006.Birner, T.: Residual Circulation and Tropopause Structure, J. Atmos. Sci.,
67, 2582–2600,
doi:10.1175/2010JAS3287.1,
2010.Birner, T., Dörnbrack, A., and Schumann, U.: How sharp is the tropopause at
midlatitudes?, Geophys. Res. Lett., 29, 1700,
doi:10.1029/2002GL015142,
2002.Birner, T., Sankey, D., and Shepherd, T. G.: The tropopause inversion layer
in models and analyses, Geophys. Res. Lett., 33, L14804,
doi:10.1029/2006GL026549,
2006.
Carlson, T. N.: Mid-Latitude Weather Systems, Am. Meteorol. Soc., 507 pp.,
1998.Chagnon, J. M., Gray, S. L., and Methven, J.: Diabatic processes
modifying potential vorticity in a North Atlantic cyclone,
Q. J. Roy. Meteor. Soc., 139, 1270–1282,
doi:10.1002/qj.2037, 2013.
Doms, G.: A Description of the Nonhydrostatic Regional COSMO-Model, Part I:
Dynamics and Numerics, Tech. rep., Deutscher Wetterdienst, Offenbach,
Germany, 2011.Erler, A. R. and Wirth, V.: The static stability of the tropopause region in
adiabatic baroclinic life cycle experiments, J. Atmos. Sci., 68, 1178–1193,
doi:10.1175/2010JAS3694.1,
2011.Fusina, F. and Spichtinger, P.: Cirrus clouds triggered by radiation, a
multiscale phenomenon, Atmos. Chem. Phys., 10, 5179–5190,
10.5194/acp-10-5179-2010, 2010.Gettelman, A. and Wang, T.: Structural diagnostics of the tropopause
inversion layer and its evolution, J. Geophys. Res.-Atmos., 120, 46–62,
doi:10.1002/2014JD021846,
2015.Grise, K. M., Thompson, D. W. J., and Birner, T.: A global survey of static
stability in the stratosphere and upper troposphere, J. Climate, 23,
2275–2292,
doi:10.1175/2009JCLI3369.1,
2010.Gutowski, W. J., Branscome, L. E., and Stewart, D. A.: Life cycles of moist
baroclinic eddies, J. Atmos. Sci., 49, 306–319,
doi:10.1175/1520-0469(1992)049<0306:LCOMBE>2.0.CO;2,
1992.Held, I. M. and Suarez, M. J.: A proposal for the intercomparison of the
dynamical cores of atmospheric general circulation models, B. Am. Meteorol.
Soc., 75, 1825–1830,
doi:10.1175/1520-0477(1994)075<1825:APFTIO>2.0.CO;2,
1994.Joos, H. and Wernli, H.: Influence of microphysical processes on the
potential vorticity development in a warm conveyor belt: a case-study with
the limited-area model COSMO, Q. J. Roy. Meteor. Soc., 138, 407–418,
doi:10.1002/qj.934, 2012.Kunkel, D., Hoor, P., and Wirth, V.: Can inertia-gravity waves persistently
alter the tropopause inversion layer?, Geophys. Res. Lett., 41, 7822–7829,
doi:10.1002/2014GL061970,
2014.Kunz, A., Konopka, P., Müller, R., Pan, L. L., Schiller, C., and
Rohrer, F.: High static stability in the mixing layer above the extratropical
tropopause, J. Geophys. Res., 114, D16305,
doi:10.1029/2009JD011840,
2009.
Madonna, E., Wernli, H., Joos, H., and Martius, O.: Warm conveyor belts in
the ERA-Interim data set (1979–2010). Part I: Climatology and potential
vorticity evolution, J. Climate, 27, 3–26, 2014Mellor, G. L. and Yamada, T.: Development of a turbulence closure model for
geophysical fluid problems, Rev. Geophys., 20, 851–875,
doi:10.1029/RG020i004p00851,
1982.Müller, S., Hoor, P., Berkes, F., Bozem, H., Klingebiel, M., Reutter, P.,
Smit, H. G. J., Wendisch, M., Spichtinger, P., and Borrmann, S.: In situ
detection of stratosphere-troposphere exchange of cirrus particles in the
midlatitudes, Geophys. Res. Lett., 42, 949–955,
doi:10.1002/2014GL062556,
2015.Murray, F. W.: On the computation of saturation vapor pressure, J. Appl.
Meteorol., 6, 203–204,
doi:10.1175/1520-0450(1967)006<0203:OTCOSV>2.0.CO;2,
1967.Olson, J. B. and Colle, B.: A modified approach to initialize an idealized
extratropical cyclone within a mesoscale model, Mon. Weather Rev., 135,
1614–1624,
doi:10.1175/MWR3364.1, 2007.Otsuka, S., Takeshita, M., and Yoden, S.: A numerical experiment on the
formation of the tropopause inversion layer associated with an explosive
cyclogenesis: possible role of gravity waves, Prog. Earth Planet. Sc., 1, 19,
doi:10.1186/s40645-014-0019-0,
2014.Peevey, T. R., Gille, J. C., Homeyer, C. R., and Manney, G. L.: The double
tropopause and its dynamical relationship to the tropopause inversion layer
in storm track regions, J. Geophys. Res.-Atmos., 119, 10194–10212,
doi:10.1002/2014JD021808,
2014.Randel, W. J. and Wu, F.: The Polar summer tropopause inversion layer, J.
Atmos. Sci., 67, 2572–2581,
doi:10.1175/2010JAS3430.1,
2010.Randel, W. J., Wu, F., and Forster, P.: The extratropical tropopause
inversion layer: global observations with GPS data, and a radiative forcing
mechanism, J. Atmos. Sci., 64, 4489–4496,
doi:10.1175/2007JAS2412.1,
2007.Riese, M., Ploeger, F., Rap, A., Vogel, B., Konopka, P., Dameris, M., and
Forster, P.: Impact of uncertainties in atmospheric mixing on simulated UTLS
composition and related radiative effects, J. Geophys. Res., 117, D16305,
doi:10.1029/2012JD017751,
2012.Ritter, B. and Geleyn, J.-F.: A comprehensive radiation scheme for numerical
weather prediction models with potential applications in climate simulations,
Mon. Weather Rev., 120, 303–325,
doi:10.1175/1520-0493(1992)120<0303:ACRSFN>2.0.CO;2,
1992.Schemm, S., Wernli, H., and Papritz, L.: Warm conveyor belts in idealized
moist baroclinic wave simulations, J. Atmos. Sci., 70, 627–652,
doi:10.1175/JAS-D-12-0147.1,
2013.Son, S.-W. and Polvani, L. M.: Dynamical formation of an extra-tropical
tropopause inversion layer in a relatively simple general circulation model,
Geophys. Res. Lett., 34, L17806,
doi:10.1029/2007GL030564,
2007.Spang, R., Günther, G., Riese, M., Hoffmann, L., Müller, R., and
Griessbach, S.: Satellite observations of cirrus clouds in the Northern
Hemisphere lowermost stratosphere, Atmos. Chem. Phys., 15, 927–950,
10.5194/acp-15-927-2015, 2015.Spichtinger, P., Gierens, K., Leiterer, U., and Dier, H.: Ice supersaturation
in the tropopause region over Lindenberg, Germany, Meteorol. Z., 12,
143–156,
doi:10.1127/0941-2948/2003/0012-0143,
2003.Sprenger, M., Croci Maspoli, M., and Wernli, H.: Tropopause folds and
cross-tropopause exchange: a global investigation based upon ECMWF analyses
for the time period March 2000 to February 2001, J. Geophys. Res.-Atmos.,
108, 8518,
doi:10.1029/2002JD002587,
2003.
Steppeler, J., Doms, G., Schättler, U., Bitzer, H. W., Gassmann, A.,
Damrath, U., and Gregoric, G.: Meso-gamma scale forecasts using the
nonhydrostatic model LM, Meteorol. Atmos. Phys., 82, 75–96,
doi:10.1007/s00703-001-0592-9,
2003.Thorncroft, C. D., Hoskins, B. J., and McIntyre, M. E.: Two paradigms of
baroclinic-wave life-cycle behaviour, Q. J. Roy. Meteor. Soc., 119, 17–55,
doi:10.1002/qj.49711950903,
1993.Tiedtke, M.: A comprehensive mass flux scheme for cumulus parameterization in
large-scale models, Mon. Weather Rev., 117, 1779–1800,
doi:10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2,
1989.Wernli, H. and Davies, H. C.: A Lagrangian-based analysis of extratropical
cyclones. I: The method and some applications, Q. J. Roy. Meteor. Soc., 123,
467–489,
doi:10.1002/qj.49712353811,
1997.Wirth, V.: Static stability in the extratropical tropopause region, J. Atmos.
Sci., 60, 1395–1409,
doi:10.1175/1520-0469(2003)060<1395:SSITET>2.0.CO;2,
2003.
Wirth, V.: A dynamical mechanism for tropopause sharpening, Meteorol. Z., 13,
477–484, 2004.
WMO: Meteorology – a three dimensional science, WMO Bulletin, 134–138,
1957.Zierl, B. and Wirth, V.: The influence of radiation on tropopause behavior
and stratosphere-troposphere exchange in an upper tropospheric anticyclone,
J. Geophys. Res., 102, 23883,
doi:10.1029/97JD01667, 1997.