Detailed Modeling of Mountain Wave PSCs

Abstract

Polar stratospheric clouds (PSCs) play a key role in polar ozone depletion. In the Arctic, PSCs can occur on the mesoscale due to orographically induced gravity waves. Here we present a detailed study of a mountain wave PSC event on 25-27 January 2000 over Scandinavia. The mountain wave PSCs were intensively observed by in-situ and remote-sensing techniques during the second phase of the SOLVE/THESEO-2000 Arctic campaign. We use these excellent data of PSC observations on 3 successive days to analyze the PSCs and to perform a detailed comparison with modeled clouds. We simulated the 3-dimensional PSC structure on all 3 days with a mesoscale numerical weather prediction (NWP) model and a microphysical box model (using best available nucleation rates for ice and nitric acid trihydrate particles). We show that the combined mesoscale/microphysical model is capable of reproducing the PSC measurements within the uncertainty of data interpretation with respect to spatial dimensions, temporal development and microphysical properties, without manipulating temperatures or using other tuning parameters. In contrast, microphysical modeling based upon coarser scale global NWP data, e.g. current ECMWF analysis data, cannot reproduce observations, in particular the occurrence of ice and nitric acid trihydrate clouds. Combined mesoscale/microphysical modeling may be used for detailed a posteriori PSC analysis and for future Arctic campaign flight and mission planning. The fact that remote sensing alone cannot further constrain model results due to uncertainities in the interpretation of measurements, underlines the need for synchronous in-situ PSC observations in campaigns.

Full text

Page 1

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
Atmos. Chem. Phys. Discuss., 3, 253–299, 2003
www.atmos-chem-phys.org/acpd/3/253/
c ? European Geosciences Union 2003
Atmospheric
Chemistry
and Physics
Discussions
Detailed modeling of mountain wave
PSCs
S. Fueglistaler1, S. Buss1, B. P. Luo1, H. Wernli1, H. Flentje2, C. A. Hostetler3,
L. R. Poole3, K. S. Carslaw4, and Th. Peter1
1Atmospheric and Climate Science, ETH Z¨ urich, Switzerland
2DLR Oberpfaffenhofen, 82230 Wessling, Germany
3NASA Langley Research Center, Hampton, VA, USA
4School of the Environment, Univ. of Leeds, Leeds, UK
Received: 6 November 2002 – Accepted: 27 December 2002 – Published: 13 January 2003
Correspondence to: S. Fueglistaler (stefanf@atmos.umnw.ethz.ch)
253

Page 2

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
Abstract
Polar stratospheric clouds (PSCs) play a key role in polar ozone depletion. In the Arc-
tic, PSCs can occur on the mesoscale due to orographically induced gravity waves.
Here we present a detailed study of a mountain wave PSC event on 25–27 January
2000 over Scandinavia. The mountain wave PSCs were intensively observed by in-situ
and remote-sensing techniques during the second phase of the SOLVE/THESEO-2000
Arctic campaign. We use these excellent data of PSC observations on 3 successive
days to analyze the PSCs and to perform a detailed comparison with modeled clouds.
We simulated the 3-dimensional PSC structure on all 3 days with a mesoscale numeri-
cal weather prediction (NWP) model and a microphysical box model (using state-of-the-
art nucleation rates for ice and nitric acid trihydrate particles). We show that the com-
bined mesoscale/microphysical model is capable to reproduce the PSC measurements
within the uncertainty of data interpretation with respect to spatial dimensions, temporal
development and microphysical properties, without manipulating temperatures or us-
ing other tuning parameters. In contrast, microphysical modeling based upon coarser
scale global NWP data, e.g. current ECMWF analysis data, cannot reproduce obser-
vations, in particular the occurrence of ice and nitric acid trihydrate clouds. Combined
mesoscale/microphysical modeling may be used for detailed a posteriori PSC analysis
and for future Arctic campaign flight and mission planning. The fact that remote sensing
alone cannot further constrain model results due to uncertainities in the interpretation
of measurements, underlines the need for synchronous in-situ PSC observations in
campaigns.
5
10
15
20
1. Introduction
Polar stratospheric clouds (PSCs) play a key role in polar ozone depletion. On the
aerosol surface, heterogeneous chemical reactions take place that activate chlorine
from inert reservoir species (ClONO2, HCl) to Cl2, which is rapidly photolyzed into
25
254

Page 3

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
atomic chlorine radicals (Solomon et al., 1986; Tolbert et al., 1987):
ClONO2+ HCl −→ Cl2+ HNO3
and
(1)
HCl + HOCl −→ Cl2+ H2O
Cl2+ hν −→ Cl + Cl
The chlorine radicals then destroy ozone in catalytic cycles (Molina and Molina , 1987;
McElroy et al., 1986).
In addition, PSCs can permanently remove HNO3through sedimentation, a pro-
cess termed denitrification, which enhances halogen-catalyzed chemical reactions
and hence further promotes ozone destruction (WMO, 1999; Waibel et al., 1999;
Tabazadeh et al., 2000; Gao et al., 2001).
In contrast to Antarctica where a stable vortex prevails each winter, in the Arctic
the vortex is less stable and synoptic temperatures remain often above the tempera-
tures required for PSC formation. However, local mechanisms such as orographically
induced gravity waves that propagate into the stratosphere can lead to adiabatic cool-
ing (and warming). In the cold phases of these waves PSCs may form even though
synoptic scale temperatures are too high. These so-called mountain wave PSCs are
located in the coldest regions over and in the lee of mountains, and often are quasi-
stationary (i.e. change slowly compared with the time-scale given by the passage of
the air through the region where temperatures allow PSC formation). Repeated occur-
rence of mountain wave PSCs can ‘process’ a significant fraction of the air in the polar
vortex, despite their relatively small spatial dimension (Carslaw et al., 1998b).
Based on lidar observations by Browell et al. (1990), Toon et al. (1990) identified
three distinct types of PSCs. Type 1a and 1b both show low to moderate backscatter
ratios (BSR). In contrast to type 1b which was later identified as supercooled ternary
(HNO3/H2SO4/H2O) solution (STS) droplets with number densities ∼ 10 cm−3(Carslaw
255
(2)
(3)
5
10
15
20
25

Page 4

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
Back Close
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
et al., 1994), type 1a shows moderate aerosol depolarization (δaerosol), which Biele
et al. (2001) interpreted as containing small number densities (n ? 10−2cm−3) of as-
pherical particles, most likely nitric acid trihydrate (NAT). Type 2 PSCs show high BSRs
and moderate to high depolarisation, attributed to water ice particles. Later, Tsias et al.
(1999) identified type 1a-enh PSCs as similar to type 1a but with enhanced NAT parti-
cle number densities (n ? 0.1 cm−3). First evidence from in-situ measurements for the
presence of NAT particles in PSCs was given by Voigt et al. (2000) from a balloon-borne
experiment during the SOLVE/THESEO-2000 campaign. During the same campaign,
Fahey et al. (2001) detected for the first time very large HNO3-containing (presumably
NAT) particles in very low number densities (so-called ‘NAT-rocks’) which can efficiently
denitrify the polar stratosphere. Apart from ternary solution droplets and NAT the exis-
tence of other HNO3containing hydrates, such as nitric acid dihydrate (NAD), is also
still under debate.
Carslaw et al. (1998a) and Wirth et al. (1999) successfully modeled lidar measure-
ments of mountain wave PSCs from flights that were parallel to the wind direction (so-
called quasi-Lagrangian measurements). They used a microphysical box model that
calculated the PSC microphysics along a manually inferred trajectory, and without ex-
act knowledge of ice or NAT nucleation rates. Using T-Matrix calculations for aspherical
solid particles (Mishchenko , 1991) and Mie calculations with an index of refraction as
a function of aerosol composition, temperature and wavelength for the liquid aerosol
(Krieger et al., 2000), it was shown that the simulated lidar signal inferred from the
modeled microphysics could be brought into good agreement with the observations,
provided that the trajectory was chosen carefully and that the number densities of ice
and NAT particles were chosen within plausible limits.
5
10
15
20
2.Mesoscale/microphysical modeling approach
25
Here we significantly improve the method of Carslaw et al. (1998a) and Wirth et al.
(1999) and extend its scope. Instead of manually inferred trajectories we use tra-
256

Page 5

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
jectories from mesoscale numerical weather prediction (NWP) models. This greatly
reduces the degree of freedom of the model. Trajectories from NWP models further-
more allow us to simulate the 3-dimensional structure of PSCs, and allow predictions
of PSC occurrences where no measurements are available. We use trajectories from
the mesoscale High Resolution Model (HRM, cf. Section 3.5) and compare them with
trajectories from ECMWF analysis data.
An updated version of the microphysical box model is used, which calculates the
nucleation of ice and NAT particles. We use the nucleation rates given by Koop et al.
(2000) for the homogeneous freezing of ice particles and those provided by Luo et al.
(2002) for the NAT nucleation on ice surfaces exposed to the gas phase, instead of
adjusting the solid particle number densities manually as was done by Carslaw et al.
(1998a) and Wirth et al. (1999). The approach chosen in this study uses the current
understanding of numerical weather prediction models and microphysical processes in
a self-consistent way without any ‘tuning’, such that the simulations can be run fully
automatically. This enables to obtain the morphology of PSCs in 2-D or 3-D calcula-
tions instead of focusing on one or two selected trajectories. We compare the results
of these simulations to measurements allowing to judge the accuracy of the state-of-
the-art modeling.
The combined mesoscale/microphysical model is applied to the period 25–27 Jan-
uary over Scandinavia. The data acquired in this period during the SOLVE/THESEO-
2000 Arctic campaign provide probably the most complete collection of data of a moun-
tain wave PSC event. For all three successive days the simulations were performed
with exactly the same setup, showing the robustness of the approach.
5
10
15
20
257

Page 6

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
3.Data and tools
3.1.Synoptic situation, 25–27 January 2000
The meteorological situation leading to the mountain wave PSCs on 25–27 January
over Scandinavia was discussed by D¨ ornbrack et al. (2002). Figure 1 shows the po-
tential vorticity (PV) field on the 345 K isentropic surface, wind speed in the upper
troposphere (250 hPa) and wind vectors in the lower troposphere (900 hPa) for 25–
27 January 2000. The isentropic PV charts are well suited, for instance, to identify
large-scale upper-tropospheric anticyclones (characterized by PV values < 2pvu) and
jet-stream regions (characterized by strong horizontal PV gradients). In agreement
with Doernbrack et al. we note for 25 January the existence of an extreme upper-level
anticyclone over the UK, Iceland and Norway with strong low-level winds towards north-
ern Norway and a northerly upper tropospheric jet over Scandinavia. In the course of
the 3 days the anticyclone shifts southward and on 26 January a very strong jet is
directly over central Scandinavia. The jet is almost parallel to strong near surface west-
erlies, which provides very favourable conditions for excitation and vertical propagation
of mountain waves. On 27 January the anticyclone shifted further south, leaving rela-
tively weak low- and upper-level winds over Scandinavia, except for the southernmost
part with upper-level wind speeds ∼ 50m/s.
3.2.PSC measurement
5
10
15
In this study lidar data is used from the NASA LaRC Aerosol Lidar (a piggy-back instru-
ment to the NASA GSFC AROTEL lidar) on board the NASA DC-8 aircraft (532 nm,
1064 nm) and the DLR OLEX lidar system on board the Falcon aircraft (354 nm,
532 nm, 1064 nm) (Flentje et al., 1999). Both systems measure total backscatter
at all wavelengths and depolarization at 532 nm, which allows discrimination of spher-
ical (STS) and aspherical (e.g. NAT and ice) particles. In addition, we will refer to the
results of the balloon-borne measurements of 25 January (Voigt et al., 2000) and the
20
25
258

Page 7

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
Back Close
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
in-situ measurements on board the stratospheric research aircraft ER-2 on 27 January
(Northway et al., 2002). Table 1 provides an overview of the platforms in operation on
25-27 January and from the data used in this study. The data is analyzed in Sect. 4.
3.3.Analysis of lidar data
A lidar system emits a laser pulse and measures the backscatter from aerosols and air
molecules as a function of time. The backscatter ratio BSR is defined as
BSR = 1 +βaerosol
βair
where βaerosolis the backscatter coefficient of aerosol and βairis the backscatter
coefficient of air molecules (Rayleigh-scattering). The lidar system measures the total
backscatter coefficient βtotal= βaerosol+ βair, and in order to obtain the backscatter
ratio, the backscatter coefficient βairhas to be calculated as a function of air density.
The depolarization of the scattered laser pulse yields information about the shape of
the scatterer. The aerosol depolarization is defined as:
βaerosol
⊥
βaerosol
?
where βaerosol
⊥
parallel aerosol backscatter coefficient. The color ratio is defined as the ratio of the
aerosol backscattering coefficient at two wavelengths λ1, λ2with λ1< λ2:
βaerosol
λ1
βaerosol
λ2
The color ratio is very sensitive to the size of the scatterers, but independent of
the number of scatterers. We will use the following notation: the backscatter ratio at
259
5
,
(4)
10
δaerosol=
(5)
is the perpendicular aerosol backscatter coefficient and βaerosol
?
is the
15
CR(λ1/λ2) =
(6)
20

Page 8

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
1064nm ≡ BSR(1064), the color ratio βaerosol
depolarization at 532nm ≡ δaer(532).
Based on long term lidar observations over Ny Alesund, Spitsbergen (78.9◦N,
11.9◦E) and T-Matrix calculations, Biele et al. (2001) presented a classification of li-
dar data. They classify PSCs into the following types: type 1a (aspherical particles,
most likely NAT, low particle number densities n ≈ 10−2cm−3), type 1a-enh (aspher-
ical particles, most likely NAT, high particle number densities n ? 0.1cm−3), type 1b
(ternary HNO3/H2SO4/H2O aerosol droplets (STS), n ≈ 10cm−3), type mix (spheri-
cal and aspherical particles, likely NAT and STS particles mixed externally and not in
thermodynamic equilibrium) and type 2 (ice particles, n =1-10cm−3). We will use this
classification as a reference in our lidar data interpretation.
The backscatter of aspherical particles may be calculated using the T-Matrix method
(Mishchenko , 1991) (with spheroids of aspect ratio 0.5-1.5 as proxies for ice and NAT
particles). The index of refraction is 1.31 for ice and 1.48 for NAT (Middlebrook et al.,
1994; Toon et al., 1994), that of STS obtained from Krieger et al. (2000).
532
/βaerosol
1064
≡ CR(532/1064), and the aerosol
5
10
15
3.4.Microphysical modeling
A microphysical box model is used to calculate the evolution of the aerosol along a
trajectory. The thermodynamics governing the condensation/evaporation kinetics of
water and nitric acid by the aqueous sulfuric acid aerosol is calculated from an ion-
interaction model (Pitzer , 1991; Luo et al., 1995; Meilinger et al., 1995). The aerosol
size distribution is assumed to be log-normal with an initial mode radius of 0.06µm and
a mode width of 1.6 at T = 210K. All calculations were performed with n = 13cm−3
background aerosol particles, the sensitvity of the model results on background aerosol
number density being small. The size distribution is modeled using 26 size bins for
the liquid aerosol. Homogeneous ice nuleation in STS droplets is calculated from the
nucleation rates by Koop et al. (2000), and the ice vapor pressure is calculated in
accordance with Marti and Mauersberger (1993). NAT nucleation on exposed ice
20
25
260

Page 9

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
surfaces is calculated in accordance with Luo et al. (2002), and the NAT vapor pressure
is taken from Hanson and Mauersberger (1988). For each timestep, the number of
nucleating ice and NAT particles is calculated for each size bin and transferred to new
bins of their own. Upon evaporation of ice and NAT particles, the resulting droplets
return into their original size bin.
The water mixing ratio in the Arctic stratosphere is taken from a linear interpolation of
measured water mixing ratios on 27 January 2000 with a mixing ratio χH2O= 4.2ppmv
at 400 K and χH2O= 6.3ppmv at 600 K (Schiller et al., 2002). In agreement with
Arctic mid-winter measurements (Kleinb¨ ohl et al., 2003), an HNO3profile with constant
volume mixing ratio of 7.5 ppmv is used (altitude-dependent deviations in the order of
30% may be expected in a real profile, but do not significantly alter model results).
5
10
3.5. Mesoscale modeling
Mesoscale numerical modeling studies (that include validation with observations) of
gravity waves over Scandinavia that propagate into the stratosphere and lead to wave-
induced PSCs were performed previously by D¨ ornbrack et al. (1999, 2001).
thermore, D¨ ornbrack et al. (2002) modeled 25-27 January 2000 episode with the
mesoscale MM5 model. Here we calculated a series of mesoscale simulations with the
High Resolution Model (HRM) (Majewski et al., 1991). A former version of this limited-
area model was used operationally by the German and Swiss Weather Services until
early 2001 and is widely used in the hindcast mode for regional climate simulations (e.g.
L¨ uthi et al. (1996)). The model integrates the set of primitive equations in the hydro-
static limit. The initial and boundary conditions are taken from ECMWF analyses, and a
radiative upper boundary condition prevents wave reflection. We use a dry physics ver-
sion of the model without radiation and a horizontal resolution of 0.125◦(corresponding
to ∼14km) and 60 vertical levels up to 4hPa (regularly spaced in logp). Further details
of the model setup can be found in Buss et al. (A gravity wave induced ice cloud over
Greenland: Model validation and investigation of dynamical mechanisms, manuscript
in preparation). Sensitivity studies showed that the simulation of the gravity wave and
261
Fur-
15
20
25

Page 10

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
the associated temperature fields is sensitive to horizontal and vertical resolution and
the initialization time, but insensitive to the height of the uppermost model level as well
as inclusion of radiatation and moisture. For the simulations presented here, the orog-
raphy was low-pass filtered with a conservative diffusion operator in order to eliminate
the grid-scale components of the gravity wave. The simulations were started every 24
hours for integration periods of 36 hours, and the simulation steps 12-36 hours were
used for the trajectory calculations.
Figure 2 shows a vertical cross-section of the HRM simulation at 15:00 UTC, 26 Jan-
uary. The propagating gravity wave becomes evident from the tilted bands of hori-
zontal divergence and convergence whose amplitude increases with height. The su-
perimposed temperature field reveals two regions with temperatures below 185 K at
21–27 km altitude, in agreement with the location of PSCs (data shown below).
5
10
4. Classification of lidar data
Figure 3 shows an overview of the lidar measurements on 25-27 January. The lidar
data was classified with the classification of Biele et al. (2001) and shows PSCs over
Scandinavia on all 3 successive days. Figure 4 shows the flight path of the selected
lidar data segments, and Fig. 5 shows the results of the classification of these data
segments according to Biele et al. (2001).
15
4.1.25 January 2000
The DC-8 crossed the Scandinavian mountain ridge three times, and on each crossing
PSCs were observed, consisting of ice, STS, and likely NAT (see Fig. 3). The LaRC
lidar measurement between 15:03 UTC and 15:38 UTC (Figs. 3 and 5a) are represen-
tative for all lidar observations over Scandinavia on this day. The classification shows
a large ice cloud over the Scandinavian mountain ridge, followed downstream by a
type ‘mix’ cloud and finally a type 1a-enh PSC. This sequence of PSC type has been
20
25
262

Page 11

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
observed before [eg. Carslaw et al. (1998a); Wirth et al. (1999)] and is interpreted as
follows. As the air approaches the mountain ridge, it is rapidly lifted and adiabatically
cooled. When temperatures reach the necessary supercooling of ∼ 3K, ice nuleation
begins, leading to ice PSCs with high particle number densities due to the high cool-
ing rates (∼ 30K/h). Very rapid cooling also prevents the background aerosol to take
up HNO3before ice nucleation sets in, and the liquid aerosol droplets are initially out
of thermodynamic equilibrium with the gas phase. Only during the evolutin of the ice
cloud, the liquid takes up HNO3and depletes the gas phase. Behind the ridge the
air sinks back to its original altitude, the temperature rises and the ice particles evap-
orate. Depending on temperature, also the liquid ternary solution droplets evaporate
HNO3and develop back to binary H2SO4/H2O droplets. Eventually, the NAT particles
evaporate when the temperature rises above the existence temperature TNAT, which is
about 8K above Tice(Hanson and Mauersberger , 1988). Zondlo et al. (2000) provide
a comprehensive overview over these processes.
Based on the LaRC lidar data, Hu et al. (2002) estimated ice particle number den-
sities in this PSC nice= 2 − 5cm−3with a mode radius r = 1.2 − 2µm. NAT par-
ticle number densities were estimated as nNAT= 0.1 − 0.5cm−3with a mode radius
r = 0.4 − 1µm, in general agreement with our analysis (not shown). Balloon-borne in-
situ measurements on this day at 20:30–22:30UTC (see Fig. 3) show the presence of
STS and NAT particles downwind of the Scandinavian mountain ridge at altitudes 22-
23km, with a NAT particle number density of n ≈ 0.1cm−3and a radius r = 0.5 −1µm
(Voigt et al., 2000).
5
10
15
20
4.2.26 January 2000
Lidar observations on board of the Falcon aircraft show the presence of large ice PSCs
over Scandinavia and patches of STS and type 1a-enh PSCs over the Atlantic ocean.
Downstream of the large ice cloud over Scandinavia (see Fig. 5b, at 200 - 600 km,
and Fig. 3) the Falcon observed a second ice PSC over Finland (see Fig. 5b, at 750-
850 km, and Fig. 3). This rather unique observation of 2 large successive PSCs was
263
25

Page 12

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
chosen as a test case for the combined mesoscale/microphysical modeling approach
(Sect. 5). The first ice PSC was also observed by ground-based FTIR, and based
on the FTIR data, Hoepfner et al. (2001) estimated the average ice particle number
densities nice=2-5cm−3, with a corresponding particle radius r = 2 − 1µm, in general
agreement with the analysis presented in Sect. 5. Directly downstream of the first ice
PSC, following a small region of type “mix”, only a tiny margin of the PSC is identified
as type 1a/1a-enh (see Fig. 5b, at 300-650 km and 22-26 km alt.). A few isolated NAT
“streaks” leave the lower part of the first ice cloud (see Fig. 5b at ∼ 700km). The PSC
upstream of the first ice cloud (see Fig. 5b, at 0-200km, altitude ≈ 22km) is classified
as 1a-enh/mix/STS and is part of several stratified PSC patches observed at ≈ 22km
altitude over the Atlantic (see Fig. 3).
The second ice cloud shows a clearer type 1a-enh signal downstream. Unfortu-
nately, the aircraft turned northward (and hence the subsequent flight leg is not quasi-
Lagrangian) just when the type 1a-enh signal appears (see right edge of Fig. 5b). The
classification of STS below the second ice cloud indicates that NAT particles (originat-
ing from the first ice cloud), if present at all, must be small (r < 0.5µm) and in very
low number densities nNAT≈ 0.01cm−3, such that the liquid droplets dominate the lidar
backscatter.
5
10
15
4.3.27 January 2000
On 27 January PSCs were sparse compared to the two preceding days. At 13:45 UTC
the Falcon observed a small (≈ 30km in wind direction) ice PSC near Kiruna, about
50 km downstream of Kebnekaise, the highest peak in northern Scandinavia (18◦33’E/
67◦53’N, elevation 2111m). Downstream of and just below the ice PSC there is STS
(see Figs. 3 and 5c). The small geographical dimensions of the cloud indicates that in
general temperatures were above the ice nucleation temperature, but single mountains
such as Kebnekaise may generate gravity waves of small horizontal dimensions and
thus localized cooling.
The DC-8 headedsouth fromKiruna
264
20
25
and observedaPSCnear

Page 13

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
Helsinki/St.Petersburg (13:00–13:15 UTC, Figs. 3 and 5d).
crossing of the mountain ridge, only STS was observed. The cloud near Helsinki
classifies as type 1a-enh, type 1a and a small section as type 1b (STS). A hypothesis
for the genesis of this NAT cloud will be presented in Sect. 6.
In addition to the lidar data, in-situ measurements on board the NASA ER-2 strato-
spheric research aircraft are available for this day. The ER-2 left Kiruna at ∼09:00 UTC
and headed south over Finland towards Russia, from where it returned to Kiruna. Dur-
ing the entire flight, the NOyinstrument found particulate matter at two positions only:
on the outbound flight at ∼ 10:00 UTC, 27.83◦E/61.44◦N, and on the incoming flight at
∼ 13:15 UTC, 29.45◦E/59.9◦N. The two locations are in agreement with the lidar ob-
servation of the type 1a-enh PSC discussed above. Based on the NOydata, Northway
et al. (2002) estimated a particle radius r ≈ 3µm (outgoing flight leg) and r ≈ 4µm
(incoming flight leg) in low number densities nNAT≈ 3 × 10−4cm−3. T-Matrix calcula-
tions for NAT particles of these sizes and number densities yield a backscatter ratio
BSR(1064)? 1.25, which is far from the observed values BSR(1064) ≈3-15. The ob-
served color ratio CR(532/1064) ≈1.1-3 of the lidar observation indicates that the cloud
mainly consists of smaller particles (from T-Matrix calculations we estimate r ≈ 0.8µm,
nNAT≈ 0.3cm−3, corresponding to ∼ 3ppmv HNO3in the solid phase). This discrep-
ancy between in-situ NOyand lidar data may be resolved by the fact that the altitude
of the ER-2 measurement (∼20km) is at the extreme bottom of the PSC observed
by the lidar (see Fig. 5d). It can be speculated whether these larger particles at the
cloud bottom result from sedimentation processes as proposed by Fueglistaler et al.
(2002a); Dhaniyala et al. (2002). In addition, the NAT number densities at the edges
of the cloud may be smaller due to less favorable nucleation conditions at cloud forma-
tion time. Both processes can lead to the observed low number density. In sum we
may conclude that this cloud consists of NAT particles with r ? 4µm, most likely with
r ≈ 0.8µm and nNAT≈ 0.3cm−3.
On the subsequent
5
10
15
20
25
265

Page 14

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
Back Close
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
5.Detailed PSC modeling, 26 January 2000
A comprehensive microphysical box model (Luo et al., 2002) was used with trajecto-
ries from NWP models. Forward and backward trajectories were calculated starting at
20.37◦E/63.75◦N, 15:00 UTC between 400 K and 650 K potential temperature in in-
crements of 2 K (corresponding to a vertical resolution ∼ 100m, yielding a total of 126
trajectories), from both ECMWF analysis data and the HRM simulation. Figure 6 shows
the flight path of the Falcon on 26 January (flight leg 3, black), ECMWF and HRM tra-
jectories at 500 K (red) and 600 K (blue) and the trajectory starting position (green).
The figure shows that the flight path is quasi-Lagrangian (i.e. parallel to the wind di-
rection) and consequently the microphysical calculations along these trajectories can
be directly compared to the observations. Minor deviations between observations and
calculated simulations should not come as a surprise, since flight path and trajecto-
ries are not perfectly aligned, and the mountain wave cannot be considered as strictly
stationary.
The simulated backscatter ratio BSR(1064) based on the results of the microphys-
ical calculations along the trajectories is shown in Fig. 7 together with the measured
backscatter ratio BSR(1064). This allows a direct comparison of simulations and mea-
surements, while the underlying microphyiscal results, such as particle types and num-
ber densities, are shown later (see Fig. 9). Model results along trajectories are plotted
in the geometry of the flight path (positions are resampled to equal distance from the
reference position where flight path and all trajectories intersect).
A comparison of the HRM-based PSC simulation (Fig. 7b) with the measured lidar
signal (Fig. 7a) shows that the simulation is in good agreement with the measure-
ment. In particular, the shape of first ice cloud over Scandinavia fits the observation
very well, and hence corroborates the HRM mesoscale simulation in combination with
the modeling of the ice nucleation. Also the second ice cloud is in general agree-
ment with measurements, although its shape shows a tilt westward with height which
is too strong compared to the measurement. Measured and simulated lidar backscat-
5
10
15
20
25
266

Page 15

ACPD
3, 253–299, 2003
Detailed modeling of
mountain wave PSCs
S. Fueglistaler et al.
Title Page
Abstract
Introduction
Conclusions
References
Tables
Figures
??
?
?
BackClose
Full Screen / Esc
Print Version
Interactive Discussion
c ? EGU 2003
ter ratios agree well and deviations are on average less than 25%. We conclude that
the mesoscale/microphysical model simulation correctly reproduces the cloud micro-
physics, in particular the particle types, number densities and sizes. It is emphasized
again at this point that the simulation uses the meteorological parameters temperature
and pressure from the HRM simulation without any modification, and that the micro-
physical box model calculates the nucleation of ice and NAT particles from nucleation
rates rather than from prescribed values as in previous studies.
Upon closer inspection of measurement and simulation we note again the wrong tilt
of the second ice cloud. A part of this tilt is an effect of the deviation between flight
path and trajectories, however it is also apparent in the temperature field of the HRM
simulation. This tilt is also observed in the main ice cloud, where it causes trajectories
to descend too soon compared to the measurement (i.e. ice particles evaporate too
soon in the simulation, compare the ice regions with BSR(1064) > 100 at 24-25 km
altitude in the measurement and simulation, Figs. 7a,b). Further we note that the HRM-
based simulation cannot reproduce the small scale waves (with wavelength ? 20km
and amplitude ? 5K) superimposed on the dominant wave number observed in the
lidar data. It is clear that the mesoscale model with a spatial resolution of ∼ 15km
cannot resolve these waves. Calculations with a manually modified trajectory with
these small scale waves superimposed on the HRM trajectory show that these waves
can affect the microphysical properties quantitatively, but do not change the qualitative
properties of the cloud (calculations shown in the Appendix).
The simulation cannot reproduce the small PSC at 22 km altitude upstream of the
first ice cloud, in Sect. 4.2 identified as STS and NAT, due to two reasons. Firstly, the
microphysical model requires ice particles to initiate NAT nucleation, but the HRM sim-
ulation temperatures do not reach the ice nucleation temperature upstream of Scandi-
navia. Secondly, the HRM backtrajectories are slightly further south than the flight path
for this section over the Atlantic (see Fig. 6). As will be shown later, a trough of air cold
enough to form STS is present in the simulation at the location where the upstream
PSC was observed, but its southern edge is just missed by the HRM trajectories.
5
10
15
20
25
267