ACPAtmospheric Chemistry and PhysicsACPAtmos. Chem. Phys.1680-7324Copernicus PublicationsGöttingen, Germany10.5194/acp-18-11041-2018A model intercomparison of CCN-limited tenuous clouds in the high ArcticA model intercomparison of CCN-limited tenuous clouds in the high ArcticStevensRobin G.robin.stevens@canada.cahttps://orcid.org/0000-0002-8737-6988LoeweKatharinahttps://orcid.org/0000-0003-3574-7203DeardenChristopherDimitrelosAntoniosPossnerAnnahttps://orcid.org/0000-0001-6996-8624EirundGesa K.https://orcid.org/0000-0001-6346-2534RaatikainenTomihttps://orcid.org/0000-0002-2603-516XHillAdrian A.ShipwayBenjamin J.WilkinsonJonathanhttps://orcid.org/0000-0002-6906-4999RomakkaniemiSamihttps://orcid.org/0000-0001-9414-3093TonttilaJuhaLaaksonenArihttps://orcid.org/0000-0002-1657-2383KorhonenHannelehttps://orcid.org/0000-0001-6264-0706ConnollyPaulLohmannUlrikehttps://orcid.org/0000-0001-8885-3785HooseCorinnahttps://orcid.org/0000-0003-2827-5789EkmanAnnica M. L.https://orcid.org/0000-0002-5940-2114CarslawKen S.https://orcid.org/0000-0002-6800-154XFieldPaul R.Institute of Climate and Atmospheric Science, School of Earth and Environment, University of Leeds, Leeds, UKInstitute of Meteorology and Climate Research, Karlsruhe Institute of Technology, Karlsruhe, GermanyCentre for Atmospheric Science, School of Earth and Environmental Sciences, University of Manchester, Manchester, UKDepartment of Meteorology, Stockholm University, Stockholm, SwedenInstitute for Atmospheric and Climate Science, Eidgenössische Technische Hochschule, Zürich, SwitzerlandDepartment of Global Ecology, Carnegie Institution for Science, Stanford, CA, USAFinnish Meteorological Institute, Helsinki, FinlandMet Office, Exeter, UKFinnish Meteorological Institute, Kuopio, Finlandnow at: Air Quality Research Division, Environment and Climate Change Canada, Dorval, Canadanow at: the Centre of Excellence for Modelling the Atmosphere and Climate, School of Earth and Environment, University of Leeds, Leeds, UKRobin G. Stevens (robin.stevens@canada.ca)8August2018181511041110715December201711December201723June201826June2018This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/4.0/This article is available from https://acp.copernicus.org/articles/18/11041/2018/acp-18-11041-2018.htmlThe full text article is available as a PDF file from https://acp.copernicus.org/articles/18/11041/2018/acp-18-11041-2018.pdf
We perform a model intercomparison of summertime high Arctic (> 80∘ N) clouds
observed during the 2008 Arctic Summer Cloud Ocean Study (ASCOS) campaign,
when observed cloud condensation nuclei (CCN) concentrations fell below
1 cm-3. Previous analyses have suggested that at these low CCN
concentrations the liquid water content (LWC) and radiative properties of the
clouds are determined primarily by the CCN concentrations, conditions that
have previously been referred to as the tenuous cloud regime. The
intercomparison includes results from three large eddy simulation models
(UCLALES-SALSA, COSMO-LES, and MIMICA) and three numerical weather prediction
models (COSMO-NWP, WRF, and UM-CASIM). We test the sensitivities of the model
results to different treatments of cloud droplet activation, including
prescribed cloud droplet number concentrations (CDNCs) and diagnostic CCN
activation based on either fixed aerosol concentrations or prognostic aerosol
with in-cloud processing.
There remains considerable diversity even in experiments with prescribed
CDNCs and prescribed ice crystal number concentrations (ICNC). The
sensitivity of mixed-phase Arctic cloud properties to changes in CDNC depends
on the representation of the cloud droplet size distribution within each
model, which impacts autoconversion rates. Our results therefore suggest
that properly estimating aerosol–cloud interactions requires an appropriate
treatment of the cloud droplet size distribution within models, as well as
in situ observations of hydrometeor size distributions to constrain them.
The results strongly support the hypothesis that the liquid water content of
these clouds is CCN limited. For the observed meteorological conditions, the
cloud generally did not collapse when the CCN concentration was held constant
at the relatively high CCN concentrations measured during the cloudy period,
but the cloud thins or collapses as the CCN concentration is reduced. The CCN
concentration at which collapse occurs varies substantially between models.
Only one model predicts complete dissipation of the cloud due to glaciation,
and this occurs only for the largest prescribed ICNC tested in this study.
Global and regional models with either prescribed CDNCs or prescribed aerosol
concentrations would not reproduce these dissipation events. Additionally,
future increases in Arctic aerosol concentrations would be expected to
decrease the frequency of occurrence of such cloud dissipation events, with
implications for the radiative balance at the surface. Our results also show
that cooling of the sea-ice surface following cloud dissipation increases
atmospheric stability near the surface, further suppressing cloud formation.
Therefore, this suggests that linkages between aerosol and clouds, as well as
linkages between clouds, surface temperatures, and atmospheric stability need
to be considered for weather and climate predictions in this region.
Introduction
A decrease in Arctic sea-ice extent and thickness has been observed within
recent decades . Further decreases in Arctic sea-ice
extent are expected to increase the fluxes of aerosol and aerosol precursor
gases as well as latent heat and sensible
heat from the open ocean surface within the Arctic .
The long-range transport of anthropogenic aerosol is currently a significant
source to the Arctic region . Therefore, future
changes in non-local sources of aerosol and long-range transport could have
significant impacts on aerosol concentrations in the Arctic. Furthermore, an
increase in shipping traffic is expected once the Arctic becomes seasonally
ice free, further increasing aerosol concentrations . This
increase in shipping traffic would also be expected to yield an increased
demand for accurate weather forecasts over the Arctic region. However, it
remains unclear whether the net effect of these changes in aerosol
concentrations and surface fluxes would result in an increase or a decrease
in cloud cover or drizzle precipitation. Changes in cloud properties
could strongly influence the radiation budget in the Arctic, resulting in
feedbacks on the rate of sea-ice loss. Arctic clouds remain poorly
understood, and the current representation of these processes in global
climate models is most likely insufficient to realistically simulate
long-term changes.
Few observations have been made of Arctic clouds relative to clouds at lower
latitudes. Field campaigns that have investigated Arctic clouds include the
International Arctic Ocean Expeditions in 1991 AOE-91;
and 1996 AOE-96;, the Arctic Ocean Experiment in 2001
AOE-01;, the First ISCCP (International
Satellite Cloud Climatology Project) Regional Experiment–Arctic Clouds
Experiment in 1998 FIRE-ACE;, the Surface Heat Budget of
the Arctic Ocean project in 1997–1998 (SHEBA; ), the
Mixed-Phase Arctic Cloud Experiment in 2004 (M-PACE; ),
the Indirect and Semi-Direct Aerosol Campaign in 2008
ISDAC;, the Arctic Summer Cloud Ocean Study in 2008
ASCOS;, the VERtical Distribution of Ice in Arctic
cloud campaign in 2012 VERDI;, the Aerosol–Cloud
Coupling and Climate Interactions in the Arctic campaign in 2013
ACCACIA;, the Arctic Clouds in Summer
Experiment in 2014 ACSE, and the Canadian Network
on Climate and Aerosols: Addressing Key Uncertainties in Remote Canadian
Environment campaign in 2014 NETCARE;. Of these
campaigns, only a few (AOE-91, AOE-96, AOE-01, ASCOS, and ACSE) have sampled
the high Arctic north of 80∘ N. These campaigns and subsequent analyses have
provided insights into the structures and radiative impacts of Arctic clouds,
including the following.
At supersaturations as high as 0.8 %,
observed cloud condensation nuclei (CCN) concentrations are usually less than
100 cm-3 in the high Arctic summer and have been observed to be as
low as 1 cm-3.
During the AOE-91, AOE-96, AOE-01, and ASCOS campaigns more than 25 %
of observed CCN concentrations were < 10 cm-3 at supersaturations
≤ 0.3 %. Additionally, more than 60 % of the low-altitude
clouds observed via aircraft during the NETCARE campaign were found to have
CCN concentrations less than 16 cm-3 at a supersaturation of
0.6 %.
Arctic clouds often have a net warming
effect on the surface, even in summer . The shortwave
(SW) radiative effect of Arctic clouds is small relative to the longwave (LW)
radiative effect due to the high albedo of sea ice and the low angle of
incoming solar radiation.
The LW surface warming effect of Arctic clouds
strongly affects the surface temperature and therefore would be expected to
impact the thickness and extent of Arctic sea ice
.
In order to better understand the processes controlling Arctic clouds and
their uncertainties in current models, we perform a model intercomparison of
summertime high Arctic (> 80∘ N) clouds. We have chosen as our case
study the final 2 days of the ice drift period of the 2008 ASCOS campaign
. During this period, a decrease in cloud
water content was observed coincident with a decrease in observed CCN
concentrations to less than 1 cm-3. The concentrations of CCN were
measured continuously using a CCN counter operating at a fixed
supersaturation of ∼ 0.2 %. Details on the quality and data
processing of ship-based CCN measurements are available in
and in .
Previous analysis has identified these clouds
as existing within the tenuous cloud regime: cloud liquid water content (LWC)
and surface radiative effects are limited by the availability of aerosol to
act as CCN. This cloud regime has been observed during the ASCOS campaign
and the NETCARE campaign . Due to
the low CCN concentrations observed in the high Arctic, this cloud regime is
expected to be a frequent occurrence in the Arctic summer.
have linked the dissipation of these clouds and the associated increase in
surface LW cooling to the onset of the autumn sea-ice freeze-up in 2008. The
tenuous cloud regime would be very sensitive to changes in aerosol
concentrations due to increased emissions from either increased human
activity in the Arctic or increased emissions due to decreasing sea ice.
Changes in these clouds would be expected to affect the surface radiative
energy balance and thereby potentially affect Arctic sea-ice extent and
thickness. The tenuous cloud regime therefore presents an important but
challenging case to represent within models.
The ASCOS ice drift period, in whole or in part, has been previously examined
using models by , ,
, , , and
. The models used by these studies were a single-column model
configuration of the Met Office Unified Model (UM), two versions of the
Arctic System Reanalysis (ASR) and the ERA-Interim reanalysis, the Integrated
Forecast System (IFS) model of the European Centre for Medium-Range Weather
Forecasts (ECMWF), the polar-optimized version of the Weather Research and
Forecasting (WRF) regional numerical weather prediction (NWP) model, the
Consortium for Small-scale Modeling (COSMO) model configured as a large eddy
simulation (LES) model, and the MISU MIT Cloud and Aerosol (MIMICA) LES
model, respectively. found that observations of surface
radiative fluxes and surface temperatures were better reproduced by the
single-column UM during the tenuous cloud regime period on
1 September 2008 if prescribed CCN concentrations were
reduced to 1 cm-3. For higher CCN concentrations, the model
produced cloud with much larger LWCs than observed.
highlighted the fact that the two configurations of ASR failed to reproduce the
observed clouds from 27 August to
1 September. They noted that this period was better
represented by ERA-Interim, and they hypothesized that this was due to
differences in the treatment of cloud microphysics.
found that, while using a constant assumed CCN concentration, increased model
vertical resolution and a newer cloud microphysics scheme including
prognostic cloud ice, rain and snow were insufficient to reproduce cloud
dissipation during the tenuous cloud periods. Similarly to ,
found that biases of the Polar WRF regional NWP model
against surface radiative flux observations for the entire ASCOS drift period
were reduced as the prescribed cloud droplet number concentration (CDNC) was
reduced from values representative of low latitudes (250 cm-3) to
values representative of pristine Arctic conditions (10 cm-3).
Biases during the periods labelled as in the tenuous cloud regime were
further reduced if the prescribed CDNC was reduced to 1 cm-3.
found that in the LES configuration of COSMO, a prescribed
CDNC of 2 cm-3 was insufficient to prevent cloud dissipation but
that a cloud could be maintained with a prescribed CDNC of 10 cm-3.
They additionally performed sensitivity studies to moisture availability and
to ice crystal number concentrations (ICNCs). The cloud LWC was found to be
sensitive to both moisture availability and ICNC, but none of the tested
water vapour profiles resulted in cloud dissipation, and an unrealistically
high ICNC was required for cloud glaciation. Using the MIMICA LES model,
found that enhanced levels of accumulation-mode particles,
if located at the cloud top, may under certain conditions be an important
source of accumulation-mode particles in the Arctic boundary layer.
Previous model studies of other Arctic mixed-phase clouds have established
the sensitivity of cloud LWC, ice water contents (IWC), and other cloud
properties in models to the interaction of ice and liquid ,
the representation of ice enhancement mechanisms , prescribed
cloud ICNC , ice-nucleating particle (INP)
concentrations , INP depletion and supply
, the size distribution of cloud ice , the
habit of cloud ice , and enhancement of CCN
concentrations by ship emissions . Additionally,
have investigated the importance of riming in mixed-phase
clouds. However, the clouds investigated in these studies had greater CDNCs
and would not be expected to show the same sensitivity to changes in CCN
concentrations as the tenuous cloud regime observed during ASCOS.
In this paper, we extend these previous studies by comparing the results of
both LES and cloud-resolving NWP models of the tenuous cloud regime observed
during ASCOS using increasingly complex representations of aerosol–cloud
interactions. We begin with simulations of liquid-phase cloud only, and we
later show results in which ice nucleation is included through prescribed ICNCs.
We show first the results of simulations in which cloud droplet activation is
represented using prescribed CDNCs, similar to the studies of
, , and . We then show the
results of simulations with cloud droplet activation calculated based on a
temporally and spatially constant aerosol size distribution. Finally, we
include in our simulations prognostic aerosol concentrations, including
aerosol uptake and removal by activation into cloud droplets, which reduces
the available CCN for activation in subsequent model time steps. In this way,
we attempt to determine the key processes contributing to the dissipation of
these clouds, and we isolate and attempt to attribute differences in model
results to differences in model processes. We then discuss the implications
for realistic representation of Arctic aerosol–cloud interactions.
Observed cloud properties, surface radiation,
and aerosol concentrations. (a) Liquid water content, (b) liquid
water path, (c) ice water content, (d) ice water path, (e) surface net longwave flux, and (f) concentrations of N50. Shaded
rectangles indicate the interquartile ranges of LWP, IWP, surface net LW
radiation, and N50 during the cloudy and nearly cloud-free periods
defined in Sect. . Dashed vertical lines indicate the
beginnings and endings of these periods.
Section shows an overview of observed meteorological
conditions during the case study period. Section
describes the models participating in this study.
Section describes the simulations performed for this
study. Section presents and discusses the results of our
liquid-phase-only simulations, and Sect. presents and
discusses the results of the simulations including cloud ice. Finally,
Sect. offers a summary and our conclusions.
Overview of the ASCOS campaign
A full description of the conditions during the ASCOS campaign is available
in . Observations during the ASCOS campaign were
obtained on-board the icebreaker Oden from two measurement sites set up on
the ice floe and by helicopter. However, helicopter observations were
restricted to outside of clouds due to safety concerns regarding icing of the
aircraft. In order to examine the tenuous cloud regime, we focus our study on
the period from 30 August to 1 September
2008. These were the last 2 days of the ice drift period, which ended at
about 87∘09 N, 11∘01 W. Observed winds were westerly at the
site, with observed wind speeds varying between 2 and 6 ms-1
during the 2-day period. Conditions were dominated by a high-pressure
system over the North Pole, yielding anticyclonic winds on the synoptic
scale. Observed surface pressures rose from ∼ 1025 to
∼ 1030 hPa during the 2-day period. Mixed-phase
stratocumulus clouds were observed during this period until approximately
20:00 UTC on 31 August, when a break in low-level cloud
cover was observed, despite observed water vapour mixing ratios at or above
saturation coincident with a decrease in observed CCN concentrations from
about 70 to < 1 cm-3. The CCN
concentrations were measured continuously using a CCN counter operating at a
fixed supersaturation of ∼ 0.2 %. A second identical CCN
counter was cycled between supersaturations of 0.11 and 0.73 %.
give further details on the quality of the data.
Near-surface air temperatures were observed to be near -4 ∘C,
falling to -13 ∘C after the break in cloud.
Figure shows cloud properties, surface radiation, and
aerosol concentrations derived from observations. Net surface LW radiation is
defined to be positive downwards (absorption by the surface) throughout this
paper. The LWC and IWC were derived from measurements using a microwave
radiometer, 35 GHz millimetre cloud radar, vertical temperature profiles from
radiosondes, and ceilometers, as detailed in . The
methodology is described further in . The observed liquid
water path (LWP) has a reported root mean square error of 25 gm-2 and the uncertainty in the observed ice water path
(IWP) could be up to a factor of 2 .
For ease of comparison with the model results, we designate the period from
21:00 UTC on 30 August to 12:00 UTC on
31 August as the “cloudy” period and the period from
00:00 to 06:00 UTC on 1 September as the “nearly cloud-free”
period. There is a clear transition in every variable shown in
Fig. between these two periods: the liquid and frozen
parts of the cloud both descend towards the surface, and the liquid and ice
water contents both decrease, causing an increase in the LW emission from the
surface. These changes are coincident with a decrease in the observed surface
concentrations of aerosol particles larger than 50 nm (N50) from
>10 to < 1 cm-3. Total aerosol concentrations as
measured by a twin differential mobility particle sizer with a lower
detection limit of 3 nm fell generally below 10 cm-3, with a
median value of 2 cm-3 during the nearly cloud-free period.
Further details on the quality and data processing of ship-based aerosol
measurements are available in . CCN concentrations
measured at supersaturations as high as 0.73 % during this period were also
below 1 cm-3. Additionally, helicopter profiles of aerosol number
concentrations were performed from 19:53 to 20:13 UTC on
31 August and from 07:32 to 07:55 UTC on
1 September using a condensation particle counter
. These indicate that the number concentrations of
aerosol larger than 14 nm were generally below 10 cm-3 up to
850 m of altitude during the 31 August profile and up
to 500 m of altitude during the 1 September profile. With
reference to Fig. , we note that these heights are similar
to the locations of the observed cloud-top heights at these time periods, and
these altitudes were also similar to temperature inversion base heights
observed via a scanning microwave radiometer .
In-cloud measurements were not performed due to aircraft icing concerns
. Additionally, CloudSat+Cloud–Aerosol Lidar with
Orthogonal Polarization (CALIOP) cloud retrievals are not available north of
82∘ N and are therefore unavailable for this case . Moderate
Resolution Imaging Spectroradiometer (MODIS) retrievals have been shown to
underestimate cloud cover in the Arctic, particularly over sea ice and for
cloud-top heights less than 2 km. We therefore
consider MODIS-derived cloud information unreliable for this case. Therefore,
no reliable observations of cloud droplet number concentrations are available
for this case.
As mentioned above, previous analysis has
identified these clouds as existing within the tenuous cloud regime: cloud
LWC is limited by the availability of aerosol to act as CCN. The hypothesis
is that at extremely low CCN concentrations, each available CCN is activated,
grows through condensation to drizzle droplet sizes, and is removed by
sedimentation. It is implicit in this hypothesis that in-cloud precipitation
occurs predominantly through liquid-phase processes, although frozen-phase
processes could contribute to precipitation formation, and glaciation would
be an alternate cause of cloud dissipation. In the following sections the
aerosol and meteorological environment will be decoupled via sensitivity
tests to assess the validity of this hypothesis.
Description of participating models
Simulations were performed using three large eddy simulation (LES) models and
three numerical weather prediction (NWP) models. LES models are fine-resolution models (horizontally several metres to hundreds of metres) with
domains typically from hundreds of metres to hundreds of kilometres capable
of resolving turbulent eddies and useful for detailed studies of clouds. NWP
models are generally coarser-resolution (horizontally hundreds of metres to
tens of kilometres) models with larger domains (tens of kilometres to global)
capable of simulating mesoscale weather systems and performing operational
forecasting. The NWP models used in this study all prognose surface
temperatures and surface sensible and latent heat fluxes, but these values
are prescribed for the LES models in this study. The NWP models can describe
the full meteorological variability and can therefore help to separate
meteorological versus aerosol effects.
The LES models participating in this study are the University of California,
Los Angeles LES with Sectional Aerosol module for Large Scale Applications
UCLALES-SALSA;, the MISU MIT Cloud and Aerosol LES
model MIMICA;, and the Consortium for Small-scale
Modeling (COSMO) model configured as an LES model
(hereafter referred to as COSMO-LES). The NWP models are v3.6.1 of the Polar
Weather Research and Forecasting model Polar WRF;, the
Met Office Unified Model with Cloud AeroSol Interacting Microphysics
UM-CASIM;, and COSMO configured as an NWP model
(hereafter referred to as COSMO-NWP). Each of the
models is described in detail in previous publications, so we will
restrict ourselves to a brief overview here. The participating models are
described and compared in Table .
UCLALES-SALSA is a combination of an LES model
UCLALES; and a sectional aerosol and cloud
microphysics module SALSA;. A detailed description of
UCLALES-SALSA can be found in . A comparison of
UCLALES-SALSA results against those of a previous model intercomparison based
on the second Dynamics and Chemistry of Marine Stratocumulus Field Study
(DYCOMSII) can also be found in . The properties and
microphysical processes of aerosol, cloud droplets, and rain are defined for
certain size sections (bins). In the current set-up, aerosol has 10 size bins
based on dry particle size and cloud droplets have 7 bins that are parallel
with the 7 largest aerosol bins. Raindrops have seven size bins which are based
on droplet size. Microphysics includes water vapour condensation and
evaporation, cloud activation, rain formation, coagulation, and deposition.
With the exception of rain formation, these processes are modelled based on
physical equations. Rain formation is based on an autoconversion scheme in
which
a log-normal size distribution (σ=1.1) is expected for each cloud bin
and droplets larger than 50 µm are moved to the first precipitation
bin. Subgrid-scale turbulence is based on the Smagorinsky–Lilly model as
described in . Radiation transfer is calculated following
the four-stream radiative transfer solver of .
MIMICA is an LES model which uses a two-moment bulk microphysics scheme with
five hydrometeor categories (cloud droplets, raindrops, ice crystals,
graupel, and snow). MIMICA also includes a two-moment aerosol module providing
the possibility to represent different aerosol populations covering a range
of size intervals and compositions . The autoconversion
parameterization and the interactions between liquid particles follow the
scheme of . Liquid–ice interactions are parameterized
according to the microphysical scheme of . The subgrid-scale
model is based on a Smagorinsky–Lilly eddy diffusivity closure
. At the surface, the model uses Monin–Obukhov similarity
theory and the momentum fluxes are computed as described in
. The CCN activation is described by the kappa-Köhler
theory . A four-stream radiative transfer solver
is used in the model. A thorough description of MIMICA is
given in . The MIMICA model has participated in the ISDAC
model intercomparison study and has also been used
to simulate the DYCOMSII case ; in both cases it
compared well with other models.
Both COSMO-LES and COSMO-NWP use the two-moment cloud microphysics scheme
described in . A fixed log-normal aerosol mode was
implemented into COSMO-LES and prognostic aerosol transport, activation, and
resuspension following hydrometeor evaporation were implemented in COSMO-NWP
following . Aerosol activation to cloud droplets is
performed following the scheme described in and
. The two-stream radiation scheme after
calculates the radiation transfer in COSMO. The boundary
layer turbulence is parameterized using a 3-D scheme in COSMO-LES
and a 1-D vertical turbulent
diffusion scheme based on in COSMO-NWP. The minimum
threshold for the eddy diffusivity in COSMO-NWP was adjusted to
0.01 m2s-1. The COSMO model participated in
the ISDAC LES model intercomparison study , and the
predicted IWP and LWP were within the range of the other models.
Description of models participating in this
study.
UCLALES-SALSAMIMICACOSMO-LESCOSMO-NWPWRFUM-CASIMDescribed in () (), (), , (), ()Condition for icenucleationNo ice nucleationSi>0.05 and qc>0.001gkg-1Si>0.05 and qc>0.001gkg-1Si>0.05 and qc>0.001gkg-1Sl>-0.001and T<-8∘CSl>-0.001and T<-8∘CNumber ofvertical levels below2 km112128124172524Finest vertical resolution (m)15.07.57.524.230.210.8Coarsest vertical resolution below 2 km (m)47.247.7228.3237.1141.9156.7Coarsest vertical resolution below 1.5 km (m)23.835.635.6202.3108.8136.7Horizontal resolution50 m62.5 m100 m1 km1 km1 kmHorizontal domain size3.15 km6 km6.4 km600 km600 km600 kmPrognostic aerosol*Sectional aerosol (10 size bins; drydiameter from3 nm to 1 µm)Two-moment bulk NoneNoneNoneTwo-moment bulk
* Only used in CCN30prog and CCN80prog simulations; described in
Sect. .
The physics options used in the Polar WRF simulations are based on the
recommendations described in . Cloud microphysical processes
are parameterized according to the double-moment scheme of
. Autoconversion of cloud droplets to rain is treated
according to the scheme of . For droplet activation in the
CCN30fixed and CCN80fixed cases (see Sect. ), the scheme
of is used assuming a fixed background concentration
of CCN. There is no prognostic treatment of aerosols in the WRF simulations.
The atmospheric boundary layer is represented by the
Mellor–Yamada–Nakanishi–Niino (MYNN) scheme , and the
rapid radiative transfer model RRTMG; is used for both
longwave and shortwave radiation.
The UM-CASIM model has been described previously in and
. However, the subgrid cloud scheme described in
was not used for this study. Boundary layer processes,
including surface fluxes of moisture and heat, are parameterized with the
blended boundary layer scheme and subgrid-scale
turbulent processes are represented with a 3-D Smagorinsky-type turbulence
scheme . A two-stream radiation scheme is
used, as described in . It is possible to run the UM-CASIM
model as a fully coupled atmosphere–ocean model, but for this study a fixed
sea-ice fraction of 100 % and a fixed sea-ice thickness of 2 m
were used. Activation of cloud droplets in simulations without prescribed
CDNCs is performed following the scheme described in
and .
Except UCLALES-SALSA and WRF, all models in this study contained five
hydrometeor classes: cloud droplets, rain, cloud ice crystals, snow, and
graupel. These hydrometeor classes are represented as gamma distributions
with prescribed shape parameters and prognosed bulk mass and number
concentrations. WRF contains the hydrometeor classes described above except
graupel. UCLALES-SALSA represents cloud droplets and raindrops using seven
sectional size bins for each species, tracking number and mass independently.
Frozen water species are not currently simulated by UCLALES-SALSA.
Sedimentation of cloud droplets is simulated only by UCLALES-SALSA, WRF, and
UM-CASIM.
Nucleation of cloud ice was conditionally permitted in each model within a
defined range of temperatures (T), cloud droplet mass mixing ratios
(qc), liquid supersaturations (Sl), and ice
supersaturations (Si). In MIMICA and the two COSMO models, ice
forms in the presence of supercooled liquid water (Si>0.05 and
qc>0.002 or 0.001 gkg-1, respectively) and for WRF and
UM-CASIM ice forms at T<-8∘C in the presence of
supercooled liquid water. These differences will have minimal impact on the
simulation, as cloud-top temperatures are generally below -8 ∘C.
For all models and all simulations, the rate of ice nucleation was
parameterized following and . The
change in ICNC due to nucleation of cloud ice in each time step was therefore
ΔICNC=max(0,ICNCfixed-ICNC),
where ICNC is the cloud ice crystal number concentration, ΔICNC is the
change in ICNC due to ice nucleation during a single model time step, and
ICNCfixed is a chosen fixed value dependent on the experiment:
1, 0.2, or 0.02 L-1 for
experiments labelled ICNC1p00, ICNC0p20, or ICNC0p02, respectively (see
Sect. ). Thus, whenever the conditions for ice formation
are met, any loss in Nice due to sedimentation, autoconversion to snow,
or scavenging will be exactly compensated for by further activation to maintain
the ICNC as ICNCfixed. For simulations labelled NOICE, the models were
run without any formation of frozen cloud water permitted.
Description of simulations
For the UM-CASIM simulations, a global simulation initialized using the
European Centre for Medium-Range Weather Forecasts (ECMWF) global analysis
was performed to produce a set of time-varying boundary conditions. The WRF
and COSMO-NWP models used boundary conditions directly from the ECMWF global
analysis. The three NWP models were then run with a
0.009∘× 0.009∘ horizontal resolution rotated grid (approximately
1 × 1 km throughout the domain) spanning a 600 × 600 km
domain centred at 87.3∘ N, 6.0∘ W. The period of interest for
this study is the transition period of the observed cloud from the cloudy
state to the nearly cloud-free state, starting approximately at 12:00 UTC on
31 August (see Sect. above). The NWP
models were therefore started at 12:00 UTC on 30 August 2008
to allow for 24 h to reach a representative state, and the total
simulation duration was 48 h.
Initial profiles of potential temperature, humidity, and wind speed for the
LES models were taken from the 31 August 05:35 UTC
radiosonde observations from the ASCOS campaign (Fig. ).
No flux of heat and moisture from or to the surface was permitted due to the
sea-ice cover. Sensible and latent heat fluxes at the surface were
<1Wm-2 in the UM-CASIM modelling results, and observed surface
fluxes were generally <5Wm-2 during the ASCOS campaign
. Surface temperatures were prescribed to be
-1.8 ∘C. Furthermore, the set-up of all LES models follows the
large-scale subsidence description of , with
divergence assumed to be constant below a height of 2 km. The value of
the divergence was chosen to be 1.5×10-6s-1.
Preliminary simulations with UCLALES-SALSA showed that a divergence of
1.5×10-6s-1 was too low in this model to balance
radiative cooling and the associated mixing, and the cloud layer would
continuously rise at a rate similar to the clouds in the COSMO-LES CDNC30
simulations (e.g. Fig. ). The increased
length of the UCLALES-SALSA simulations compared to the COSMO-LES
simulations (discussed next paragraph) allows the cloud layer to rise to
unrealistic altitudes. A larger value of 5×10-6s-1 was
therefore used instead for the subsidence in the UCLALES-SALSA simulations.
While we do not investigate sensitivities to prescribed subsidence in this
study, other studies have shown that differences in prescribed subsidence
affect Arctic mixed-phase cloud LWP and IWP . Within
UCLALES-SALSA, subsidence only affects the tendencies of temperature and
water vapour and does not directly alter advection of air parcels, aerosols,
cloud droplets or rain.
Due to numerical instabilities, the COSMO-LES simulations are restricted to a
duration of 16 h, including 2 h of spin-up during which ice formation
is not permitted. These instabilities are visible in the full model results
as waves in the upper atmosphere. These waves do not reach the boundary layer
during the simulations, and thus they do not influence the cloud in the
boundary layer. In order to focus on the transition period starting
approximately at 12:00 UTC on 31 August, the COSMO-LES
simulations were therefore started at 06:00 UTC, 31 August.
UCLALES-SALSA simulations were run from 00:00 UTC on 31 August
for 36 h, including 3 h of spin-up, during which coagulation,
sedimentation, and autoconversion are disabled. MIMICA simulations were run
from 12:00 UTC, 30 August for 72 h, including 2 h
of spin-up, but we only show results from the first 48 h in this study.
As we have not prescribed any time-varying surface fluxes or large-scale
forcings for the LES models and the diurnal cycles in this case are weak,
the LES model results are largely independent of the start time for this
case.
31 August 05:35 UTC radiosonde
observations of (a) absolute temperature, (b) potential
temperature,
and (c) relative humidity from the ASCOS campaign.
Several sensitivity experiments with different treatments and concentrations
of CCN and ICNC were carried out (Table ). The values chosen
for the sensitivity studies were based on observations of aerosol
concentrations during the ASCOS campaign. First, to make the models as
similar as possible, we performed simulations with prescribed CDNCs. We first
prescribed a CDNC of 30 cm-3 (CDNC30), as mean CCN concentrations
at a supersaturation of 0.2 % were observed to be 26.55 cm-3 over
the ice drift period . Then, in order to test the
sensitivity to reduced aerosol concentrations, we perform simulations with
the CDNC reduced to 3 cm-3 (CDNC03).
Description of simulations performed. The last six
columns indicate which models performed simulations of each case.
UCL: UCLALES-SALSA, MIM: MIMICA, COL: COSMO-LES,
CON: COSMO-NWP,
UMC: UM-CASIM.
We then performed simulations in which cloud droplet activation was calculated
based on an aerosol size distribution. We represented the aerosol size
distribution using the log-normal fit of . A single log-normal
mode was fit to observations of accumulation-mode particles made on-board the
icebreaker Oden using a twin differential mobility particle sizer with an
inlet height around 20–25 m above the surface .
Further details on the quality and data processing of ship-based aerosol
measurements are available in . This yielded a median
diameter of 94 nm and a geometric standard deviation of 1.5. For
simplicity, we assume that the aerosol particles are composed entirely of
ammonium sulfate, but in reality 43 % of the non-refractory aerosol mass was
observed to be organic with low hygroscopicity
. We initially chose an aerosol number concentration of
30 cm-3 (CCN30) to represent the cloudy period based on the
observed CCN concentrations. However, preliminary simulations with
UCLALES-SALSA indicated that an initial CCN concentration of
30 cm-3 would result in the dissipation of the cloud (as will be shown
in Sect. ), so a larger value of 80 cm-3 (CCN80) was
chosen as a sensitivity study. Additionally, we chose a value of
3 cm-3 to test the sensitivity of our results to further reductions
in the CCN concentration. In order to assess the sensitivity to the removal
of aerosol by cloud processes within the models, we perform simulations with
either constant aerosol or with prognostic aerosol processing. In the
CCN30fixed and CCN80fixed cases, the aerosol concentration remains constant
in space and time and is not affected by cloud processes. Cloud droplet
activation occurs only if the number of newly activated cloud droplets
exceeds the current number of cloud droplets in a given grid cell, in which
case the CDNC is updated to the number calculated by the activation
parameterization. In the prognostic aerosol simulations (CCN03prog, CCN30prog
and CCN80prog), aerosol is removed through activation into cloud droplets,
resuspended upon evaporation, and transported by advection.
In addition to the sensitivity to CCN, we also investigated the sensitivity
of the clouds within the models to ICNC. Observations of ice-nucleating
particles (INPs) are not available for this period, as the concentrations at
the surface were below the detection limit of the instrument
. Following , we chose a prescribed ICNC of
0.2 L-1 as our control simulation (ICNC0p20) based on previous
observations of INP in the Arctic from AOE-91 and AOE-96
. Additionally, we performed a liquid-phase-only
sensitivity study with no ice nucleation (NOICE) and additional sensitivity
studies with prescribed ICNCs of 0.02 (ICNC0p02) and
1 L-1 (ICNC1p00).
We begin by discussing the CDNC30_NOICE case.
Figure shows the LWCs and the mass mixing
ratios of cloud droplets and rain predicted by the MIMICA, COSMO-LES,
COSMO-NWP, WRF, and UM-CASIM models. Results in this figure and throughout
the paper are shown at the centre of the domain for all models. We note that
the direct comparison of results between LES and NWP models is not trivial:
the LES models in this study used wrapped boundary conditions and
time-invariant surface fluxes and therefore would always be expected to tend
towards some equilibrium cloud state. The NWP models, however, simulate the
advection of different air masses with different histories through the
domain, and changes due to differences in air masses can be conflated with
the temporal evolution of a single cloud system. With these challenges in
mind, we note that the surface is homogeneously covered in sea ice in all
models, and we expect that the centre of the domain will be representative
for our case study. In order to assess this, we show statistics from the NWP
models over a 100 km2 area in the centre of the domain in the
Supplement (Figs. S1–S3). Figure S1 shows characteristics of the
distribution of LWP and IWP within the specified 100 km2 area as
simulated by the three NWP models for the CDNC30_ICNC0p20 case. Figures S2
and S3 show statistics of LWP, IWP, and net surface longwave radiation
for the NWP models for all of the sensitivity studies. We note that the
centre-of-domain values are nearly always within the interquartile range of
the 100 km2 area values. Furthermore, centre-of-domain values are
sufficiently close to the domain medians and have similar enough responses
to changes in CDNC, CCN, and ICNC as not to change the conclusions of our
study. We expect less spatial variability in the LES models than the NWP
models, which were run with periodic boundary conditions and fixed surface
fluxes. Thus, the centre-of-domain points are representative for the domain
in both NWP and LES models.
All models produce clouds near 1 km of altitude. Despite no inclusion of
ice processes, the predicted LWC values are generally within a factor of 2
of those observed during the cloudy period. In all models, the cloud
droplet mass mixing ratios generally increase with altitude within the cloud.
The MIMICA model predicts the thickest cloud (cloud depth ∼ 600 m) with the largest cloud droplet mass mixing ratios, reaching
values greater than 0.5 gkg-1 at cloud top. The cloud depths
simulated by WRF and UM-CASIM are slightly thinner (∼ 500 m), and the cloud droplet mass mixing ratios are smaller
(∼ 0.3 gkg-1). The cloud depths produced by
COSMO-NWP are similar to those produced by WRF and UM-CASIM, but the cloud
droplet mass mixing ratios are much smaller (∼ 0.05 gkg-1). The cloud-top height predicted by COSMO-NWP is
greater than for any other model. This is consistent for all cases in this
study simulated by COSMO-NWP. We note that COSMO-NWP has the coarsest
vertical resolution of all the models participating in this study. The
COSMO-LES model produces the thinnest clouds (cloud depth ∼ 400 m) with the lowest cloud droplet mass mixing ratios
(<0.2gkg-1). COSMO-LES produces a consistent layer of rain below
cloud with mass mixing ratios ∼ 0.04 gkg-1. The
other four models, however, produce less rain with more variability.
Liquid water content and cloud
droplet and rain mass mixing ratios in the simulations with the following: a prescribed CDNC
of 30 cm-3 and no cloud ice permitted (CDNC30_NOICE); and liquid
water content derived from observations. (a) Liquid water content.
(b) Mass mixing ratios of cloud droplets. (c) Mass mixing
ratios of rain. Results are shown (from left to right) from the MIMICA,
COSMO-LES, COSMO-NWP, WRF, and UM-CASIM models. Observed liquid water
contents are shown in the rightmost column. Vertical dashed lines indicate
the beginnings and endings of the cloudy and nearly cloud-free periods.
None of the models predict the observed dissolution of the cloud during the
second half of the examined period, except perhaps UM-CASIM. We will show in
Sect. that this is generally true even if cloud
ice is included in the models. UM-CASIM predicts thinning of the cloud during
the last 6 h of simulation, suggesting a possible meteorological
contribution to dissipation, but the other two NWP models do not predict this
thinning. Previous analysis of this case has identified these clouds as
existing within the tenuous cloud regime and has suggested that the
dissipation of the cloud is related to extremely low (<1cm-3)
observed CCN concentrations. The prescribed CDNC cases would not be expected
to reproduce this effect, as the parameterization of the cloud droplet
activation is not linked to CCN availability. However, other potential causes
of the transition could be resolved by the models. In particular, the NWP
models would be expected to yield more realistic changes in meteorological
conditions due to advective transport through changes with time in the
boundary conditions applied to these models. However, the vertical
atmospheric structure at the interiors of the domains will evolve to be
different than at the boundaries. Nevertheless, the absence of this
transition in these modelling results supports the interpretation that the
LWC of these clouds is CCN limited. We will discuss this further in
Sect. when we discuss the lower prescribed CDNC
case.
In order to explain the differences between the results of the different
models, Fig. shows the liquid-phase process
rates for this simulation (autoconversion of cloud droplets to rain,
sedimentation of cloud droplets, and sedimentation of rain). The larger mass
mixing ratios of rain and the thinner cloud predicted by COSMO-LES is due to
the larger autoconversion tendencies (>1×10-4gm-3s-1 vs. 10-6 to
10-5gm-3s-1 in other models). Autoconversion rates
greater than 2×10-6gm-3s-1 exist even in regions
where the cloud droplet mass concentration is less than
0.01 gcm-3, the lower limit of the colour scale shown in
Fig. . Autoconversion rates and cloud
droplet mass mixing ratios both decrease by about 2 orders of magnitude
from their maximums near cloud top to the layer between 200 and 700 m.
Tendencies of mixing ratios of
cloud water and rainwater due to liquid-phase processes for simulations with a
prescribed CDNC of 30 cm-3 and no cloud ice permitted
(CDNC30_NOICE). (a) Autoconversion of cloud droplets to rain, (b) sedimentation of cloud droplets, and
(c) sedimentation of
raindrops. Results are shown (from left to right) from the MIMICA, COSMO-LES,
COSMO-NWP, WRF, and UM-CASIM models. Note that only WRF and UM-CASIM simulate
the
sedimentation of cloud droplets.
By dividing the mass of cloud droplets by the autoconversion rates from each
model, an autoconversion timescale can be estimated for each model. This
autoconversion timescale is less than 1 h for COSMO-LES, on the order of
several hours for COSMO-NWP, approximately 1 day for WRF and UM-CASIM, and
several days for MIMICA for this case. The COSMO-LES model also has the
greatest tendencies due to rain sedimentation
(>10-4gm-3s-1). These large sedimentation tendencies are
partially explained by the fact that COSMO-LES produces a greater mass of
rain of all the models for this case. The higher cloud droplet mixing ratios
seen in the MIMICA results are due to a combination of lower autoconversion
tendencies and a lack of cloud droplet sedimentation in this model.
Figure shows scatter plots of the
autoconversion tendencies plotted against the cloud droplet mass mixing
ratios in order to allow us to examine this process in more detail. The
large differences in autoconversion tendencies are despite the fact that the
same autoconversion scheme is used in MIMICA, COSMO-LES,
COSMO-NWP, and WRF. COSMO-LES, COSMO-NWP, and WRF all prescribed the same
maximum cloud droplet radius to be used for autoconversion (40 µm),
and MIMICA used a similar value (50 µm). The initial difference in
autoconversion tendencies between COSMO-LES and COSMO-NWP can be explained
primarily by the difference in cloud droplet mass mixing ratios: as there is
more mass of cloud droplets available to form rain in COSMO-LES,
autoconversion tendencies are greater. The autoconversion
scheme also predicts greater autoconversion rates if rain constitutes a
greater proportion of the liquid water mass within a given grid cell. This
results in a positive feedback on any other model differences that affect
autoconversion rates, enhancing differences in autoconversion rates between
COSMO-LES and COSMO-NWP. Autoconversion tendencies per unit mass of cloud
droplets are clearly greater in COSMO-LES and COSMO-NWP than in WRF, MIMICA,
and UM-CASIM for the CDNC30_NOICE case. As cloud droplet activation is
prescribed in this case, autoconversion is similarly treated in all models
except for UM-CASIM, and no frozen processes are permitted in this case, we
believe that the differences in autoconversion rates per unit cloud droplet
mass are due primarily to the differences in the representation of the cloud
droplet size distribution. MIMICA, COSMO-LES, COSMO-NWP, and WRF
represent the cloud droplet size distribution using a gamma distribution
defined by
dNdx=Axν1exp(-λ1xμ),
where x is the cloud droplet mass, and μ and ν1 are shape
parameters. The intercept and slope parameters A and λ1 are
defined by
A=μCDNCΓν1+1μλ1ν1+1μ,λ1=Γν1+1μΓν1+2μx‾-μ,
where x‾ is the mean cloud droplet mass. However, the prescribed shape
parameters are different between the different models: COSMO-LES and
COSMO-NWP used shape parameters μ=0.33 and ν1=0, and MIMICA used
μ=0.33 and ν1=1. In the WRF model, μ=0.33, and ν1 is
diagnostically calculated based on . UM-CASIM used a
different form of the gamma distribution:
dNdD=CDNC1Γ(1+ν2)λ2(1+ν2)Dν2exp(-λ2D),
where D is the cloud droplet diameter, and ν2 and λ2 are shape
and slope parameters distinct in meaning from ν1 and λ1.
λ2 is defined by
λ2=Γ(4+ν2)Γ(1+ν2)πρ6x‾1/3.
For the purposes of calculating autoconversion, UM-CASIM used a diagnostic
ν2 based on .
Tendencies of mixing ratios of cloud
water due to autoconversion vs. cloud droplet mass mixing ratios for
simulations with prescribed CDNCs and no cloud ice permitted.
Results are shown from the (a) MIMICA, (b) COSMO-LES, (c) COSMO-NWP, (d) WRF, and (e) UM-CASIM models.
Results from simulations with a prescribed CDNC of 3 cm-3 are shown
as blue circles and those with a prescribed CDNC of 30 cm-3 are
shown as red squares.
Sensitivity to prescribed CDNC
Next, we examine the CDNC03_NOICE case in order to investigate the
sensitivity of the model results to a reduction in prescribed CDNC from 30 to
3 cm-3. Figure shows the mass mixing ratios
of cloud droplets and rain. All models produce thinner clouds with lower LWCs
compared to the higher CDNC case. A stable cloud is produced by MIMICA, with
cloud thickness reduced to ∼ 300 m and cloud-top cloud
droplet mass mixing ratios reduced to ∼ 0.2 gkg-1,
but mixing ratios of rain are similar to those produced with the larger
prescribed CDNC. In COSMO-LES, two clouds are produced initially at
∼ 200 and ∼ 900 m. Available
water is removed by precipitation, and the clouds begin to dissipate towards
the end of the simulation (note that COSMO-LES simulations have ended at 34 h
since 30 August 2008, 12:00 UTC). COSMO-NWP produces a
cloud with cloud droplet mass mixing ratios reduced to ∼ 0.02 gkg-1 that thins and temporarily dissipates towards the end
of the simulation. UM-CASIM produces a stable cloud with cloud-top cloud
droplet mass mixing ratios reduced to ∼ 0.2 gkg-1,
with rain mass mixing ratios larger than those predicted when using a
prescribed CDNC of 30 cm-3. WRF produces a fog layer between the
surface and ∼ 500 m. The reduced LWPs predicted by WRF
early in the simulation, compared to the CDNC30_NOICE case, allow for greater
longwave cooling of the surface, ultimately creating an inversion layer that
tracks the top of the fog layer. This effect would not be reproduced by
COSMO-LES, despite the dissolution of the cloud, as the surface temperature
in COSMO-LES was prescribed for this study.
Liquid water contents, cloud and rain mass
mixing ratios, and tendencies due to liquid-phase processes for simulations
with a prescribed CDNC of 3 cm-3 and no cloud ice permitted
(CDNC03_NOICE). (a) Liquid water contents. (b) Mass mixing
ratios of cloud droplets. (c) Mass mixing ratios of rain. (d, e, f) Tendencies of mixing ratios of cloud droplets or rain due to
(d) autoconversion of cloud droplets to rain, (e) sedimentation of
cloud droplets, and (f) sedimentation of rain. Results are shown (from left to right)
from the MIMICA, COSMO-LES, COSMO-NWP, WRF, and UM-CASIM
models. Note that only WRF and UM-CASIM simulate the sedimentation of cloud
droplets.
Figure also shows the liquid-phase process rates for
the CDNC03_NOICE case. The reduction in the prescribed CDNC values results
in an increase in the autoconversion to rain tendencies in MIMICA and
UM-CASIM. It can be seen in Fig. that the
autoconversion tendencies are increased even after accounting for changes in
cloud droplet mass mixing ratios. Within WRF and UM-CASIM, cloud droplet
sedimentation tendencies remain similar in magnitude to those in the
CDNC30_NOICE simulation. The mass of cloud droplets available to sediment is
reduced by increased autoconversion to rain, but this is compensated for by
increased fall speeds due to the increased size of the cloud droplets. Rain
sedimentation tendencies in WRF and UM-CASIM are also similar in magnitude to
the CDNC30_NOICE case. The rates of autoconversion to rain and the rates of sedimentation
of rain predicted in COSMO-NWP are similar to those in the CDNC30_NOICE case
when the cloud thickness and LWC are greatest, but diminish to much smaller
values as the cloud dissipates. Compared to the higher CDNC case, the MIMICA
model predicts larger losses within the cloud through the sedimentation of
rain due to the larger mixing ratio of rain predicted in this case. The
COSMO-LES model predicts lower mixing ratios of rain for this case relative
to the CDNC30_NOICE case as the cloud dissipates. The changes in mass due to
sedimentation are therefore lower than in the CDNC30_NOICE case.
Sensitivity to activation scheme
We will now discuss the CCN30fixed_NOICE and CCN80fixed_NOICE cases. These
cases differ from the prescribed CDNC cases in that cloud droplet activation
is predicted based on a constant background aerosol concentration of either
80 or 30 cm-3 with median diameter 94 nm and
geometric standard deviation of 1.5 instead of being prescribed to be
30 or 3 cm-3.
Figure shows time-averaged profiles of cloud
properties for the CDNC30_NOICE, CDNC03_NOICE, CCN30fixed_NOICE, and
CCN80fixed_NOICE cases. We average over the period from 12:00 to 24:00 UTC
on 31 August (24–36 h since 12:00 UTC,
30 August) in order to exclude the initial period before a
stable cloud forms in the NWP models. We note that for the CDNC03 and CDNC30
cases in COSMO-LES, COSMO-NWP, and UM-CASIM, the CDNC is prescribed through
activation, but is permitted to vary within cloud due to evaporation and
transport. When the background CCN concentration is set to be
30 cm-3, the CDNC within cloud (column a) is ∼ 15 cm-3 in MIMICA, ∼ 25 cm-3 in COSMO-LES,
∼ 15 cm-3 in COSMO-NWP, ∼ 20 cm-3 in WRF, and ∼ 20 cm-3 in UM-CASIM.
The differences in activation fractions are more pronounced for the CCN80
cases: MIMICA, COSMO-LES, COSMO-NWP, WRF, and UM-CASIM predict in-cloud CDNCs
of ∼ 25, ∼ 60,
∼ 20, ∼ 40, and
∼ 60 cm-3, respectively. This diversity in CDNC of
15–20 or 20–60 cm-3 for the same constant CCN
concentrations underscores the variability that exists in model results and
model sensitivities to perturbations in aerosol concentrations. Unless the
models are constrained through common forcings and common scientific choices,
there will remain diversity in model results and model sensitivity for both
LES and NWP models.
Time-averaged profiles of cloud properties
and tendencies due to liquid-phase processes for simulations with no cloud
ice permitted. The first letter in subplot labels refers to column and the second to
the row. (a) Cloud droplet number concentrations. (b) Liquid water
contents. (c) Cloud droplet mass mixing ratios. (d) Rain mass
mixing ratios. Rightmost three columns: tendencies of mixing ratios of cloud
droplets or rain due to autoconversion of cloud droplets to rain (e),
sedimentation of cloud droplets (f), and sedimentation of rain (g). Results are shown (from top to bottom) from the
MIMICA (i),
COSMO-LES (ii), COSMO-NWP (iii), WRF (iv), and UM-CASIM (v)
models. The blue dotted line indicates the CDNC03 case, the red dash-dotted line
indicates the CDNC30 case, the purple dashed line indicates the CCN30fixed
case, and the solid turquoise line indicates the CCN80fixed case. Note that
only WRF and UM-CASIM simulate the sedimentation of cloud droplets.
There are many model differences making it difficult to assign variations to
particular processes, but one pair of models provides some insight as we
shall see. These differences are due in part to differences in the activation
schemes used in the different models: the activation scheme described in
is used in MIMICA, the scheme described in
and is used in COSMO-LES and
COSMO-NWP, and the scheme described in and
is used in WRF and UM-CASIM. These differences may
also be due to differences in the representation of small-scale turbulence
within the models: COSMO-NWP, WRF, and UM-CASIM have horizontal resolutions
too coarse to resolve individual updrafts. WRF and UM-CASIM therefore assume
minimum updraft velocities for activation as 0.1 ms-1. COSMO-NWP
parameterizes the updraft velocity used for activation by adding 0.8×TKE to the grid-resolved updraft velocity; TKE is the
turbulent kinetic energy. The fine resolutions of MIMICA and COSMO-LES allow
them to resolve these updrafts explicitly. Differences in sink terms across
models, such as the collision–coalescence of cloud droplets and cloud droplet
sedimentation, would also be expected to contribute to these differences. As
WRF and UM-CASIM have the same activation scheme and the same minimum
updraft velocity, we infer that remaining differences in CDNC are due to
differences in sink terms. For the CCN30fixed case, CDNCs are similar in both
models, but CDNCs simulated by UM-CASIM are greater in the CCN80fixed case.
Therefore, CDNC sinks must be similar in the CCN30fixed case, but faster for
WRF in the CCN80fixed case.
As cloud properties within the tenuous cloud regime are expected to be
dependent on CCN concentrations via changes in CDNC, it is informative to
examine how cloud properties are related to the modelled CDNC for these four
cases. With the exception of the CDNC03 case, the vertical cloud extent and
cloud droplet mass mixing ratios (column c) are similar across the different
cases in MIMICA, COSMO-LES, and UM-CASIM (differences < 100 m and
< 0.1 gkg-1, respectively). The COSMO-NWP model (subplot c.iii) shows
higher cloud altitudes and cloud droplet mass mixing ratios for the CDNC30
case. The WRF model results (subplot c.iv) generally show an increase in both
cloud vertical thickness and cloud height correlated with increasing CDNC.
The mass mixing ratios of rain within MIMICA (subplot d.i) and UM-CASIM
(subplot d.v) clearly increase with decreasing CDNCs due to increases in
autoconversion from cloud droplets (subplots e.i and e.v), mitigated somewhat
by increases in rain sedimentation rates (subplots g.i and g.v). For the
CCN80fixed case, the CDNC is sufficiently high in UM-CASIM to reduce
concentrations of rain below 10-3gkg-1. This
effect is present within WRF (subplot d.iv), but is more difficult to discern
because of coincident changes in cloud height and thickness. Within COSMO-LES
(subplot d.ii), there is a weak increase in rain mass mixing ratios with
decreasing CDNC until CDNC is reduced to 3 cm-3, at which point
rain mass mixing ratios are reduced due to cloud dissipation.
Sensitivity to prognostic aerosol
We now consider the CCN80prog_NOICE case. In these simulations, the aerosol
is initialized as in the CCN80fixed_NOICE case, but is then allowed to
evolve with time due to advection, removal by cloud droplet activation, and
resuspension upon evaporation. Figure shows
profiles vs. time of the mass mixing ratios of cloud droplets and rain, CDNC,
N50 concentrations, and potential temperature.
Cloud properties in the simulations with
prognostic aerosol and an initial CCN concentration of 80 cm-3
(CCN80prog_NOICE). (a) Cloud droplet mass mixing ratio, (b) rain mass mixing ratio,
(c) cloud droplet number
concentration, (d) N50 concentration, and (e) potential temperature. Results are shown
(from left to right) from the UCLALES-SALSA, MIMICA, COSMO-NWP, and UM-CASIM
models. Note that aerosol concentrations and CDNCs are fixed during the
2 h spin-up period in MIMICA, and N50 concentrations are not available
for COSMO-NWP.
For this case, the MIMICA and COSMO-NWP models produce results very similar
to those for the CCN80fixed_NOICE case. In COSMO-NWP, the resuspension of
aerosol upon the evaporation of cloud droplets and raindrops leads to a build-up
of aerosol below cloud, leading to an enhancement of CDNCs at cloud base,
particularly after 24 h of simulation time. The UM-CASIM model produces a
cloud that is reduced in vertical extent and liquid water content, with more
rain compared to the case without aerosol processing. The reduction in
available CCN by activation reduces CDNC, leading to larger cloud droplets
and increased autoconversion to rain. The UCLALES-SALSA model also produces a
stable cloud with cloud-top height near 1 km and cloud droplet mixing
ratios of ∼ 0.3 gkg-1, but with no autoconversion to
rain. Unlike the other models included in this study, UCLALES-SALSA does not
assume a gamma distribution for cloud droplets, and instead uses seven sectional
bins to represent the cloud droplet size distribution and explicitly
calculates drop–drop collisions using the bin representation (see
Sect. ). Therefore, the UCLALES-SALSA model does not
necessarily produce any large (>50µm) cloud droplets upon
activation, as would be implicitly assumed by a gamma distribution. The
UCLALES-SALSA model resolves narrower cloud droplet size distributions than
those represented by the other models in this study, with no cloud droplets
large enough to trigger partitioning into the rain category. Differences in
cloud thickness between MIMICA and UCLALES-SALSA (thickening in MIMICA and
thinning with time in UCLALES-SALSA) for this case are primarily due to the
different subsidence rates as described in Sect. .
Simulations performed by UCLALES-SALSA using the same lower subsidence rate
as the MIMICA simulations yielded a cloud layer with a similar LWP to the
MIMICA simulation (∼ 125 and
140 gm-2, respectively), but the cloud layer rose at an
unrealistic rate.
When the initial CCN concentration is reduced to 30 cm-3
(CCN30prog_NOICE; Fig. ), the UCLALES-SALSA model
no longer maintains a stable cloud. Instead, the larger size of cloud
droplets allows for partitioning into rain, which subsequently removes the
available aerosol by sedimentation. As the cloud thins, radiative cooling of
the cloud top weakens, resulting in less generation of turbulence. The
above-cloud temperature inversion subsequently descends due to subsidence.
Within UCLALES-SALSA, subsidence only affects the tendencies of temperature
and water vapour and does not directly alter advection of aerosols.
Therefore the temperature inversion descends into the aerosol-depleted layer,
suppressing any further entrainment of aerosol from above the cloud. The
reduction in aerosol concentrations further reduces CDNCs, leading to larger
cloud droplets and further enhances conversion to rain. The depletion of
aerosol therefore results in a positive feedback loop that ends with total
dissipation of the cloud. The MIMICA, COSMO-NWP, and UM-CASIM models,
conversely, do maintain clouds to the end of the simulation, although the
water content of the clouds is reduced. The COSMO-NWP model shows the
weakest sensitivity to the decrease in CCN concentrations, similarly to the
weak sensitivity shown in Sect. to changes in
prescribed CDNC. The vertical extent of the cloud simulated by the MIMICA
model decreases with time. This cloud has similar cloud droplet mass mixing
ratios to the case with fixed aerosol concentrations (CCN30fixed_NOICE), but
is thinner (∼ 300 m vs. ∼ 500 m).
The CDNC decreases during the simulation to ∼ 2 cm-3
after 48 h, resulting in faster autoconversion rates and larger mixing
ratios of rain. Results from the UM-CASIM model are qualitatively similar to
those with the higher initial aerosol concentration, but cloud droplet mass
mixing ratios and CDNC are lower (0.1 vs. 0.15 gkg-1 and 5 vs.
20 cm-3). The concurrent reductions in both cloud droplet mass
mixing ratios and CDNC yield only small changes in cloud droplet sizes,
so there are no large changes in rain autoconversion rates, cloud droplet
sedimentation rates, mass mixing ratios of rain, or rain sedimentation rates.
Cloud properties in the simulations with
prognostic aerosol and an initial CCN concentration of 30 cm-3
(CCN30prog_NOICE). (a) Cloud droplet mass mixing ratio, (b) rain mass mixing ratio,
(c) cloud droplet number
concentration, (d) N50 concentration, and (e) potential temperature. Results are shown
(from left to right) from the UCLALES-SALSA, MIMICA, COSMO-NWP, and UM-CASIM
models. Note that aerosol concentrations and CDNCs are fixed during the
2 h spin-up period in MIMICA, and N50 concentrations are not available
for COSMO-NWP.
Further reducing the initial CCN concentration to 3 cm-3
(CCN03prog_NOICE, Fig. ) results in dissipation of
the original cloud in UCLALES-SALSA, MIMICA, and UM-CASIM. The results of
UCLALES-SALSA are qualitatively similar to the results with an initial CCN
concentration of 30 cm-3, except that the cloud dissipates much
more quickly. The cloud completely dissipates after less than 6 h into
the simulation, while in the simulation with an initial CCN concentration of
30 cm-3, the formation of rain started after 6 h of simulation,
and complete dissipation of the cloud did not occur until the end of the 36 h simulation. The original cloud layer in the MIMICA model dissipates
after about 36 h of simulation time. A second cloud layer forms 12 h
from the beginning of the simulation at around 200 m from the surface
and rises to 700 m by the end of the simulation. Rain falling from the
upper cloud layer evaporates before reaching the lower cloud layer. This
transports moisture and aerosol vertically closer to the lower cloud layer,
where they are subsequently mixed into the lower cloud layer by turbulence.
COSMO-NWP maintains a drizzling cloud throughout most of the simulation.
Evaporation of cloud droplets and raindrops transports aerosol below cloud,
resulting in larger aerosol concentrations and larger CDNCs at cloud base
than those predicted by the other models. In UM-CASIM, reduction of the
initial aerosol concentration to 3 cm-3 results in dissipation of
the cloud by drizzle. The formation and dissipation of the cloud is not
visible in the centre-of-domain results shown here, but the aerosol number
concentrations remain depleted in the air mass where the cloud formed, which
passes through the centre of the domain 24–36 h from the start of
the simulation. The thinning of the cloud layer allows for the cooling of the surface
via longwave emission, creating a stable layer near 200 m. This
restricts any cloud from forming above this layer. This feedback will not
occur in the LES models due to the prescribed surface conditions and fluxes used
in our study.
Cloud properties in the simulations with
prognostic aerosol and an initial CCN concentration of 3 cm-3
(CCN03prog_NOICE). (a) Cloud droplet mass mixing ratio, (b) rain mass mixing ratio,
(c) cloud droplet number
concentration, (d) N50 concentration, and (e) potential temperature. Results are shown
(from left to right) from the UCLALES-SALSA, MIMICA, and UM-CASIM models.
Note that aerosol concentrations and CDNCs are fixed during the 2 h
spin-up period in MIMICA, and N50 concentrations are not available for
COSMO-NWP.
The timescale of aerosol removal depends strongly on the model and the
initial CCN concentration. UCLALES-SALSA predicts that below-cloud N50
concentrations would be unaffected for initial CCN concentrations of 3 or
80 cm-3 due to a lack of mixing to the surface after cloud
dissipation in the former case and a lack of precipitation in the latter
case. If the initial CCN concentration is 30 cm-3, UCLALES-SALSA
predicts that N50 concentrations throughout the boundary layer fall below
1 cm-3 after 36 h. The MIMICA model predicts a steady decrease
in surface N50 concentrations for all three prognostic cases simulated,
ranging from ∼ 0.4 cm-3h-1 for the
80 cm-3 case to 0.05 cm-3h-1 for the 3 cm-3
case. Aerosol removal rates are difficult to diagnose from COSMO-NWP and
UM-CASIM due to the advection of different air masses being simultaneous with
aerosol processing.
The model results shown above demonstrate a robust relationship between
decreases in CDNC, either through direct prescription or from the effects of
activation and processing, and the thinning or even collapse of the cloud
layer. However, the sensitivity of the cloud layer to decreases in CDNC
differs between models due to differences in the partitioning of cloud liquid
between cloud droplets and rain and differences in the representation of
surface properties. In the next section we build on these liquid-only results
by adding the complication of ice interactions.
Sensitivity to ice formationBase case
Figure shows the liquid and ice water contents
from the models when the CDNC is prescribed as 30 cm-3 and the ICNC
is prescribed as 0.2 L-1 (CDNC30_ICNC0p20). The IWCs predicted by
the models vary by an order of magnitude between the models, with COSMO-LES
and WRF predicting IWCs less than 0.002 gm-3, but MIMICA producing
highly variable IWCs often as great as 0.02 gm-3. We note that the
IWCs derived from observations are often greater than 0.05 gm-3,
but the uncertainty could be as great as a factor of 2, as stated in
Sect. . Even accounting for this uncertainty, all models
underestimate the IWC for this case. Any model bias in IWC does not seem to
be related to biases in LWC.
Liquid and ice water content in the
simulations with a prescribed CDNC of 30 cm-3 and a prescribed ICNC
of 0.2 L-1 (CDNC30_ICNC0p20), and liquid and ice water content derived from observations. (a) Liquid water contents, (b) ice water contents. Results are shown
(left to right) from the MIMICA, COSMO-LES, COSMO-NWP, WRF, and UM-CASIM
models. Values derived from observations are shown in the rightmost column.
Vertical dashed lines indicate the beginnings and endings of the cloudy and
nearly cloud-free periods.
Figure shows the mass mixing ratios
of cloud droplets, rain, cloud ice crystals, snow, and graupel from the
models when the CDNC is prescribed as 30 cm-3 and the ICNC is
prescribed as 0.2 L-1 (CDNC30_ICNC0p20). We note with comparison
to Fig. that the introduction of ice does
not change cloud height or cloud depth by more than 100 m in any
model, and cloud mass mixing ratios change by less than 20 % in all
models. However, mass mixing ratios of rain are reduced in the results of the
MIMICA and WRF models.
Cloud mass mixing ratios in
the simulations with a prescribed CDNC of 30 cm-3 and a prescribed
ICNC of 0.2 L-1 (CDNC30_ICNC0p20). (a) Mass mixing ratios of
cloud droplets, (b) mass mixing ratios of rain, (c) mass
mixing ratios of cloud ice crystals, (d) mass mixing ratios of snow,
and (e) mass mixing ratios of graupel. Results are shown (left
to right) from the MIMICA, COSMO-LES, COSMO-NWP, WRF, and UM-CASIM models. Note that
WRF does not possess a graupel category.
The form of frozen mass depends on which model is used: only MIMICA produces
a significant amount of graupel, and only WRF predicts that most frozen water
would be snow. COSMO-LES, COSMO-NWP, and UM-CASIM predict the frozen water to
exist predominantly as cloud ice crystals, but UM-CASIM also predicts a small
amount of mass in the snow category. Within MIMICA, any collision between a
liquid hydrometeor and a frozen hydrometeor will move the resulting mass to
the graupel category. Within all other models, collisions between cloud ice
crystals smaller than 160 µm and cloud droplets do not form
graupel. Collisions between ice crystals larger than 160 µm and
cloud droplets can produce graupel in COSMO-LES and COSMO-NWP, but the
collision and sticking efficiencies are small. So even if large ice crystals
are present, this remains a negligible source of graupel. Since cloud ice
crystals are the dominant form of frozen hydrometeors in all other models
aside from WRF and cloud droplets are the dominant form of liquid
hydrometeors in all models, no graupel is formed in COSMO-LES, COSMO-NWP, or
UM-CASIM. As mentioned in Sect. , the set-up of WRF used
in this study does not possess a graupel category, so riming by snow will
increase the mass of snow instead of forming graupel in WRF.
In order to examine the causes and implications of these differences in ice
between the models, Fig. shows time-averaged profiles of
process rates affecting ice crystals and snow for each of the models for the
prescribed CDNC cases. We average over 31 August 2008
from 12:00 to 24:00 UTC in order to exclude the initial period of the NWP
models before a stable cloud forms. We note that mass mixing ratios of snow
(column b) are often an order of magnitude less than cloud ice mass mixing
ratios (column a), but even these small amounts of snow can have significant
effects on cloud species or water vapour mixing ratios (columns f and h).
Within COSMO-LES, except the CDNC30_ICNC1p00 case, insignificant
autoconversion to snow occurs (subplot c.ii) and nearly all frozen cloud mass
remains as cloud ice crystals (subplot a.ii). The cloud ice grows by deposition
within cloud and sublimates below cloud (subplot g.ii), frequently sublimating
completely before reaching the surface. COSMO-NWP (row iii) behaves similarly
to COSMO-LES, but the cloud ice grows by deposition throughout the boundary
layer (subplot g.iii). As stated previously, only WRF maintains significant
mixing ratios of snow (subplot b.iv). Autoconversion to snow proceeds more
quickly than in the other models for the same cloud ice crystal mixing ratios
(compare subplots c.iv and a.iv). The snow that is produced through
autoconversion subsequently grows efficiently by the riming of cloud droplets
(subplot e.iv) and deposition of water vapour (subplot h.iv). UM-CASIM simulates
the greatest autoconversion rates of all the models (subplot c.v). This is in
part due to UM-CASIM producing the greatest cloud ice crystal mixing ratios
of all the models (subplot a.v), but autoconversion proceeds more quickly even
for similar cloud ice mixing ratios. The snow produced by UM-CASIM grows
efficiently by the deposition and collection of cloud water (subplots h.v and e.v),
but also sediments more quickly per unit mass than in any other model
(subplot f.v) and sublimates quickly below cloud (subplot h.v), and thus the
mass of snow maintained in the atmosphere is small.
Tendencies of ice and snow mass due to processes
affecting frozen cloud mass for the prescribed CDNC simulations. The first letter
in subplot labels refers to column and the second to the row. Mass mixing ratios of
cloud ice (a) and snow (b), tendencies of cloud ice and snow
mass due to autoconversion to snow (c), riming by cloud ice (d), riming by snow (e),
sedimentation of snow (f), deposition+sublimation of cloud ice (g), and deposition+sublimation
of snow (h). Results shown (from top to bottom) for MIMICA (i),
COSMO-LES (ii), COSMO-NWP (iii), WRF (iv), and UM-CASIM (v).
The differences in process rates between models are due to both differences
in the parameterization of the physical processes and differences in
the representation of the size distributions of the frozen cloud species in
the different models. Additional contributions to these differences would
come from differences in model meteorology and model resolution. In the next
section we will examine the sensitivity to ICNC in the context of CCN and
CDNC changes.
Sensitivity to CDNC, CCN, and ICNCPrescribed CDNC and fixed aerosol simulations
In order to summarize our results with different prescribed ICNCs,
Fig. shows box plots of the LWP (including cloud
droplets and rain), IWP (including cloud ice crystals, snow, and graupel),
and surface net LW radiation from each model for all of the CDNC30, CDNC03,
CCN30fixed, and CCN80fixed cases during the period after
31 August 2008, 12:00 UTC. For the three NWP models, we
show a similar figure with spatial statistics for a 100 km2 area as
Fig. S2. The NWP models show more variation across time because they include
time-varying large-scale features not considered by the LES models. We note
that we do not expect the prescribed CDNC or prescribed CCN cases to capture
the cloudy to nearly cloud-free transition, so we do not attempt to
sample the models during these observed time periods. However, if the tenuous
cloud hypothesis is correct, the cloud states resulting in each model for the
cases with greater prescribed CDNC and CCN concentrations would be expected
to be more representative of the cloudy period, and the cloud states for
the cases with lesser prescribed CDNC and CCN concentrations would be
expected to be more representative of the nearly cloud-free period. Our
choice of time period allows 24 h for the models to reach a
representative state and consists of 24 h of modelled time for the three
NWP models and MIMICA. The COSMO-LES results include 7 h of model
time before averaging, and the averaging period covers 9 h of modelled
time. We note that the choice of averaging period is arbitrary, but our
conclusions are not sensitive to changes in the averaging period, with a few
exceptions: first, the initial period required for each NWP model to form a
liquid cloud above the surface must be excluded (6–18 h). Second, the
MIMICA model predicts increased glaciation of the cloud with time in the two
ICNC1p00 cases, with LWP, IWP, and surface net LW radiation steadily
decreasing in magnitude with time. Third, the UM-CASIM model predicts that
the cloud altitude decreases after ∼ 36 h of simulation for
all cases in which a cloud is simulated, as can be seen in e.g.
Figs. , ,
, and . This leads to
decreases in LWP, IWP, and the magnitude of the surface net LW radiation if
this time period is included. Fourth, the COSMO-NWP model predicts a stable
frozen cloud in the CDNC30_ICNC1p00, CDNC03_ICNC0p20, and CDNC03_ICNC1p00
cases until 30 h of simulation time (31 August,
18:00 UTC). After 30 h a drizzling mixed-phase cloud forms, similar to the
results shown after 30 h of simulation in
Fig. . All of these effects will be
discussed further later in this section.
Water paths and net longwave radiation
for all simulations without aerosol processing. (a) Liquid water path,
(b) ice water path, (c) surface net longwave radiation. Each
subplot shows results from a single model. From left to right: MIMICA,
COSMO-LES, COSMO-NWP, WRF, and UM-CASIM. Simulations with prescribed CDNCs of
3 (CDNC03) and 30 cm-3 (CDNC30) are shown as blue
and red boxes, respectively, and simulations with prescribed CCN
concentrations of 30 (CCN30fixed) and 80 cm-3
(CCN80fixed) are shown as purple and turquoise boxes, respectively. Within
each subplot, the ICNC is increased from left to right as 0,
0.02, 0.2, and 1 L-1. Boxes show the
interquartile range over model results after 31 August
12:00 UTC, and the black horizontal lines denote the medians. Hatched regions
indicate observed interquartile range for the cloudy period, and the green
shaded regions indicate the range for the nearly cloud-free period.
Figure also shows the observed interquartile range
for the cloudy and nearly cloud-free periods as hatched and shaded
regions, respectively. These periods are defined and discussed in
Sect. . The interquartile range plotted accounts for time
variance in the observations. We do not explicitly account for observational
error, but random observational error will contribute to this time variance.
The median LWP predicted by the models spans nearly 2 orders of magnitude,
from 2.5 gm-2 for the COSMO-LES CDNC03_ICNC1p00 simulation to
190 gm-2 for the MIMICA CDNC30_ICNC0p02 case. The MIMICA model
tends to produce the largest LWPs. COSMO-LES produces the smallest LWPs for
the CDNC03 cases, and COSMO-NWP produces the smallest LWPs for all other
cases simulated. Every model for every value of ICNC shows an increase
in LWP as CDNC is increased from 3 to 30 cm-3, and
almost every model shows an increase in LWP as the fixed CCN concentration is
increased from 30 to 80 cm-3. However, the magnitude
of this increase varies greatly from model to model. Notably, COSMO-NWP shows
the smallest differences in LWP between different cases, with no significant
change in LWP between the CCN30fixed and CCN80fixed cases. We noted earlier
in Sect. that a greater fraction of cloud droplet
mass autoconverts to rain in COSMO-NWP compared to MIMICA, WRF, or UM-CASIM,
regardless of the prescribed CDNC value chosen for activation. Therefore, a
larger fraction of the liquid in the COSMO-NWP results consists of rain as
opposed to cloud droplets compared to the other models. As the CDNC is
decreased, either through changes in the prescribed CDNC or changes in the
CCN concentration, further losses in cloud droplet mass mixing ratios are
partially compensated for by increases in rain mass, reducing differences in
total LWP. We showed in Sect. that MIMICA generally predicts
less autoconversion than the other models; as a result the proportion of
the LWP composed of rain in MIMICA is less, so it shows the greatest
sensitivity to changes in CDNC.
In general, the model results show decreases in LWP with increasing ICNC, but
these changes are generally small relative to the sensitivity to our choice
of representation of cloud droplet activation. Larger prescribed ICNCs
increase the removal of liquid water through riming and through deposition via
the Wegener–Bergeron–Findeisen process (see Fig. ).
The MIMICA model predicts almost complete glaciation for
ICNC=1L-1, and therefore produces a much reduced LWP for those cases.
LWPs within COSMO-NWP are reduced to near zero for the first 30 h of the
CDNC30_ICNC1p00, CDNC03_ICNC0p20, and CDNC03_ICNC1p00 COSMO-NWP
simulations due to glaciation of the cloud, but after 30 h a drizzling
cloud forms with LWP not strongly dependent on the prescribed ICNC
concentration.
Median IWPs predicted by the models for non-zero ICNC range from ice free for
the MIMICA CDNC03_ICNC0p02 case to 7.2 gm-2 for the UM-CASIM
CDNC30_ICNC1p00 case. The model results show increases in IWP with
prescribed ICNC, except the MIMICA ICNC1p00 cases in which the cloud
glaciates and dissipates. If a shorter averaging period was used, the IWPs
for these two cases would be larger than those for the ICNC0p20 cases. The
IWPs predicted by WRF and UM-CASIM are roughly linear with respect to the
prescribed ICNC concentration over the range used here: each 10-fold
increase in ICNC increases the IWP by roughly a factor of 10. Within
COSMO-LES, increases in IWP are sub-linear with respect to increases in ICNC:
the IWP increases by a factor between 5.3 and 7.6 as the prescribed ICNC is
increased by a factor of 10 from 0.02 to 0.2 L-1.
IWPs are also sub-linear with respect to ICNC in COSMO-NWP: the IWP increases
by a factor of either 2.8 or 3.3 as the prescribed ICNC is increased by a
factor of 5 from 0.2 to 1 L-1. Median IWPs also
generally increase with increases in CDNC or increases in CCN concentrations
due to the increased cloud water available to freeze and form ice.
The net surface LW radiation within each model is generally well correlated
with the LWP within each model. As has been discussed in
, Arctic clouds have a net warming effect over sea ice
due to the high albedo of the surface and the low angle of incoming solar
radiation. Variability in the surface net LW is greater for cases with lower
LWPs than for cases with high LWPs, as LW emission by clouds saturates
for large values of LWP. The LW dependence on LWP is stronger in the LES
models than in the NWP models. This is primarily due to the experimental
set-up: within the NWP models the surface temperature is predicted in part
based on radiative flux balance, whereas it is held fixed in the LES models.
When there is less cloud, less LW radiation is re-emitted back towards the
surface, and the surface would be expected to cool more quickly, which would
then reduce the LW emission from the surface.
For the MIMICA, COSMO-LES, WRF, and UM-CASIM models, a CDNC between
3 and 30 cm-3 could be prescribed that yields an LWP
within the interquartile range of observed LWP during the cloudy period,
but this prescribed CDNC value is not consistent across models.
Unfortunately, in-cloud CDNC measurements were not available for the period
studied here, so the models cannot be constrained based on this measurement.
Also, as discussed above, the CDNC–LWP relationship for this case appears to
be dominated by the partitioning of liquid water between cloud droplets and
rain within each model, which is often tunable through the cloud droplet size
distribution parameters or a parameter in the autoconversion scheme such as
the maximum cloud droplet size. LWPs consistent with those observed during
the nearly cloud-free period were produced by simulations in which the cloud
dissipated, regardless of the mechanism of cloud dissipation. The cloud
glaciates in MIMICA simulations with a prescribed ICNC of 1 L-1,
and the cloud temporarily glaciates in the CDNC30_ICNC1p00,
CDNC03_ICNC0p20, and CDNC03_ICNC1p00 COSMO-NWP simulations. The cloud rains
out in COSMO-LES simulations with a prescribed CDNC of 3 cm-3 and
in the COSMO-NWP simulation with a prescribed CDNC of 3 cm-3 and no
cloud ice.
The median IWP from each model for every case is less than the median
observed IWP for the cloudy period. However, as discussed in
Sect. , there is a large uncertainty in the observed IWP,
which is partially responsible for the large time variance in the observed
IWP. For COSMO-LES, COSMO-NWP, and UM-CASIM, a prescribed ICNC of
1 L-1 is required to produce a median IWP within the interquartile
range of the observed IWP. The MIMICA model produces an IWP within this range
with an ICNC of 0.2 L-1. As noted previously, the MIMICA model
predicts glaciation and dissipation if an ICNC of 1 L-1 is
prescribed, and the averaging period used here includes the dissipation of
the cloud. If a shorter averaging period was used, the IWP for these two
cases would be larger than those for the ICNC0p20 cases.
Median surface net LW radiation from nearly all WRF and UM-CASIM simulations
with LWP>75gm-2 is consistent with the observations for the
cloudy period. However, despite larger LWPs, MIMICA predicts too much LW
emission. This is due in part to the prescribed surface temperatures in our
experimental set-up being too warm, as described above. This also contributes
to the discrepancy between the LW emission observed during the
nearly cloud-free period and the MIMICA and COSMO-LES results with LWPs
consistent with the nearly cloud-free period.
Prognostic aerosol simulations
Figure shows a similar plot to
Fig. for the cases with prognostic aerosol
processing. For the COSMO-NWP and UM-CASIM models, we show a similar figure
with spatial statistics for a 100 km2 area as Fig. S3. Here we also
include N50 concentrations at 20 m from the surface for consistency
with the measurement inlet height. Note that N50 concentrations are not
available from COMSO-NWP. N50 concentrations from the other three models for
the CCN30prog simulations overlap those observed for the cloudy
period, except for the UM-CASIM CCN30prog_NOICE case, which yields greater
N50 concentrations. N50 concentrations from MIMICA and UM-CASIM for the
CCN03prog cases overlap those observed for the nearly cloud-free
period. The UCLALES-SALSA CCN03prog_NOICE simulation predicts very little
depletion of N50 from the initial values, as discussed in Sect. .
The MIMICA and UM-CASIM models simulate clouds with reduced vertical extents,
lower LWCs, and therefore lower LWPs with prognostic aerosol than with
time-invariant aerosol concentrations. Similarly, IWPs are also lower due to
the lower amount of liquid water available to freeze. The LWPs simulated by
MIMICA with an initial CCN concentration of 30 cm-3 are consistent
with observations during the cloudy period. UM-CASIM and UCLALES-SALSA
produce LWPs consistent with the cloudy period with initial CCN
concentrations of 80 cm-3. All simulations in which the cloud layer
dissipated (initial CCN concentration of 3 cm-3 in all models and
initial CCN concentration of 30 cm-3 with UCLALES-SALSA) produce
LWPs within measurement error of the nearly cloud-free period.
Cloud and surface properties for all
simulations with prognostic aerosol. (a) Liquid water path,
(b) ice water path, (c) surface net longwave radiation, and (d) N50
concentrations at 20 m from the surface. Each subplot shows results
from a single model. From left to right: UCLALES-SALSA, MIMICA, COSMO-NWP,
and UM-CASIM. Simulations with an initial CCN concentration of
3 (CCN03prog), 30 (CCN30prog), and
80 cm-3 (CCN80prog) are shown as purple, orange, and green boxes,
respectively. Within each subplot, the ICNC is increased from left to right
as 0, 0.02, and 0.2 L-1. Boxes show the
interquartile range over model results after 31 August
12:00 UTC, and the black horizontal lines denote the medians. Hatched regions
indicate the observed interquartile range for the cloudy period, and the green
shaded regions indicate the range for the nearly cloud-free period. Note
that N50 concentrations are not available from the COSMO-NWP model.
When the initial CCN concentration is 80 cm-3, UCLALES-SALSA,
MIMICA, and UM-CASIM predict that below-cloud N50 concentrations remain above
50 % of initial N50 concentrations (see Fig. ). This
reduction in aerosol number is due to in-cloud processing and drizzle
deposition to the surface offset by the resuspension of aerosol from
evaporation and sublimation of hydrometeors. An initial CCN concentration of
30 cm-3 yields N50 concentrations at 20 m consistent with
observations for all cases in which this information is available, except for
the UCLALES-SALSA case and the UM-CASIM case with no ice nucleation. In the
former, the cloud dissipates and N50 is depleted throughout the boundary
layer. The latter case produces the least rain of all the cases simulated
with an initial CCN concentration of 30 cm-3 and has the least
removal of aerosol to the surface. Median N50 at 20 m for all cases
with an initial CCN concentration of 3 cm-3 is below
1 cm-3, except for the UCLALES-SALSA results, in which N50 is depleted
in cloud, but no mixing of the depleted layer with lower layers occurs
following cloud dissipation (see Fig. ). There is no
clear effect across models of changes in prescribed ICNC on modelled N50
concentrations.
Conclusions
In this study, we have compared the results of three LES models and three NWP
models for a tenuous cloud regime case study observed during the 2008 ASCOS
field campaign. We began with simulations using prescribed CDNC and
prescribed ICNC, progressed to simulations with prognostic CDNC based on a
constant aerosol size distribution, and finally showed simulations using
prognostic aerosol processing along with prognostic CDNC. Our key findings
are the following.
Our modelling results strongly support the hypothesis that the LWC, and hence
the radiative effects, of these clouds are highly sensitive to CCN
concentrations; in order words, they are CCN limited. For the observed
meteorological conditions, all models predict that the cloud does not
collapse as observed when the CCN concentration is held constant at the value
observed during the cloudy period, but the clouds thin or collapse as the CCN
concentration is reduced. Cloud dissipation due to glaciation is predicted
only by the MIMICA model and only for a prescribed ICNC of 1 L-1,
the largest value tested in this study. As the IWP was generally
underestimated compared to the observed IWP, it is possible that the
contribution of glaciation to dissipation was also underestimated. Global and
regional models with either prescribed CDNCs or prescribed aerosol
concentrations would not reproduce cloud dissipation due to low CCN
concentrations and therefore would not capture this source of variability in cloud
LWC and hence cloud radiative effects. This suggests that linkages
between aerosol and clouds need to be considered for weather and climate
predictions in this region. In particular, we recommend that studies are
carried out to determine if CCN-controlled cloudiness has a remote effect on
important weather phenomena such as mid-latitude blocking. If it does, then
we recommend that aerosol–cloud interactions be included to capture the
impact on the more populated mid-latitude regions.
All models predict increasing LWP with increasing CDNC, either through
prescribed CDNC values or changes in available CCN concentrations. The
increases in LWP and subsequent decreases in surface net LW radiation with
increasing CCN concentrations or prescribed CDNC suggest that increased
aerosol concentrations in the high Arctic during the clean summer period
would have a warming effect on the surface, potentially resulting in more
thinning of sea ice or a delay in autumn freeze-up events. Our results
suggest this effect would be most dramatic when CCN concentrations increase
beyond the threshold value required to prevent cloud dissipation.
Most models simulate increasing IWP with increasing prescribed ICNC and
decreasing LWP with increasing ICNC due to increased efficiency of the WBF
process with increased ICNC. This is consistent with the results of previous
investigations of the sensitivity of Arctic mixed-phase cloud to the
representation of ice nucleation e.g.. However, the effects of changes in ICNC on LWP and surface net LW
were generally weaker than the effects of changes in CDNC or CCN across the
ranges tested in this study. This is consistent with results found by
in which the total water path and net surface LW were to
first order determined by CDNC or CCN concentrations, rather than INP
concentrations, for CCN and INP perturbations of similar magnitude as
considered in this study. However, for larger INP perturbations (exceeding
1 L-1) in a low-INP regime, INP perturbations were seen to
potentially offset, if not reverse, the cloud response to CCN perturbations.
If INP concentrations in the Arctic were to increase beyond 1 L-1
due to changes in transport from low latitudes or increases in local
emissions, these could induce large changes in cloud properties. However,
this value is greater than those observed previously in the high Arctic
.
Despite some common model behaviours, there is large inter-model diversity in
the sensitivities of the models to changes in CDNC or CCN concentrations. The
change in LWP due to an increase in prescribed CDNC from 3 to
30 cm-3 varies from ∼ 10 to
∼ 100 gm-2 depending on the choice of model alone.
Cloud dissipation was predicted by the COSMO-LES, COSMO-NWP, and WRF models
for a prescribed CDNC of 3 cm-3, suggesting that the critical CDNC
for these models was between 3 and 30 cm-3. The critical CDNC for
the other models must be less than 3 cm-3. In the prognostic
aerosol cases, the critical initial CCN concentration was between 30 and
80 cm-3 for the UCLALES-SALSA model and between 3 and
30 cm-3 for the MIMICA and UM-CASIM models. The COSMO-NWP model did
not predict dissipation of the cloud for any of the prognostic aerosol cases.
We did not test the sensitivity of these critical values to model processes,
but it is likely that they are sensitive to the specific set-up of each model
used in this study, specifically regarding cloud droplet size distributions
and the representation of the autoconversion of cloud droplets to rain. Faster
autoconversion rates per unit cloud droplet mass are associated with lower
sensitivities in all cloud properties to changes in prescribed CDNC or CCN
concentrations. Large differences in autoconversion rates per unit cloud
droplet mass were simulated despite a similar treatment of autoconversion in
four of the models, even in cases with prescribed cloud droplet activation
and no frozen cloud processes permitted. Our results therefore suggest that
some caution is necessary in interpreting the results of any single model,
including the sensitivities of model results to perturbations in aerosol
concentrations. Properly estimating aerosol–cloud interactions requires
careful consideration regarding the representation of cloud droplet size
distributions, as well as the choice of autoconversion scheme and the
parameters set therein if an empirical formulation is chosen. Our results
also suggest that observations should aim to constrain the representation of
rain formation and extend the validity of parameterizations to the Arctic
domain. We therefore recommend that future observational campaigns aim to
perform in situ observations of cloud LWC, IWC, and hydrometeor size
distributions, as well as aerosol size and concentration profiles above and
below cloud.
The strength of aerosol sources will be critical for the stability of tenuous
Arctic clouds. When aerosol removal by activation into cloud droplets was
included in the simulations, this decreased simulated CDNCs and LWPs. The
rate of depletion of potential CCN within the boundary layer varied strongly
between different models and depending on the initial aerosol concentration.
For greater initial aerosol concentrations, precipitation formation was
suppressed, decreasing the removal of aerosol to the surface. This supports a
positive feedback mechanism whereby increasing aerosol concentrations
suppress drizzle formation, reducing the sink of aerosol to the surface. We
note that we did not investigate the replenishment of CCN by surface sources
or by aerosol nucleation and growth, but have shown
that cloud-top entrainment is important for CDNC (and hence cloud radiative
properties) in this case. Entrainment would be included in the results
presented here, but as we applied constant initial CCN concentrations
throughout the simulated atmosphere, the above-cloud aerosol concentration
available for entrainment was identical to the initial boundary layer aerosol
concentration.
A potentially important feedback is that cooling of the sea-ice surface
following cloud dissipation increases atmospheric stability near the surface,
further suppressing cloud formation. Surface fluxes were predicted to be
small by the NWP models so long as a sufficiently thick cloud layer was
simulated (surface fluxes were prescribed in the LES models). However, under
thin-cloud or cloud-free conditions, the cooling of the surface due to LW
emission increased the stability of the near-surface atmospheric layer. The
WRF model with a prescribed CDNC of 3 cm-3 predicts that any
subsequent cloud will be constrained to a shallow mixed layer at the surface,
resulting in surface fog (Fig. ). This effect can also
be seen in the potential temperature profiles predicted by UM-CASIM for the
CCN03prog_NOICE case (Fig. ). Therefore, this
suggests that linkages between clouds, surface temperatures, and atmospheric
stability may need to be considered for weather and climate predictions in
this region.
We primarily focus on cloud microphysical processes in this work, but it is
important to also note the contribution of large-scale atmospheric
circulation patterns to cloud cover and thickness e.g. as
well as sea ice e.g.. However, our results highlight
the sensitivity of high Arctic clouds to CCN concentrations, the importance
of the model representation of rain formation in clouds for correctly
capturing this sensitivity, and the interactions between clouds, surface
temperatures, and atmospheric stability. Future studies of the interactions
between Arctic clouds, sea ice, and climate must take account of all of these
findings.
There are many aspects of high Arctic aerosol–cloud interactions that were
beyond the scope of this study to address. Future studies should aim to
address the possible role of aerosol replenishment by new-particle formation,
surface sources, and transport using models that include coupled aerosols and
chemistry with active sources and sinks. The formation of new clouds or fog
after dissipation events as aerosol concentrations are replenished also needs
to be investigated. More case studies based on additional observational
campaigns need to be performed. Uncertainty analyses are necessary to explore
the simultaneous contributions of multiple compensating factors. More
investigation of surface thermodynamics and feedbacks is also necessary.
The intercomparison
model output used for our analysis is available at 10.5281/zenodo.1326922 (Stevens et al., 2018).
The supplement related to this article is available online at: https://doi.org/10.5194/acp-18-11041-2018-supplement.
RS
led the experiment design and analysis and was the primary author of the paper.
Experiments using the UCLALES-SALSA, MIMICA, COSMO-LES,
COSMO-NWP, WRF, and UM-CASIM models were performed by TR, AD, KL, AP and GE,
CD, and RS, respectively. Necessary development and support for
the UCLALES-SALSA, MIMICA, COSMO-LES, COSMO-NWP, WRF, and UM-CASIM
models were provided by TR, SR, JT, AL, and HK, AD and AE, KL and CH,
AP, GE, and UL, CD and PC, and RS, AH, BS, JW, and PF,
respectively. RS, KL, CD, AD, AP, GE, TR, CH, AE, KC, and PF all contributed
to the experiment design, analysis, and writing of the paper.
The authors declare that they have no conflict of
interest.
This article is part of the special
issue “BACCHUS – Impact of Biogenic versus Anthropogenic emissions on
Clouds and Climate: towards a Holistic UnderStanding (ACP/AMT/GMD
inter-journal SI)”. It is not associated with a conference.
Acknowledgements
We thank the two anonymous reviewers for their comments on this paper.
We gratefully acknowledge support from the European Union's Seventh Framework
Programme (FP7/2007-2013) with the project Impact of Biogenic versus Anthropogenic emissions
on Clouds and Climate: towards a Holistic UnderStanding (BACCHUS; grant no. 603445) and the
European Research Council projects ECLAIR (grant no. 646857) and C2Phase (grant no. 714062). We acknowledge the
use of the MONSooN
system, a collaborative facility supplied under the Joint Weather and Climate
Research Programme, a strategic partnership between the UK Met Office and the
Natural Environment Research Council. We also acknowledge the use of the JASMIN
system operated by Centre for Environmental Data Archival (CEDA) as well as
the Swiss National Supercomputing Centre (CSCS).
Birgit Wehner, Douglas Orsini, Maria Martin, and Staffan Sjögren are much
appreciated for providing the size-resolved
particle number and the CCN observations. Caroline Leck and Michael Tjernström are specifically
thanked for their coordination of ASCOS. The
Swedish Polar Research Secretariat provided access to the icebreaker Oden and
logistical support. We would like to thank Joseph Sedlar, Thorsten Mauritsen,
and Matthew Shupe for the observational data reprinted in this paper
and for their comments on an early version of the paper.
Edited by: Daniel J. Cziczo
Reviewed by: two anonymous referees
ReferencesAbdul-Razzak, H. and Ghan, S. J.: A parameterization of aerosol activation: 2.
Multiple aerosol types, J. Geophys. Res.-Atmos., 105,
6837–6844, 10.1029/1999JD901161, 2000.Abdul-Razzak, H., Ghan, S. J., and Rivera-Carpio, C.: A parameterization of
aerosol activation: 1. Single aerosol type, J. Geophys. Res.-Atmos., 103, 6123–6131, 10.1029/97JD03735, 1998.Avramov, A. and Harrington, J. Y.: Influence of parameterized ice habit on
simulated mixed phase Arctic clouds, J. Geophys. Res., 115,
1–14, 10.1029/2009JD012108, 2010.Bigg, E. K.: Ice forming nuclei in the high Arctic, Tellus B, 48, 223–233,
10.1034/j.1600-0889.1996.t01-1-00007.x, 1996.Bigg, E. K. and Leck, C.: Cloud-active particles over the central Arctic
Ocean, J. Geophys. Res.-Atmos., 106, 32155–32166,
10.1029/1999JD901152, 2001.Bigg, E. K., Leck, C., and Nilsson, D.: Sudden changes in arctic atmospheric
aerosol concentrations during summer and autumn, Tellus B, 48, 254–271,
10.1034/j.1600-0889.1996.t01-1-00009.x, 1996.Birch, C. E., Brooks, I. M., Tjernström, M., Shupe, M. D., Mauritsen, T., Sedlar, J., Lock, A. P., Earnshaw, P., Persson, P. O. G.,
Milton, S. F., and Leck, C.: Modelling atmospheric structure, cloud and their response to CCN in the central Arctic: ASCOS case studies,
Atmos. Chem. Phys., 12, 3419–3435, 10.5194/acp-12-3419-2012, 2012.Boisvert, L. N. and Stroeve, J. C.: The Arctic is becoming warmer and wetter
as revealed by the Atmospheric Infrared Sounder, Geophys. Res. Lett., 42, 4439–4446, 10.1002/2015GL063775, 2015.Chan, M. A. and Comiso, J. C.: Arctic Cloud Characteristics as Derived from
MODIS, CALIPSO , and CloudSat, J. Climate, 26, 3285–3306,
10.1175/JCLI-D-12-00204.1, 2013.Chang, R. Y.-W., Leck, C., Graus, M., Müller, M., Paatero, J., Burkhart, J. F., Stohl, A., Orr, L. H., Hayden, K., Li, S.-M., Hansel, A., Tjernström, M.,
Leaitch, W. R., and Abbatt, J. P. D.: Aerosol composition and sources in the central Arctic Ocean during ASCOS, Atmos. Chem. Phys., 11,
10619–10636, 10.5194/acp-11-10619-2011, 2011.Clough, S. A., Shephard, M. W., Mlawer, E. J., Delamere, J. S., Iacono, M. J.,
Cady-Pereira, K., Boukabara, S., and Brown, P. D.: Atmospheric radiative
transfer modeling: A summary of the AER codes, J. Quant. Spectrosc. Ra., 91, 233–244,
10.1016/j.jqsrt.2004.05.058, 2005.Corbett, J. J., Lack, D. A., Winebrake, J. J., Harder, S., Silberman, J. A., and Gold, M.: Arctic shipping emissions inventories and future
scenarios, Atmos. Chem. Phys., 10, 9689–9704, 10.5194/acp-10-9689-2010, 2010.Curry, J. A., Schramm, J. L., and Ebert, E. E.: Impact of clouds on the
surface radiation balance of the Arctic Ocean, Meteorol. Atmos. Phys., 51, 197–217, 10.1007/BF01030494, 1993.Curry, J. A., Hobbs, P. V., King, M. D., Randall, D. A., Minnis, P., Isaac,
G. A., Pinto, J. O., Uttal, T., Bucholtz, A., Cripe, D. G., Gerber, H.,
Fairall, C. W., Garrett, T. J., Hudson, J., Intrieri, J. M., Jakob, C.,
Jensen, T., Lawson, P., Marcotte, D., Nguyen, L., Pilewskie, P., Rangno, A.,
Rogers, D. C., Strawbridge, K. B., Valero, F. P. J., Williams, A. G., and
Wylie, D.: FIRE arctic clouds experiment, B. Am. Meteorol. Soc., 81, 5–29,
10.1175/1520-0477(2000)081<0005:FACE>2.3.CO;2, 2000.Ekman, A. M. L., Wang, C., Ström, J., and Krejci, R.: Explicit
Simulation of Aerosol Physics in a Cloud-Resolving Model: Aerosol Transport
and Processing in the Free Troposphere, J. Atmos. Sci.,
63, 682–696, 10.1175/JAS3645.1, 2006.Fan, J., Ovtchinnikov, M., Comstock, J. M., McFarlane, S. A., and Khain, A.:
Ice formation in Arctic mixed-phase clouds: Insights from a 3-D
cloud-resolving model with size-resolved aerosol and cloud microphysics,
J. Geophys. Res., 114, D04205, 10.1029/2008JD010782,
2009.Fountoukis, C. and Nenes, A.: Continued development of a cloud droplet
formation parameterization for global climate models, J. Geophys. Res.-Atmos., 110, 1–10, 10.1029/2004JD005591, 2005.Fridlind, A. M., van Diedenhoven, B., Ackerman, A. S., Avramov, A., Mrowiec,
A., Morrison, H., Zuidema, P., and Shupe, M. D.: A FIRE-ACE/SHEBA Case Study
of Mixed-Phase Arctic Boundary Layer Clouds: Entrainment Rate Limitations on
Rapid Primary Ice Nucleation Processes, J. Atmos. Sci.,
69, 365–389, 10.1175/JAS-D-11-052.1, 2012.Fu, Q. and Liou, K. N.: Parameterization of the Radiative Properties of Cirrus
Clouds, J. Atmos. Sci., 50, 2008–2025,
10.1175/1520-0469(1993)050<2008:POTRPO>2.0.CO;2, 1993.Furtado, K. and Field, P.: The Role of Ice Microphysics Parametrizations in
Determining the Prevalence of Supercooled Liquid Water in High-Resolution
Simulations of a Southern Ocean Midlatitude Cyclone, J. Atmos. Sci., 74, 2001–2021, 10.1175/JAS-D-16-0165.1, 2017.Garratt, J.: Review: the atmospheric boundary layer, Earth-Sci. Rev.,
37, 89–134, 10.1016/0012-8252(94)90026-4, 1994.Grosvenor, D. P., Field, P. R., Hill, A. A., and Shipway, B. J.: The relative importance of macrophysical and cloud albedo
changes for aerosol-induced radiative effects in closed-cell stratocumulus: insight from the modelling of a case study,
Atmos. Chem. Phys., 17, 5155–5183, 10.5194/acp-17-5155-2017, 2017.
Halliwell, C.: Unified Model Documentation Paper 028 Subgrid Turbulence
Scheme, Unified Model Documentation Paper 28, 2014.Harrington, J. Y., Reisin, T., Cotton, W. R., and Kreidenweis, S. M.: Cloud
resolving simulations of Arctic stratus part II: Transition-season clouds,
Atmos. Res., 51, 45–75, 10.1016/S0169-8095(98)00098-2, 1999.Heintzenberg, J. and Leck, C.: The summer aerosol in the central Arctic 1991–2008: did it change or not?,
Atmos. Chem. Phys., 12, 3969–3983, 10.5194/acp-12-3969-2012, 2012.
Herzog, H.-J., Schubert, U., Vogel, G., Fiedler, A., and Kirchner, R.: LLM
–
the high-resolving nonhydrostatic simulation model in the DWD – project
LITFASS. Part I modelling technique and simulation method, COSMO Technical
Report No. 4, 1–68, Deutscher Wetterdienst, Offenbach, Germany,
2002a.Herzog, H. J., Vogel, G., and Schubert, U.: LLM - a nonhydrostatic model
applied to high-resolving simulations of turbulent fluxes over heterogeneous
terrain, Theor. Appl. Climatol., 73, 67–86,
10.1007/s00704-002-0694-4, 2002b.Hines, K. M. and Bromwich, D. H.: Simulation of Late Summer Arctic Clouds
during ASCOS with Polar WRF, Mon. Weather Rev., 145, 521–541,
10.1175/MWR-D-16-0079.1, 2017.Hines, K. M., Bromwich, D. H., Bai, L., Bitz, C. M., Powers, J. G., and
Manning, K. W.: Sea Ice Enhancements to Polar WRF, Mon. Weather Rev.,
143, 2363–2385, 10.1175/MWR-D-14-00344.1, 2015.Igel, A. L., Ekman, A. M., Leck, C., Tjernström, M., Savre, J., and
Sedlar, J.: The free troposphere as a potential source of arctic boundary
layer aerosol particles, Geophys. Res. Lett., 44, 7053–7060,
10.1002/2017GL073808, 2017.Intrieri, J. M., Fairall, C. W., Shupe, M. D., Persson, P. O. G., Andreas,
E. L., Guest, P. S., and Moritz, R. E.: An annual cycle of Arctic surface
cloud forcing at SHEBA, J. Geophys. Res., 107, 8039,
10.1029/2000JC000439, 2002.Jiang, H., Cotton, W. R., Pinto, J. O., Curry, J. A., and Weissbluth, M. J.:
Cloud Resolving Simulations of Mixed-Phase Arctic Stratus Observed during
BASE: Sensitivity to Concentration of Ice Crystals and Large-Scale Heat and
Moisture Advection, J. Atmos. Sci., 57, 2105–2117,
10.1175/1520-0469(2000)057<2105:CRSOMP>2.0.CO;2, 2000.Kapsch, M. L., Graversen, R. G., Tjernström, M., and Bintanja, R.: The
effect of downwelling longwave and shortwave radiation on Arctic summer sea
ice, J. Climate, 29, 1143–1159, 10.1175/JCLI-D-15-0238.1,
2016.Kay, J. E. and Gettelman, A.: Cloud influence on and response to seasonal
Arctic sea ice loss, J. Geophys. Res.-Atmos., 114,
1–18, 10.1029/2009JD011773, 2009.Khvorostyanov, V. I. and Curry, J. A.: Aerosol size spectra and CCN activity
spectra: Reconciling the lognormal, algebraic, and power laws, J. Geophys. Res.-Atmos., 111, 1–15, 10.1029/2005JD006532,
2006.Klein, S. A., McCoy, R. B., Morrison, H., Ackerman, A. S., Avramov, A.,
de Boer, G., Chen, M., Cole, J. N. S., Del Genio, A. D., Falk, M., Foster,
M. J., Fridlind, A., Golaz, J.-C., Hashino, T., Harrington, J. Y., Hoose, C.,
Khairoutdinov, M. F., Larson, V. E., Liu, X., Luo, Y., McFarquhar, G. M.,
Menon, S., Neggers, R. A. J., Park, S., Poellot, M. R., Schmidt, J. M.,
Sednev, I., Shipway, B. J., Shupe, M. D., Spangenberg, D. A., Sud, Y. C.,
Turner, D. D., Veron, D. E., von Salzen, K., Walker, G. K., Wang, Z., Wolf,
A. B., Xie, S., Xu, K.-M., Yang, F., and Zhang, G.: Intercomparison of model
simulations of mixed-phase clouds observed during the ARM Mixed-Phase Arctic
Cloud Experiment. I: single-layer cloud, Q. J. Roy. Meteor. Soc., 135, 979–1002, 10.1002/qj.416, 2009.Klingebiel, M., de Lozar, A., Molleker, S., Weigel, R., Roth, A., Schmidt, L., Meyer, J., Ehrlich, A., Neuber, R., Wendisch, M.,
and Borrmann, S.: Arctic low-level boundary layer clouds: in situ measurements and simulations of mono- and bimodal supercooled
droplet size distributions at the top layer of liquid phase clouds, Atmos. Chem. Phys., 15, 617–631, 10.5194/acp-15-617-2015, 2015.Kokkola, H., Korhonen, H., Lehtinen, K. E. J., Makkonen, R., Asmi, A., Järvenoja, S., Anttila, T., Partanen, A.-I., Kulmala, M., Järvinen, H.,
Laaksonen, A., and Kerminen, V.-M.: SALSA – a Sectional Aerosol module for Large Scale Applications, Atmos. Chem. Phys., 8, 2469–2483,
10.5194/acp-8-2469-2008, 2008.Kupiszewski, P., Leck, C., Tjernström, M., Sjogren, S., Sedlar, J., Graus, M., Müller, M., Brooks, B., Swietlicki, E., Norris, S.,
and Hansel, A.: Vertical profiling of aerosol particles and trace gases over the central Arctic Ocean during summer, Atmos. Chem. Phys.,
13, 12405–12431, 10.5194/acp-13-12405-2013, 2013.Lannerfors, H., Heintzenberg, J., and Hansson, H. C.: A comprehensive study of
physical and chemical parameters of the Arctic summer aerosol; results from
the Swedish expedition Ymer-80, Tellus B, 35, 40–54,
10.1111/j.1600-0889.1983.tb00006.x, 1983.Leaitch, W. R., Korolev, A., Aliabadi, A. A., Burkart, J., Willis, M. D., Abbatt, J. P. D., Bozem, H., Hoor, P., Köllner, F., Schneider, J.,
Herber, A., Konrad, C., and Brauner, R.: Effects of 20–100 nm particles on liquid clouds in the clean summertime Arctic, Atmos. Chem. Phys., 16,
11107–11124, 10.5194/acp-16-11107-2016, 2016.Leck, C. and Svensson, E.: Importance of aerosol composition and mixing state for cloud droplet activation over the Arctic pack
ice in summer, Atmos. Chem. Phys., 15, 2545–2568, 10.5194/acp-15-2545-2015, 2015.Leck, C., Bigg, E. K., Covert, D. S., Heintzenberg, J., Maenhaut, W., Nilsson,
E. D., and Wiedensohler, A.: Overview of the atmospheric research program
during the International Arctic Ocean Expedition of 1991 (IAOE-91) and its
scientific results, Tellus B, 48,
136–155, 10.3402/tellusb.v48i2.15833, 1996.Leck, C., Nilsson, E. D., Bigg, E. K., and Bäcklin, L.: Atmospheric
program on the Arctic Ocean Expedition 1996 (AOE-96): An overview of
scientific goals, experimental approach, and instruments, J. Geophys. Res.-Atmos., 106, 32051–32067,
10.1029/2000JD900461, 2001.Leck, C., Norman, M., Bigg, E. K., and Hillamo, R.: Chemical composition and
sources of the high Arctic aerosol relevant for cloud formation, J. Geophys. Res., 107, 4135, 10.1029/2001JD001463, 2002.Leck, C., Tjernström, M., Matrai, P., Swietlicki, E., and Bigg, K.: Can
marine micro-organisms influence melting of the Arctic pack ice?, EOS T. Am. Geophys. Un., 85, 25–36, 10.1029/2004EO030001,
2004.Lilly, D. K.: A proposed modification of the Germano subgrid-scale closure
method, Phys. Fluids A-Fluid, 4, 633–635,
10.1063/1.858280, 1992.Lloyd, G., Choularton, T. W., Bower, K. N., Crosier, J., Jones, H., Dorsey, J. R., Gallagher, M. W., Connolly, P., Kirchgaessner, A. C. R.,
and Lachlan-Cope, T.: Observations and comparisons of cloud microphysical properties in spring and summertime Arctic stratocumulus clouds
during the ACCACIA campaign, Atmos. Chem. Phys., 15, 3719–3737, 10.5194/acp-15-3719-2015, 2015.Lock, A. P., Brown, A. R., Bush, M. R., Martin, G. M., and Smith, R. N. B.: A
New Boundary Layer Mixing Scheme. Part I: Scheme Description and
Single-Column Model Tests, Mon. Weather Rev., 128, 3187–3199,
10.1175/1520-0493(2000)128<3187:ANBLMS>2.0.CO;2, 2000.
Lock, A. P., Edwards, J. M., and Boutle, I. A.: Unified Model Documentation
Paper 024 The Parametrization of Boundary Layer Processes, Met Office, 2015.Loewe, K., Ekman, A. M. L., Paukert, M., Sedlar, J., Tjernström, M., and Hoose, C.: Modelling micro- and macrophysical contributors to the
dissipation of an Arctic mixed-phase cloud during the Arctic Summer Cloud Ocean Study (ASCOS), Atmos. Chem. Phys.,
17, 6693–6704, 10.5194/acp-17-6693-2017, 2017.
Manners, J., Hill, P. G., and Ex, D.: Unified Model Documentation Paper 023
The Radiation Code, Met Office, 2016.Martin, G. M., Johnson, D. W., and Spice, A.: The Measurement and
Parameterization of Effective Radius of Droplets in Warm Stratocumulus
Clouds, J. Atmos. Sci., 51, 1823–1842,
10.1175/1520-0469(1994)051<1823:TMAPOE>2.0.CO;2, 1994.Martin, M., Chang, R. Y.-W., Sierau, B., Sjogren, S., Swietlicki, E., Abbatt, J. P. D., Leck, C., and Lohmann, U.: Cloud condensation
nuclei closure study on summer arctic aerosol, Atmos. Chem. Phys., 11, 11335–11350, 10.5194/acp-11-11335-2011, 2011.Mauritsen, T., Sedlar, J., Tjernström, M., Leck, C., Martin, M., Shupe, M., Sjogren, S., Sierau, B., Persson, P. O. G.,
Brooks, I. M., and Swietlicki, E.: An Arctic CCN-limited cloud-aerosol regime, Atmos. Chem. Phys., 11, 165–173, 10.5194/acp-11-165-2011, 2011.McFarquhar, G. M., Ghan, S., Verlinde, J., Korolev, A., Strapp, J. W., Schmid,
B., Tomlinson, J. M., Wolde, M., Brooks, S. D., Cziczo, D., Dubey, M. K.,
Fan, J., Flynn, C., Gultepe, I., Hubbe, J., Gilles, M. K., Laskin, A.,
Lawson, P., Leaitch, W. R., Liu, P., Liu, X., Lubin, D., Mazzoleni, C.,
MacDonald, A. M., Moffet, R. C., Morrison, H., Ovchinnikov, M., Shupe, M. D.,
Turner, D. D., Xie, S., Zelenyuk, A., Bae, K., Freer, M., and Glen, A.:
Indirect and semi-direct aerosol campaign: The impact of arctic aerosols on
clouds, B. Am. Meteorol. Soc., 92, 183–201,
10.1175/2010BAMS2935.1, 2011.Mellor, G. L. and Yamada, T.: A Hierarchy of Turbulence Closure Models for
Planetary Boundary Layers, J. Atmos. Sci., 31,
1791–1806, 10.1175/1520-0469(1974)031<1791:AHOTCM>2.0.CO;2, 1974.Miltenberger, A. K., Field, P. R., Hill, A. A., Rosenberg, P., Shipway, B. J., Wilkinson, J. M., Scovell, R., and Blyth, A. M.:
Aerosol-cloud interactions in mixed-phase convective clouds – Part 1: Aerosol perturbations, Atmos. Chem. Phys., 18, 3119–3145, 10.5194/acp-18-3119-2018, 2018.Morrison, H., Shupe, M. D., and Curry, J. A.: Modeling clouds observed at
SHEBA using a bulk microphysics parameterization implemented into a
single-column model, J. Geophys. Res.-Atmos., 108,
4255, 10.1029/2002JD002229, 2003.Morrison, H., Curry, J. A., and Khvorostyanov, V. I.: A New Double-Moment
Microphysics Parameterization for Application in Cloud and Climate Models.
Part I: Description, J. Atmos. Sci., 62, 1665–1677,
10.1175/JAS3446.1, 2005a.Morrison, H., Shupe, M. D., Pinto, J. O., and Curry, J. A.: Possible roles of
ice nucleation mode and ice nuclei depletion in the extended lifetime of
Arctic mixed-phase clouds, Geophys. Res. Lett., 32, 1–5,
10.1029/2005GL023614, 2005b.Morrison, H., Zuidema, P., Ackerman, A. S., Avramov, A., De Boer, G., Fan,
J., Fridlind, A. M., Hashino, T., Harrington, J. Y., Luo, Y., Ovchinnikov,
M., and Shipway, B.: Intercomparison of cloud model simulations of Arctic
mixed-phase boundary layer clouds observed during SHEBA/FIRE-ACE, J. Adv. Model. Earth Sy., 3, 1–23, 10.1029/2011MS000066,
2011.Nakanishi, M. and Niino, H.: An improved Mellor-Yamada Level-3 model: Its
numerical stability and application to a regional prediction of advection
fog, Bound.-Lay. Meteorol., 119, 397–407,
10.1007/s10546-005-9030-8, 2006.Nenes, A. and Seinfeld, J. H.: Parameterization of cloud droplet formation in
global climate models, J. Geophys. Res., 108, 4415,
10.1029/2002JD002911, 2003.Ovchinnikov, M., Korolev, A., and Fan, J.: Effects of ice number concentration
on dynamics of a shallow mixed-phase stratiform cloud, J. Geophys. Res.-Atmos., 116, 1–15, 10.1029/2011JD015888,
2011.Ovchinnikov, M., Ackerman, A. S., Avramov, A., Cheng, A., Fan, J., Fridlind,
A. M., Ghan, S., Harrington, J., Hoose, C., Korolev, A., McFarquhar, G. M.,
Morrison, H., Paukert, M., Savre, J., Shipway, B. J., Shupe, M. D., Solomon,
A., and Sulia, K.: Intercomparison of large-eddy simulations of Arctic
mixed-phase clouds: Importance of ice size distribution assumptions, J. Adv. Model. Earth Sy., 6, 223–248,
10.1002/2013MS000282, 2014.
Paatero, J., Vaattovaara, P., Vestenius, M., Meinander, O., Makkonen, U., Kivi,
R., Hyvärinen, A., Asmi, E., Tjernström, M., and Leck, C.:
Finnish contribution to the Arctic summer cloud ocean study (ASCOS)
expedition, Arctic ocean 2008, Geophysica, 45, 119–146, 2009.Paukert, M. and Hoose, C.: Modeling immersion freezing with aerosol-dependent
prognostic ice nuclei in Arctic mixed-phase clouds, J. Geophys. Res., 119, 9073–9092, 10.1002/2014JD021917, 2014.Peters, G. P., Nilssen, T. B., Lindholt, L., Eide, M. S., Glomsrød, S., Eide, L. I., and Fuglestvedt, J. S.: Future emissions from
shipping and petroleum activities in the Arctic, Atmos. Chem. Phys., 11, 5305–5320, 10.5194/acp-11-5305-2011, 2011.Petters, M. D. and Kreidenweis, S. M.: A single parameter representation of hygroscopic growth and cloud condensation nucleus activity,
Atmos. Chem. Phys., 7, 1961–1971, 10.5194/acp-7-1961-2007, 2007.Pinto, J. O.: Autumnal mixed-phase cloudy boundary layers in the Arctic, J.
Atmos. Sci., 55, 2016–2038,
10.1175/1520-0469(1998)055<2016:AMPCBL>2.0.CO;2, 1998.Possner, A., Zubler, E., Fuhrer, O., Lohmann, U., and Schär, C.: A Case
Study in Modeling Low-Lying Inversions and Stratocumulus Cloud Cover in the
Bay of Biscay, Weather Forecast., 29, 289–304,
10.1175/WAF-D-13-00039.1, 2014.Possner, A., Ekman, A. M. L., and Lohmann, U.: Cloud response and feedback
processes in stratiform mixed-phase clouds perturbed by ship exhaust,
Geophys. Res. Lett., 44, 1964–1972, 10.1002/2016GL071358,
2017.Prenni, A. J., Harrington, J. Y., Tjernstöm, M., DeMott, P. J., Avramov,
A., Long, C. N., Kreidenweis, S. M., Olsson, P. Q., and Verlinde, J.: Can
ice-nucleating aerosols affect arctic seasonal climate?, B. Am. Meteorol. Soc., 88, 541–550, 10.1175/BAMS-88-4-541,
2007.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, 10.1175/1520-0493(1992)120<0303:ACRSFN>2.0.CO;2, 1992.Sand, M., Samset, B. H., Balkanski, Y., Bauer, S., Bellouin, N., Berntsen, T. K., Bian, H., Chin, M., Diehl, T., Easter, R., Ghan, S. J.,
Iversen, T., Kirkevåg, A., Lamarque, J.-F., Lin, G., Liu, X., Luo, G., Myhre, G., Noije, T. V., Penner, J. E., Schulz, M., Seland, Ø.,
Skeie, R. B., Stier, P., Takemura, T., Tsigaridis, K., Yu, F., Zhang, K., and Zhang, H.: Aerosols at the poles: an AeroCom Phase II multi-model
evaluation, Atmos. Chem. Phys., 17, 12197–12218, 10.5194/acp-17-12197-2017, 2017.Savre, J., Ekman, A. M. L., and Svensson, G.: Technical note: Introduction to
MIMICA, a large-eddy simulation solver for cloudy planetary boundary layers,
J. Adv. Model. Earth Sy., 6, 1–20,
10.1002/2013MS000292, 2014.Sedlar, J., Tjernström, M., Mauritsen, T., Shupe, M. D., Brooks, I. M.,
Persson, P. O. G., Birch, C. E., Leck, C., Sirevaag, A., and Nicolaus, M.: A
transitioning Arctic surface energy budget: The impacts of solar zenith
angle, surface albedo and cloud radiative forcing, Clim. Dynam., 37,
1643–1660, 10.1007/s00382-010-0937-5, 2011.Seifert, A. and Beheng, K. D.: A two-moment cloud microphysics
parameterization for mixed-phase clouds. Part 1: Model description,
Meteorol. Atmos. Phys., 92, 45–66,
10.1007/s00703-005-0112-4, 2006.Seifert, A., Nuijens, L., and Stevens, B.: Turbulence effects on warm-rain
autoconversion in precipitating shallow convection, Q. J. Roy. Meteor. Soc., 136, 1753–1762, 10.1002/qj.684, 2010.Serreze, M. C. and Stroeve, J. C.: Arctic sea ice trends, variability and
implications for seasonal ice forecasting, Philos. T.
Roy. Soc. A, 373,
20140159, 10.1098/rsta.2014.0159, 2015.Shindell, D. and Faluvegi, G.: Climate response to regional radiative forcing
during the twentieth century, Nat. Geosci., 2, 294–300,
10.1038/ngeo473, 2009.Shupe, M. D., Persson, P. O. G., Brooks, I. M., Tjernström, M., Sedlar, J., Mauritsen, T., Sjogren, S., and Leck, C.: Cloud and boundary
layer interactions over the Arctic sea ice in late summer, Atmos. Chem. Phys., 13, 9379–9399, 10.5194/acp-13-9379-2013, 2013.Shupe, M. D., Turner, D. D., Zwink, A., Thieman, M. M., Mlawer, E. J., and
Shippert, T.: Deriving arctic cloud microphysics at Barrow, Alaska:
Algorithms, results, and radiative closure, J. Appl. Meteorol. Clim., 54, 1675–1689, 10.1175/JAMC-D-15-0054.1, 2015.Solomon, A., Morrison, H., Persson, O., Shupe, M. D., and Bao, J.-W.:
Investigation of Microphysical Parameterizations of Snow and Ice in Arctic
Clouds during M-PACE through Model-Observation Comparisons, Mon. Weather Rev., 137, 3110–3128, 10.1175/2009MWR2688.1, 2009.Solomon, A., Feingold, G., and Shupe, M. D.: The role of ice nuclei recycling in the maintenance of cloud ice in Arctic
mixed-phase stratocumulus, Atmos. Chem. Phys., 15, 10631–10643, 10.5194/acp-15-10631-2015, 2015.Sotiropoulou, G., Sedlar, J., Forbes, R., and Tjernström, M.: Summer
Arctic clouds in the ECMWF forecast model: an evaluation of cloud
parameterization schemes, Q. J. Roy. Meteor.
Soc., 142, 387–400, 10.1002/qj.2658, 2015.Steppeler, J., Doms, G., Schättler, U., Bitzer, H. W., Gassmann, A., U.,
D., and Gregoric, G.: Meso-gamma scale forecasts using the nonhydrostatic
model LM, Meteorol. Atmos. Phys., 82, 75–96,
10.1007/s00703-001-0592-9, 2003.Stevens, B., Moeng, C., and Sullivan, P.: Large-eddy simulations of
radiatively driven convection: Sensitivities to the representation of small
scales, J. Atmos. Sci., 56, 3963–3984,
10.1175/1520-0469(1999)056<3963:LESORD>2.0.CO;2, 1999.Stevens, B., Moeng, C.-H., Ackerman, A. S., Bretherton, C. S., Chlond, A.,
de Roode, S., Edwards, J., Golaz, J.-C., Jiang, H., Khairoutdinov, M.,
Kirkpatrick, M. P., Lewellen, D. C., Lock, A., Müller, F., Stevens,
D. E., Whelan, E., and Zhu, P.: Evaluation of Large-Eddy Simulations via
Observations of Nocturnal Marine Stratocumulus, Mon. Weather Rev., 133,
1443–1462, 10.1175/MWR2930.1, 2005.Stevens, R. G., Loewe, K., Dearden, C., Dimitrelos, A., Possner, A., Eirund, G. K., Raatikainen, T., Hill,
A. A., Shipway, B. J., Wilkinson, J., Romakkaniemi, S., Tonttila, J., Laaksonen, A., Korhonen, H., Connolly, P.,
Lohmann, U., Hoose, C., Ekman, A. M. L., Carslaw, K. S., and Field, P. R.:
Model output from “A model intercomparison of CCN-limited tenuous clouds in the high Arctic” [Data set],
Zenodo,
10.5281/zenodo.1326922, 2018.
Stratton, R. A., Willett, M. R., Derbyshire, S., Wong, R., and Whithall, M.:
Unified Model Documentation Paper 027 Convection Schemes, Met Office, 2015.Struthers, H., Ekman, A. M. L., Glantz, P., Iversen, T., Kirkevåg, A., Mårtensson, E. M., Seland, Ø., and Nilsson, E. D.: The effect of
sea ice loss on sea salt aerosol concentrations and the radiative balance in the Arctic, Atmos. Chem. Phys., 11, 3459–3477, 10.5194/acp-11-3459-2011, 2011.Tjernström, M., Leck, C., Persson, P. O. G., Jensen, M. L., Oncley,
S. P., and Targino, A.: The summertime arctic atmosphere: Meteorological
measurements during the Arctic Ocean Experiment 2001, B. Am. Meteorol. Soc., 85, 1305–1321,
10.1175/BAMS-85-9-1305, 2004.Tjernström, M., Birch, C. E., Brooks, I. M., Shupe, M. D., Persson, P. O. G., Sedlar, J., Mauritsen, T., Leck, C., Paatero, J.,
Szczodrak, M., and Wheeler, C. R.: Meteorological conditions in the central Arctic summer during the Arctic Summer Cloud
Ocean Study (ASCOS), Atmos. Chem. Phys., 12, 6863–6889, 10.5194/acp-12-6863-2012, 2012.Tjernström, M., Leck, C., Birch, C. E., Bottenheim, J. W., Brooks, B. J., Brooks, I. M., Bäcklin, L., Chang, R. Y.-W., de Leeuw, G., Di Liberto, L.,
de la Rosa, S., Granath, E., Graus, M., Hansel, A., Heintzenberg, J., Held, A., Hind, A., Johnston, P., Knulst, J., Martin, M., Matrai, P. A.,
Mauritsen, T., Müller, M., Norris, S. J., Orellana, M. V., Orsini, D. A., Paatero, J., Persson, P. O. G., Gao, Q., Rauschenberg, C., Ristovski, Z.,
Sedlar, J., Shupe, M. D., Sierau, B., Sirevaag, A., Sjogren, S., Stetzer, O., Swietlicki, E., Szczodrak, M., Vaattovaara, P., Wahlberg, N., Westberg, M.,
and Wheeler, C. R.: The Arctic Summer Cloud Ocean Study (ASCOS): overview and experimental design, Atmos. Chem. Phys., 14, 2823–2869, 10.5194/acp-14-2823-2014, 2014.Tjernström, M., Shupe, M. D., Brooks, I. M., Persson, P. O. G.,
Prytherch, J., Salisbury, D. J., Sedlar, J., Achtert, P., Brooks, B. J.,
Johnston, P. E., Sotiropoulou, G., and Wolfe, D.: Warm-air advection, air
mass transformation and fog causes rapid ice melt, Geophys. Res. Lett., 42, 5594–5602, 10.1002/2015GL064373, 2015.Tonttila, J., Maalick, Z., Raatikainen, T., Kokkola, H., Kühn, T., and Romakkaniemi, S.: UCLALES-SALSA v1.0: a large-eddy model with
interactive sectional microphysics for aerosol, clouds and precipitation, Geosci. Model Dev., 10, 169–188, 10.5194/gmd-10-169-2017, 2017.Uttal, T., Curry, J. A., McPhee, M. G., Perovich, D. K., Moritz, R. E.,
Maslanik, J. A., Guest, P. S., Stern, H. L., Moore, J. A., Turenne, R.,
Heiberg, A., Serreze, M. C., Wylie, D. P., Persson, O. G., Paulson, C. A.,
Halle, C., Marison, J. H., Wheeler, P. A., Makshtas, A., Welch, H., Shupe,
M. D., Intrieri, J. M., Stamnes, K., Lindsey, R. W., Pinkel, R., Pegau,
W. S., Stanton, T. P., and Grenfeld, T. C.: Surface heat budget of the
arctic ocean, B. Am. Meteorol. Soc., 83, 255–275,
10.1175/1520-0477(2002)083<0255:SHBOTA>2.3.CO;2, 2002.Vaughan, D., Comiso, J., Allison, I., Carrasco, J., Kaser, G., Kwok, R., Mote,
P., Murray, T., Paul, F., Ren, J., Rignot, E., Solomina, O., Steffen, K., and
Zhang, T.: Observations: Cryosphere, in: Climate Change 2013: The Physical
Science Basis. Contribution of Working Group I to the Fifth Assessment Report
of the Intergovernmental Panel on Climate Change, edited by: Stocker, T., Qin,
D., Plattner, G.-K., Tignor, M., Allen, S., Boschung, J., Nauels, A., Xia,
Y., Bex, V., and Midgley, P., Cambridge University Press,
Cambridge, United Kingdom and New York, NY, USA, 317–382,
10.1017/CBO9781107415324.012, 2013.
Verlinde, J., Harrington, J. Y., McFarquhar, G. M., Yannuzzi, V. T., Avramov,
A., Greenberg, S., Johnson, N., Zhang, G., Poellot, M. R., Mather, J. H.,
Turner, D. D., Eloranta, E. W., Zak, B. D., Prenni, A. J., Daniel, J. S.,
Kok, G. L., Tobin, D. C., Holz, R., Sassen, K., Spangenberg, D., Minnis, P.,
Tooman, T. P., Ivey, M. D., Richardson, S. J., Bahrmann, C. P., Shupe, M.,
DeMott, P. J., Heymsfield, A. J., and Schofield, R.: The mixed-phase arctic
cloud experiment, B. Am. Meteorol. Soc., 88,
205–221, 10.1175/BAMS-88-2-205, 2007.Vogel, B., Vogel, H., Bäumer, D., Bangert, M., Lundgren, K., Rinke, R., and Stanelle, T.: The comprehensive model system COSMO-ART – Radiative
impact of aerosol on the state of the atmosphere on the regional scale, Atmos. Chem. Phys., 9, 8661–8680, 10.5194/acp-9-8661-2009, 2009.
Wang, C. and Chang, J. S.: A Three-Dimensional Numerical Model of Cloud
Dynamics, Microphysics, and Chemistry 1. Concepts and Formulation, J. Geophys. Res., 98, 14827–14844, 1993.Wesslén, C., Tjernström, M., Bromwich, D. H., de Boer, G., Ekman, A. M. L., Bai, L.-S., and Wang, S.-H.: The Arctic summer atmosphere:
an evaluation of reanalyses using ASCOS data, Atmos. Chem. Phys., 14, 2605–2624, 10.5194/acp-14-2605-2014, 2014.Westwater, E. R., Han, Y., Shupe, M. D., and Matrosov, S. Y.: Analysis of
integrated cloud liquid and precipitable water vapor retrievals from
microwave radiometers during the Surface Heat Budget of the Arctic Ocean
project, J. Geophys. Res.-Atmos., 106,
32019–32030, 10.1029/2000JD000055, 2001.Young, G., Jones, H. M., Choularton, T. W., Crosier, J., Bower, K. N., Gallagher, M. W., Davies, R. S., Renfrew, I. A., Elvidge, A. D., Darbyshire, E.,
Marenco, F., Brown, P. R. A., Ricketts, H. M. A., Connolly, P. J., Lloyd, G., Williams, P. I., Allan, J. D., Taylor, J. W., Liu, D., and Flynn, M. J.:
Observed microphysical changes in Arctic mixed-phase clouds when transitioning from sea ice to open ocean, Atmos. Chem. Phys., 16, 13945–13967, 10.5194/acp-16-13945-2016, 2016.Young, G., Connolly, P. J., Jones, H. M., and Choularton, T. W.: Microphysical sensitivity of coupled springtime Arctic stratocumulus to
modelled primary ice over the ice pack, marginal ice, and ocean, Atmos. Chem. Phys., 17, 4209–4227, 10.5194/acp-17-4209-2017, 2017.Young, G., Connolly, P. J., Dearden, C., and Choularton, T. W.: Relating large-scale subsidence to convection development in Arctic mixed-phase
marine stratocumulus, Atmos. Chem. Phys., 18, 1475–1494, 10.5194/acp-18-1475-2018, 2018.