Introduction
Deep convection has a significant influence on the state of the
atmosphere and climate through shortwave and longwave radiative
interactions, heat transfer through the release of latent heat and
global heat redistribution. It also plays an important part in the
hydrological cycle through the conversion of water vapour to
precipitation. One major way that aerosols can influence the
properties of deep convection is through their effect on cloud
microphysics. By acting as cloud condensation nuclei (CCN),
increased aerosol loading can lead to an increase in cloud droplet
number concentration (CDNC) and a subsequent reduction in cloud
droplet size, which in turn has been hypothesised to suppress
warm-phase precipitation . Some theoretical
e.g. and cloud-resolving (or
cloud-system-resolving) modelling studies e.g.amongst
many others have suggested that
under certain conditions, precipitation suppression in the liquid
phase may lead to an invigoration of deep convection and
a subsequent enhancement of convective precipitation. The detection
of positive correlations between satellite-observed aerosol
optical depth (AOD) and precipitation or convective cloud
properties e.g. might suggest
observational evidence of convective invigoration by
aerosols. However, factors such as meteorological covariation and
retrieval errors may contribute to or even dominate such
correlations . Complex process interactions in ice- and
mixed-phase microphysics, along with coupling to surface and
radiative feedbacks and dynamics over a range of spatiotemporal
scales, means that understanding and quantifying aerosol impacts
on deep convection remains a significant challenge
e.g..
Representing cloud microphysical processes, which occur on length
scales of microns to millimetres, has always been a significant
challenge for atmospheric models. Even in cloud-resolving models,
horizontal grid lengths tend to be on the order of kilometres to
a few hundred metres at best, so it is impossible for such
models to explicitly simulate microphysical processes. There is
a long history of microphysical parameterisation seefor
a comprehensive review, and microphysics schemes today
tend to fall into one of two categories: bin models, in which the
size distribution of each hydrometeor class is explicitly
calculated e.g.,
and bulk models, in which a size distribution function is typically
used to represent each hydrometeor class and one (or several)
moments of the size distribution function are calculated
explicitly e.g.amongst many others. Bulk
models are therefore very computationally efficient compared to
bin models often by at least 2 orders of
magnitude; and are used as standard in many
atmospheric modelling systems today. Although certain aspects of
cloud processes and aerosol indirect effects cannot be reproduced
well in bulk schemes seefor a detailed
analysis, there nevertheless remains a trade-off
between how completely the hydrometeor size spectra are
represented and the physical domain size that can then be used in
a simulation. For most applications, full bin microphysics (which
can even resolve the autoconversion process of cloud water to
rain) are only feasible using small domains and idealised
simulations, which then cannot represent the dynamical feedbacks
that can occur on larger domains a notable exception,
proving the cost of such simulations, are the multiple month-long
case study simulations using bin microphysics presented
by. Thus, studies using bulk and bin microphysics
representations provide differently imperfect and thus
complementary information. Indeed, bulk schemes remain as standard
in global models, and successful studies of aerosol indirect
effects in global models have been performed using bulk
microphysics e.g..
Whilst early bulk microphysics schemes were single moment only
(predicting only the k=1 moment of the particle size distribution
equation, mass), a significant development has been predicting two
moments of the size distribution equation (k=0, number
concentration, and k=1, mass) e.g., which has been shown
to have improved results compared to single-moment schemes
e.g.. Indeed, although not widely used at present,
three-moment schemes have been shown to further improve
representations of large hail
and precipitation reflectivities .
However, bulk schemes make a priori assumptions about the shape of
the particle size distributions (usually approximated by
exponential or gamma distributions and more rarely by lognormal
functions), whereas bin schemes calculate particle size
distributions by solving explicit microphysical equations and make
no a priori assumption about the particle size distribution
shapes. This can lead to significant differences in the cloud and
precipitation simulated by bin vs. bulk schemes. For example, bulk
schemes have been shown to underestimate areas of weak and
stratiform rain in an MCS compared to a bin scheme which performed
better against observations .
showed that a one-moment bulk scheme was
shown to be worse at partitioning rain into stratiform and
convective components in a continental squall line compared to
a bin scheme (although many studies have shown that two-moment
schemes are a significant improvement on single-moment schemes;
e.g.). found that, while all schemes
overestimated maximum rain rates in a simulated MCS, all bulk
schemes tested overpredicted average and maximum rain rates by
a factor of 2 to 3, while bin schemes overestimated maximum rain
rates by about 20 %. In idealised supercell simulations,
found that the
double-moment bulk scheme produced 2 times more accumulated
surface rain than a bin scheme, while found that
the bulk scheme also produced twice as much
surface rain as the same bin scheme used by in
simulations of the same supercell. Investigations of the shape of
the cloud droplet size distribution in large-eddy simulations of
non-precipitating shallow cumulus clouds with a bin
and bulk scheme showed the
importance of the cloud droplet size distribution shape
parameter. In the bulk scheme, evaporation rates were much more
sensitive to the value of the shape parameter than to the
condensation rates, and thus the shape parameter strongly impacted
cloud properties such as droplet number concentration, mean
droplet diameter and cloud fraction . Bin scheme
simulations suggested that the shape parameter should be based on
the relationship between local values of the cloud droplet
concentration and the relative width of the cloud droplet size
distribution rather than cloud mean values, as are traditionally
used . Further, showed that
despite other fundamental differences between the bin and bulk
condensation parameterisations, differences in condensation rates
could be predominantly explained by accounting for the width of
the cloud droplet size distributions simulated by the bin scheme.
found that the most important factor in
achieving agreement in concentrations and mass contents between
bulk and bin schemes in simulations of continental and tropical
maritime clouds was accurate representation of warm-phase
autoconversion. Sensitivity tests of four different autoconversion
parameterisations conducted by
and showed that errors in predicting cloud water content
in bulk schemes could be attributed to the saturation adjustment
used in the calculation of evaporation and condensation. Likewise,
also showed, using four different types of
autoconversion scheme, that saturation adjustment was the leading
order factor in discrepancies of prediction of cloud water content
by bulk schemes. found that tropical
cyclones showed weak sensitivity to aerosol due to the use of
saturation adjustment. In the ice phase,
found artificial spikes in heating rates from deposition and
sublimation due to the saturation adjustment
scheme. found that even at very high resolution,
convective cores in an idealised squall line simulation remained
undiluted due to the saturation adjustment used in the bulk
microphysics scheme. However, saturation adjustment alone is
insufficient to explain all differences between bin and bulk
schemes: in idealised supercell simulations using bulk
microphysics both with saturation adjustment and without (in which
the scheme was modified to include an explicit representation of
supersaturation predicted over each time step),
found that the use of saturation adjustment was able to explain
differences between a bulk and bin scheme in the response of cold
pool evolution and convective dynamics under polluted conditions,
but was not sufficient to explain the large differences in the
response of surface precipitation to aerosol loading.
Differences between bin and bulk schemes can often be traced to
their different process representations. For example, some studies
have found rain evaporation in bulk schemes to be too fast
compared to bin schemes . Bulk schemes have been found to have
higher condensation and evaporation rates but similar rates of
freezing and melting compared to bin schemes . compared rates of diffusional
growth, collisions, sedimentation and surface precipitation in
several bulk schemes against results from the Tel Aviv University
bin scheme and found that precipitation peaks
in the bulk schemes were too sharp and too narrow compared to the
bin scheme, whereas the bin scheme produced weaker precipitation
covering an overall larger area than that in the bulk
schemes. tested three different
parameterisations of the coalescence process in the
bulk scheme against a bin scheme, under
different aerosol loadings and in both warm stratocumulus and warm
cumulus clouds, and found that for both the bulk and bin scheme,
each representation of the coalescence process led to different
averaged rain contents and mean raindrop
diameters. showed that, because bulk schemes do
not represent size-resolved ice particle fall speeds, they were
unable compared to bin schemes to simulate the reduced fall
velocities of ice and snow at upper levels from clean to polluted
conditions in tropical, mid-latitude coastal and mid-latitude
summertime inland continental deep convective
clouds. also suggested that bulk schemes tended to
artificially freeze large raindrops due to the use of a fixed
gamma distribution.
In some cases, tuning particular processes in bulk schemes has led
to better agreement with bin schemes, e.g. tuning evaporation
rates and fall velocities of graupel in single-moment bulk scheme
simulations of a continental squall line . Similarly, although no active tuning was performed,
found that precipitation rates and accumulated
precipitation values were in close agreement between simulations
of continental and tropical maritime clouds in high and low CCN
conditions using a bin and bulk scheme, with agreement between the
bulk and bin scheme even greater in the high CCN case compared
to the low CCN case.
Not only do bin and bulk schemes often produce different results
in terms of cloud and precipitation, but found
that the use of fixed CCN in a bulk scheme led to opposite CCN
effects on convection and heavy rain compared to CCN effects when
using a bin scheme. Similarly, found an opposite
response of accumulated surface rain to CCN in idealised supercell
simulations using a bulk and bin scheme.
found a difference in the response of an idealised supercell to
aerosol perturbations when a bin and bulk scheme was used, with
the bulk scheme producing stronger updraughts and greater average
precipitation than the bin scheme and with the left-moving storm
prevailing in the bulk simulation, while the right-moving storm
prevailed in the bin simulation. The differences were attributed
to differences in the vertical velocities in the bin vs. bulk
schemes, which led to hydrometeors ascending to different
altitudes with different directions of background flow.
Nevertheless, bulk schemes have shown sensitivity to aerosol. In
simulations of tropical deep convection,
found an ice-phase response to aerosol in which cloud top heights
and anvil ice mixing ratios increase under polluted conditions
due to increased freezing of larger numbers of cloud droplets and
subsequent higher ice particle concentrations with smaller sizes
and reduced fall speeds. Indeed, a similar mechanism was later
confirmed in bin scheme simulations by , who
performed month-long simulations of deep convection over the
tropical western Pacific, southeastern China and the US southern
Great Plains. Further, found that
autoconversion of cloud water to rain decreased under polluted
conditions, and subsequently near-surface rain and hail particles
increased in size due to enhanced collection of cloud droplets. In
simulations of deep convection over Florida using a bin-emulating
bulk scheme, found that updraught
strengths increased and anvil areas became smaller but better
organised and with increased condensate mixing ratios. Similarly,
in simulations of summertime convection over Germany using
a two-moment bulk scheme, found a strong
aerosol effect on cloud properties such as condensate amounts and
glaciation.
Unlike liquid cloud and rain drops (well described by spheres of
constant density), ice particles have a wide range of densities
and shapes, making the representation of ice-phase microphysics in
parameterisations much more difficult than the liquid
phase. Traditionally, the approach in both bin
e.g. and bulk schemes
e.g.etc. was to
partition ice particles into one of a fixed number of categories
(e.g. cloud ice, snow, hail and graupel) each with its own
specified density, shape distribution and physical parameters such
as fall speeds. However, such partitioning oversimplifies the
complex nature of ice-phase processes, requiring thresholds and
parameters – often chosen on a relatively ad hoc basis – to
determine the partitioning of ice particles into each category and
for converting between categories. As such, it is unsurprising
that simulations have been found to be highly sensitive to
particle fall speeds and densities e.g.,
the description of dense precipitating ice as hail or graupel
categories e.g. and
changes in thresholds or rates for converting between ice
categories e.g.. Differences in ice-phase
microphysics in bulk schemes have been shown to affect cloud
biases, especially at upper levels , and to
affect ice–cloud–radiation feedbacks with impacts on
tropospheric stability, triggering of deep convection and surface
precipitation . Such limitations have led to the
development in more recent years of new representations of ice
microphysics in bulk schemes, such as approaches which separately prognose ice mass mixing
ratios grown by riming and vapour deposition (Morrison and Grabowski, 2008), approaches where
particle habit evolution is predicted by prognosing the mixing ratios of ice crystal axes (Harrington et al., 2013)
and approaches where ice-phase particles are represented by several physical properties that evolve freely in
time and space (Morrison and Milbrandt, 2015). Although these
developments are relatively new, they have already been shown to
improve simulations of observed squall lines and orographic
precipitation when compared to traditional two-moment bulk schemes
.
Evaluations of microphysics schemes frequently involve comparison
against observations of a real precipitation event
e.g.. Often, multiple microphysics
schemes are compared against each other and against observations
e.g.. Another common approach is to evaluate a single
microphysics scheme against observations and then use different
aerosol concentrations in the model to test the sensitivity of the
observed storm to aerosol processes
e.g.. However, studies of
different convective events in different regions using different
models with different microphysics schemes often produce
conflicting results on the nature of the storm response to
aerosol. Mesoscale studies of Florida convection found that cloud
water mass, updraught strength and surface precipitation tend to
increase with increased aerosol concentration, while anvil areas
decreased but contained greater condensate mass
. Studies of summertime convective
precipitation in Germany found that increased aerosol
concentrations had a strong effect on cloud microphysical (and
therefore radiative) properties but that the combined effects of
microphysical and dynamical processes resulted in relatively
little effect on surface precipitation . This is
similar to the findings of in
idealised and continental-scale simulations.
Detailed process modelling studies of aerosol–convection
interactions often focus on the sensitivity of a single idealised
model configuration (without large-scale meteorology or surface
and radiative interactions) to perturbations using either CCN
spectra e.g. or CDNC
values e.g. as a proxy
variable to test the sensitivity of the microphysics to
aerosol. Many types of idealised models are used, ranging from
flow over a 2-D mountain e.g., to 2-D
cloud-system-resolving studies of interacting convective clouds
e.g. to 3-D simulations of idealised
supercell storms e.g.. With such a wide range of model configurations,
convective and large-scale environments, microphysics
parameterisations (bin and bulk models are both frequently used in
idealised studies of aerosol–convection interactions) and proxy
variables used to represent aerosol processes, it is perhaps not
surprising that a consistent response of idealised convection to
aerosol has not been seen; indeed, due to environment and
regime dependence, it may not exist. Idealised flow over a 2-D
mountain using CDNC values to represent aerosol amounts showed
that cloud water content increased with CDNC and drizzle content
decreased , while a similar study using an
idealised supercell configuration found that differences in the
accumulated surface precipitation and convective mass flux between
polluted and pristine values of CDNC were very small
. In studies using modified CCN spectra to
represent different levels of aerosol in a two-moment scheme, 2-D
ensemble simulations of interacting convective clouds have found
that although cloud top heights and anvil ice increase under
polluted conditions, convection actually weakens slightly compared
to pristine conditions . However, similar 3-D
simulations also using a two-moment microphysics scheme have shown
that for isolated convective cells, increased aerosol leads to
reduced total precipitation and updraught velocity; for
multicell systems it leads to increased secondary convection,
total precipitation and updraught velocities, whilst supercell
systems are relatively insensitive to aerosol
. Additionally, environmental wind shear
has been shown to have a role in determining the response of
convective systems to aerosol, with increased aerosol loading
invigorating convection under weak shear conditions and
suppressing convection under strong shear in simulations performed
with both bin and bulk
microphysics schemes.
The focus of this work is to show within a single modelling
framework that uncertainty in cloud impacts through the choice of
microphysics scheme can far exceed any aerosol effect seen within
a single scheme and that this is a consistent finding across
different types of convection in different environments and types
of simulation all of which are known to impact the effect
of aerosol loading on cloud development,
e.g.. Although we use two bulk microphysics
schemes to show this, there is a body of literature which
identifies signals of aerosol impact on cloud in bulk schemes
e.g.,
albeit not always convective invigoration see
especially, and in bin-emulating bulk schemes
e.g.. Nevertheless,
using a two-moment bulk scheme to simulate a single cumulonimbus
in an environment characterised by high CAPE and low wind shear,
found higher overshooting tops and larger
sizes with increased aerosol loading, indicating that in some
environments bulk schemes are able to produce invigoration
effects. In some cases, aerosol effects may be relatively small
less than 15 %; e.g.. However,
while some argue (fairly) that this is at least in part due to the
limitations of bulk schemes to fully represent aerosol–cloud
interactions such as saturation adjustment;,
others argue that this is consistent with the concept of clouds as
a buffered system hypothesised by . Month-long
simulations approaching the climatological scale using bin
microphysics performed by also showed aerosol
impacts on precipitation on the order of a few percent. However, those
authors showed a significant aerosol impact on rain rates rather
than total rain amount, observing a shift towards heavier rain
rates and fewer light rain rates under polluted conditions in two
regions (a tropical environment and mid-latitude coastal
environment), although the response in a mid-latitude inland
summertime continental environment varied temporally over the
simulation. Similarly to the environmental dependence found by
, showed that even in an idealised
simulation of a supercell using open boundaries and bulk
microphysics, the relative humidity and shear used in the initial
profile had an impact on the aerosol effects observed in the
simulation.
We perform high-resolution convection-permitting simulations with
the Weather Research and Forecast (WRF) model in three
configurations: a real-data simulation of deep convection in the
Congo basin, an idealised supercell case and a shallow convection
large-eddy simulation (LES). In each case we compare hydrometeor
development in two commonly used double-moment bulk schemes and
investigate the response of each model configuration to CDNC
perturbations. Our focus is not to provide a detailed process
study of aerosol effects on convection per se (to do so in the
context of multiple model configurations is beyond the scope of
this paper), but rather to explore and identify uncertainty in the
cloud and precipitation response to CDNC perturbations across
a range of model configurations. We acknowledge that, due to
a lack of fully coupled aerosol–cloud processes
e.g. supersaturation representation, droplet activation,
wet deposition and buffering processes;, the magnitude of the response of bulk
microphysics schemes to CDNC perturbations may differ from that in
schemes that explicitly treat the cloud processing of aerosol. Our
goal is therefore to highlight the large uncertainty in cloud and
precipitation responses to perturbations of CDNC in
convection-permitting models, even between multiple configurations
of the same widely used model.
List of model configurations.
Model settings
Congo
Supercell
RICO LES
Horizontal grid length (km)
4
4
0.1
Number of grid points (W–E and S–N)
525
400
129
Number of vertical levels
30
30
100
Model top
5000 Pa
20 km
4 km
Time step (s)
12
12
1
Simulation length
10 days
2 h
24 h
LW radiation scheme
RRTM
–
–
SW radiation scheme
Goddard
–
–
PBL scheme
YSU
–
–
List of microphysics configurations tested and the abbreviations used for each run.
Prescribed CDNC
Congo MORR
Congo THOM
Supercell MORR
Supercell THOM
RICO MORR
RICO THOM
100 cm-3
CONGO-M100
CONGO-T100
SUPER-M100
SUPER-T100
RICO-M100
RICO-T100
250 cm-3
CONGO-M250
CONGO-T250
SUPER-M250
SUPER-T250
RICO-M250
RICO-T250
2500 cm-3
CONGO-M2500
CONGO-T2500
SUPER-M2500
SUPER-T2500
RICO-M2500
RICO-T2500
Results
WRF Congo basin
Maps of simulated outgoing longwave radiation (OLR) and surface
precipitation at 07:00 UTC on 7 August 2007 (7 days into
the simulation) indicate that the cloud morphological and
precipitation differences for different microphysics schemes are
much greater than the cloud and precipitation response within each
scheme to different CDNC values (Fig. ). In the
CONGO-MORR simulations, low OLR values (indicating cold, high
cloud) are distributed across the domain. Precipitation at this
time occurs only in cloud north of 3∘ S, but there is
a large band of non-precipitating cold cloud across the south of
the domain. There is little discernable response of the morphology
of the OLR and precipitation in the CONGO-MORR simulations to
different CDNC values (Fig. a–c). In comparison,
cold cloud in the CONGO-THOM simulations occurs mostly north of
3∘ S (Fig. d–f). Less cloud forms in CONGO-THOM compared
to CONGO-MORR, and the cloud generally has greater OLR values than
that in CONGO-MORR. Some non-precipitating cloud occurs south of
3∘ S in the CONGO-THOM simulations, but the band is
significantly weaker and warmer than in CONGO-MORR. The
differences at this snapshot are representative of differences
that persist throughout the simulation. Frequency distributions of
OLR over the entire 10-day simulation period show that CONGO-MORR
has a much higher frequency of occurrence of colder, higher cloud
(values of about 120 Wm2) than CONGO-THOM (which
increases in frequency slightly with increased CDNC), while
CONGO-THOM has a much higher frequency of occurrence of warmer
cloud (values of about 270 Wm2) than CONGO-MORR
(Fig. a). When compared to
observations of OLR from the Geostationary Earth Radiation Budget
GERB; over the same region and period, CONGO-THOM
represents warm cloud more consistently with GERB than CONGO-MORR,
despite overpredicting colder cloud somewhat, while CONGO-MORR
overpredicts higher cloud and underpredicts warm cloud compared to
the observations
(Fig. a). However, despite
a poorer prediction of cloud radiative properties, CONGO-MORR
predicts surface precipitation better than CONGO-THOM when
compared to observations from the Tropical Rainfall Measuring
Mission TRMM; merged product. Both schemes
significantly overpredict surface precipitation compared to
observations from the TRMM 3B42 product (although the spatial
patterns of precipitation are reasonably similar); however, total
accumulated surface precipitation over the 10-day simulation
period is much greater in CONGO-THOM than CONGO-MORR
(Fig. ). Further differences are seen when
the distributions of precipitation rates are compared, with
CONGO-THOM overpredicting and CONGO-MORR underpredicting the
occurrence of low precipitation rates compared to TRMM, CONGO-MORR
overpredicting and CONGO-THOM underpredicting moderate rates, and
CONGO-THOM overpredicting the frequency of occurrence of very high
precipitation rates
(Fig. b). That CONGO-MORR
overpredicts the frequency of moderate rain rates and CONGO-THOM
overpredicts the frequency of very high rain rates likely explains
why both schemes overpredict total accumulated surface rain
compared to the observations. Additionally, the overprediction of
the frequency of very high precipitation rates by CONGO-THOM is
likely the reason that the total accumulated surface precipitation
is much greater in this scheme than in CONGO-MORR
(Fig. a and b).
Congo case: zonal mean vertical sections of hydrometeor classes (colour
contours) from 1 to 10 August 2007. Hydrometeor mass mixing ratios are
contoured at 10-6 kgkg-1.
Congo case: 10-day histogram for the period 1–10 August 2007 of model
reflectivities derived from hydrometeor fields passed through the QuickBeam
radar simulator ; thresholded at values greater than
-20 dBZ for (a) CONGO-M250, (b) CONGO-T250 and
(c) the CloudSat 2B-GEOPROF product.
In panels (a) and (b) the models have been sampled at the times of the nearest CloudSat overpasses.
Further to the significant difference between the two schemes in
their reproduction of cold cloud and precipitation rates, the
updraught dynamics respond very differently to aerosol
loading. Joint histograms of cloud top height in the convective
updraughts and the radius of the updraughts show that the most
significant dynamical difference between the simulations comes
from the choice of microphysics scheme: the Morrison scheme has
a tendency towards higher frequencies of wider updraught radii
with higher cloud tops than the Thompson scheme (Fig. S1 in the
Supplement). Under increased values of CDNC, convection in the
CONGO-MORR simulation shifts towards wider cores and higher core
tops for midsized cores (radius 11 to 22 km), whilst
there is a reduction in the frequency of smaller cores of all core
top heights
(Fig. c). Conversely,
convection under polluted conditions in the CONGO-THOM simulation
shows a reduced frequency of occurrence of the highest updraught
cloud tops for all updraught radii under polluted conditions with
an increased frequency of occurrence of small updraught radii with
lower cloud tops
(Fig. d). Therefore,
a consistent aerosol response is observed in CONGO-THOM, resulting
in smaller and lower convective updraughts (i.e. weakened
convection under polluted conditions). Interestingly, both of
these effects contradict the findings of , who
found an ice-phase response to aerosol in which cloud top heights
and anvil ice mixing ratios increase under polluted conditions due
to increased freezing of larger numbers of cloud droplets and
subsequent higher ice particle concentrations with smaller sizes
and reduced fall speeds. However, we note that we consider
different values of CDNC/CCN to and that
responses may be nonmonotonic . We also
consider a different case of convection (indeed, our 10-day Congo
simulation covers many convective lifecycles). We note that the
response of the convective updraughts to aerosol loading in these
two bulk schemes cannot be attributed to saturation adjustment
alone the suggested effects of which on updraught
invigoration are detailed in because both schemes
use this method.
Congo case: mean vertical profiles of hydrometeor mass mixing ratios (MMRs)
averaged over the period 1–10 August 2007. (a) CONGO-M250 cloudy
column domain mean, (b) CONGO-T250 domain mean, (c) the
difference in the domain-mean hydrometeor mixing ratio profiles (CONGO-M250
minus CONGO-T250), (d) CONGO-M250 mean over condensed points only,
(e) CONGO-T250 mean over condensed points only for each hydrometeor
class and (f) the difference in the condensate-mean hydrometeor
mixing ratio profiles (CONGO-M250 minus CONGO-T250). Note the logarithmic
horizontal axis used in panels (a–c) due to the total difference between
the hydrometeor classes simulated by the two schemes spanning several orders
of magnitude.
Not only does the simulated cloud and precipitation morphology
differ significantly between microphysics schemes irrespective of
the CDNC values used in the comparison, zonal-mean vertical
sections of the mass mixing ratios of the different hydrometeor
classes show significant differences in the hydrometeor classes
(due to microphysics) between CONGO-MORR and CONGO-THOM
(Fig. ). The most significant difference
between the two microphysics schemes is that south of
3∘ S, CONGO-MORR produces a large amount of high ice
cloud between 300 and 150 hPa (Fig. a–c). Analysis of these vertical sections at hourly intervals
throughout the simulation in conjunction with hourly maps of OLR,
as in Fig. , show that this upper-level ice is
transported from the convective anvils in the north of the domain
to the non-convective region in the south of the domain (not
shown). In comparison, CONGO-THOM produces significantly less ice
with almost no ice visible at this contour value
(Fig. d–f). However, all three CONGO-THOM
simulations form a large amount of non-precipitating low-level
(950 to 850 hPa) liquid cloud south of 3∘ S. The
bands of cloud seen south of 3∘ S in
Fig. are therefore high ice cloud in the
CONGO-MORR simulations and low liquid cloud in the CONGO-THOM
simulations, illustrating not only a cloud morphological
difference between the microphysics schemes but also a significant
difference in the simulated hydrometeor classes and in the
vertical distribution of hydrometeors. Even in the convective
precipitating region in the north of the domain, the simulated
hydrometeor classes differ significantly between the microphysics
configurations with the CONGO-MORR simulations generating more
ice and less liquid cloud (Fig. a–c) and
the CONGO-THOM simulations producing less ice and more liquid
cloud (Fig. d–f). Rain is confined to the
convective region in the north in CONGO-THOM, while in CONGO-MORR
it is also present at low levels in the non-convective southern
region of the domain which is dominated by liquid cloud in
CONGO-THOM. We explain the mechanisms behind these differences
later, but here we highlight that it is clear from
Fig. that the differences in the simulated
hydrometeors between microphysics schemes are much greater than
the differences due to different levels of CDNC.
Because the partitioning of water into liquid and ice phases in
the full-physics model configuration appears to depend strongly on
the microphysics scheme, vertical sections of reflectivity
occurrences derived from model hydrometeor fields passed through
the QuickBeam radar simulator are compared
against equivalent reflectivity occurrences from the CloudSat
2B-GEOPROF product
(Fig. ). The histograms are derived from
the reflectivity fields thresholded to include all values greater
than -20 dBZ. The largest reflectivity values produced by the
model occur in the convective region in the north of the domain
where the largest reflectivity values are detected by the
satellite radar (Fig. ), which is also in agreement
with the TRMM precipitation observations
(Fig. ). However, both CONGO-MORR and
CONGO-THOM have a large positive bias in reflectivity compared to
the observations (Fig. ), which is indicative of
limitations in the ability of both bulk microphysics schemes to
represent the observed vertical cloud structure in this geographic
region over this time period. In general, CONGO-MORR has a much
larger positive bias in reflectivity than CONGO-THOM
(Fig. ). The CloudSat observations show
a small frequency of occurrence of reflectivities detected at
altitudes of 10 to 15 km in the south of the domain, which
is well represented by CONGO-THOM and indicates the overproduction
of ice in CONGO-MORR (Fig. ).
Differences in the simulated hydrometeor classes between the
schemes persist throughout the simulation and are illustrated by
mean profiles of hydrometeor mass mixing ratios
(Fig. ). There is significantly
more ice-phase condensate in the CONGO-M250 configuration
(Fig. a), whereas the CONGO-T250
profile is dominated by a large amount of liquid cloud mass
between the near surface and 750 hPa
(Fig. b). The differences in the
total cloud water mass between the schemes are very large: at
950 hPa (the altitude with the greatest liquid cloud mass
in CONGO-T250; Fig. ), cloud water
mass contents are about 140 times greater in CONGO-T250. The
liquid cloud mass is always greater in CONGO-T250 than CONGO-M250
(Fig. c) by several orders of
magnitude at some levels, but despite this the liquid phase does
not appear to drive differences in precipitation between the
microphysics schemes: CONGO-M250 has about 4 times more rain
mass in the mid-levels and 2 times more rain mass near the
surface than CONGO-T250
(Fig. a and b). In the ice phase,
CONGO-M250 has only slightly more snow mass than CONGO-T250 but up
to 10 times more graupel mass
(Fig. a and b), and while ice is
a significant hydrometeor at upper levels in CONGO-M250,
CONGO-T250 has almost no cloud ice at all
(Fig. a and b). We note that the
magnitude of the difference due to the choice of scheme is the same
when a bin scheme is used (Figs. S2 and S3 in the Supplement).
Mean profiles over all condensed points (i.e. representing the
mean values of each hydrometeor type but not accounting for
changes in absolute quantities across the model domain) show that
CONGO-T250 has consistently more cloud water through the depth of
the mean cloud compared to CONGO-M250
(Fig. a and b), while CONGO-M250
produces more rain (Fig. a). That
rain production in CONGO-M250 occurs mostly through the depths of
the atmosphere where cloud water persists suggests that
a significant proportion of the rain may be produced though
autoconversion in CONGO-M250, although note that these mean cloud
profiles are calculated over the entire domain and therefore
incorporate both the deep convective region in the north and the
warm-cloud region in the south, as seen in
Fig. . Further, the two schemes show
differences in the frozen hydrometeors with the mean cloud in
CONGO-M250 containing more graupel and less snow than CONGO-T250
(Fig. c). This may be a result of
the use of distinct and different definitions of ice-phase
hydrometeor categories in the two schemes, which have been shown
to cause deficiencies in simulations of observed squall lines
.
Congo case: difference in the mean hydrometeor mixing ratio profiles under polluted and pristine conditions averaged over the period 1–10 August 2007. (a) CONGO-M2500 cloudy column domain mean minus CONGO-M100 domain-mean, (b) CONGO-T2500 domain-mean minus CONGO-T100 domain mean, (c) CONGO-M2500 mean over all condensed points of each hydrometeor class minus CONGO-M100 mean over all condensed points and (d) ONGO-T2500 mean over all condensed points minus CONGO-T100 mean over all condensed points.
Not only does the partitioning of ice amongst the hydrometeor
classes differ between schemes, the response of the hydrometeors
to CDNC perturbations also differs between schemes
(Fig. ). First note that the scale of
the hydrometeor response to CDNC perturbations in the CONGO-MORR
simulations is an order of magnitude smaller than the scale of the
response in the CONGO-THOM simulations. Over the entire domain,
liquid cloud mass appears insensitive to CDNC perturbations in the
CONGO-MORR configuration (Fig. a),
although a reduction in mean-cloud liquid cloud mass under
polluted conditions (Fig. c)
indicates that there must be very few liquid cloud points in the
CONGO-MORR simulation compared to other hydrometeor types, notably
ice (Fig. a). Very weak decreases in
domain-mean near-surface rain mass may be evident under polluted
conditions in CONGO-MORR, but this difference is on the order of
10-8 kgkg-1
(Fig. a) A reduction in rain mass
under polluted conditions is more evident in the mean rain profile
(Fig. c), again indicating how few
rainy points exist compared to other hydrometeor types in
CONGO-MORR when considering the entire domain
(Fig. a). Nearly all of the
hydrometeor response in CONGO-MORR occurs in the ice-phase
processes: graupel mass decreases significantly under polluted
conditions (Fig. a and c), while ice mass
increases at upper levels in both a domain mean and ice mean sense
(Fig. a and c). In contrast, the
hydrometeor response to CDNC perturbations in the CONGO-THOM
configuration is an order of magnitude greater than in CONGO-MORR
and the dominant hydrometeor response to CDNC perturbations in
CONGO-THOM occurs in the liquid phase. Not only does the
CONGO-THOM configuration generate significantly more liquid
cloud than the CONGO-MORR configuration
(Fig. c), but the liquid cloud
mass also increases under polluted conditions by an order of magnitude
more than any other hydrometeor response
(Fig. b and d). Rain mass is relatively
insensitive to increased CDNC in CONGO-THOM
(Fig. b and d). The significant
difference between the response of the two schemes to
perturbations in CDNC, with CONGO-MORR producing less liquid cloud
and rain under polluted conditions while CONGO-THOM produces more
cloud water, indicates significant differences in the cloud
processes represented by the two schemes in this meteorological
regime.
WRF idealised supercell
The results from the real-data Congo basin simulations indicate
that the development of the simulated hydrometeor classes and the
response of the hydrometeors to CDNC perturbations depend strongly
on the choice of microphysics scheme. Although some previous
studies have focused on the response of real-data case studies to
both microphysics scheme and CDNC response
e.g., there is a much larger
body of literature that investigates the response of idealised
supercell simulations to CDNC (or CCN) perturbations
e.g.. We therefore place our study in the wider context
of the existing literature by investigating the response of
a single isolated idealised supercell under both the MORR and THOM
microphysics configurations to the same CDNC perturbations used in
our Congo simulations, simultaneously allowing us to explore the
case dependence of the deep convective response to aerosol
effects.
Idealised supercell: mean vertical profiles of hydrometeor mass mixing ratios
(MMRs), as in Fig. , averaged over the
2 h of the supercell simulation. (a) SUPER-M250 domain mean,
(b) SUPER-T250 cloudy column domain mean, (c) SUPER-M250
domain mean minus SUPER-T250 domain mean, (d) SUPER-M250
condensate mean of each hydrometeor class, (e) SUPER-T250
condensate mean and (f) SUPER-M250 condensate mean minus SUPER-T250
condensate mean.
Idealised supercell: difference in the mean hydrometeor mixing ratio profiles
under polluted and pristine conditions, as in
Fig. , averaged over the 2 h of the
supercell simulation. (a) SUPER-M2500 cloudy column domain mean
minus SUPER-M100 domain mean, (b) SUPER-T2500 domain mean minus
SUPER-T100 domain mean, (c) SUPER-M2500 condensate mean of each
hydrometeor class minus SUPER-M100 condensate mean and (d)
SUPER-T2500 condensate mean minus SUPER-T100 condensate mean.
Idealised supercell: (a) vertical profiles of domain-mean total
latent heating rate (LHR) over the 2 h of the supercell simulation
for SUPER-MORR and SUPER-THOM for CDNC values of 100, 250 and
2500 cm-3. (b) Difference in the total latent heating
contributions over the 2 h of the supercell simulation for
SUPER-M2500 minus SUPER-M100 and SUPER-T2500 minus SUPER-T100.
Figure shows mean hydrometeor
profiles from the idealised supercell model configurations under
“moderately polluted” prescribed CDNC values of
250 cm-3. As in the Congo basin case, it is clear that
the simulated hydrometeor classes differ significantly between
schemes. In contrast to the Congo basin configuration, both the
SUPER-MORR and SUPER-THOM configurations show similar behaviour in
the liquid phase, producing similar profiles of liquid cloud mass
and rain mass in both a domain mean and hydrometeor-class mean
sense (Fig. a, d and b, e),
and instead the most significant differences occur in the ice
phase. Graupel dominates as the frozen precipitating hydrometeor
in the SUPER-M250 configuration, amounting to about 4 times the
snow and ice masses at their peak amounts
(Fig. a and d). In contrast, snow
is the dominant frozen precipitating hydrometeor in the SUPER-T250
configuration, amounting to about 1.5 times the graupel mass at
peak amounts and virtually no ice present
(Fig. b and e). Although there is
very little difference between the SUPER-MORR and SUPER-THOM
configurations in the liquid phase (except for the SUPER-MORR
configuration producing about 2×10-7 kgkg-1 less domain-mean rain mass at
the surface than SUPER-THOM;
Fig. c), the SUPER-MORR
configuration forms significantly more ice, more graupel and less
snow than SUPER-THOM (highlighting that the partitioning of
ice-phase hydrometeors into categories is very different, by
design, in different microphysics schemes). Greater total
quantities of frozen hydrometeors are present between 600 and
about 150 hPa in SUPER-MORR compared to SUPER-THOM
(Fig. c and f). This is
a significant difference from the Congo real-data configuration
in which the dominant contribution to the difference between the
CONGO-MORR and CONGO-THOM configurations came from the liquid
cloud (Fig. c).
There is a more significant aerosol impact on hydrometeor mass in
the supercell case than in the Congo case for both microphysics
schemes, with mean responses over each hydrometeor type an order
of magnitude greater in the supercell case
(Fig. compared to
Fig. ). Although many past
studies have shown that aerosol impacts depend on cloud dynamics
and thermodynamics e.g., we note
that not only do the individual schemes respond differently to
CDNC in different cases of convection (as expected), but the
way the schemes differ from each other in their response to CDNC
is also significantly different in the supercell case compared to the
Congo case. The SUPER-MORR and SUPER-THOM cases differ
qualitatively from the CONGO-MORR and CONGO-THOM cases,
respectively, both in the altitudes at which the response occurs
and the sign of the response of some of the hydrometeors. In the
SUPER-MORR configuration, cloud water mass increases under
polluted conditions, and rain mass is suppressed at mid-levels
(between 600 and 450 hPa) but shows negligible response at
the surface (Fig. a and c). In the
ice phase, cloud ice increases under polluted conditions in
SUPER-MORR, while graupel and snow decrease
(Fig. a and c). Similarly, the
hydrometeor response of the SUPER-THOM case to CDNC perturbations
also differs in sign and in altitude to CONGO-THOM. In SUPER-THOM,
cloud water mass increases and rain mass decreases under polluted
conditions (Fig. b and d), but
unlike SUPER-MORR the decrease in rain is evident at the
surface. Graupel mass decreases under polluted conditions in
SUPER-THOM, similarly to SUPER-MORR, but occurs over a much
larger range of heights
(Fig. b and d); this is unlike
CONGO-THOM, which shows very little response to polluted
conditions (Fig. c). Interestingly,
this is in contrast to , who found an increase in
graupel mass with increased CDNC in the Thompson scheme. However,
their study was of 2-D idealised squall line simulations and
considered CDNC values of 100, 500 and 100 drops per
cm-3. The dominant domain-mean response to increased
CDNC perturbations in SUPER-THOM is an increase in snow mass
between 550 and 150 hPa
(Fig. b), which likely comes from
lofting of an increased mass of cloud water
(Fig. d). This is in contrast
both to SUPER-MORR in which the dominant hydrometeor response
occurred in the ice class
(Fig. a), despite an almost equal
increase in lofted cloud water
(Fig. c), and to CONGO-THOM
in which the dominant hydrometeor response occurred in the liquid
cloud (Fig. b). That both schemes
show an increased lofting of cloud water under polluted conditions
(Fig. c and d), but SUPER-MORR responds
by generating more cloud ice
(Fig. a and c) while SUPER-THOM shows an
increase in snow (Fig. b and d), suggests
differences in the processes that convert cloud ice to snow. This
is explored later in Sect. . We emphasise
that our main result shows that the variability due to
microphysics scheme dominates any aerosol impacts on
microphysics. Results using the WRF-SBM in the idealised supercell
case show that aerosol impacts in the bin scheme are of equal
magnitude to those in the bulk schemes (Fig. S4 in the
Supplement).
To further investigate the importance of the difference in
microphysics representations and the difference in their response
to CDNC perturbations,
Fig. includes the
domain-mean total latent heating (sum of the latent heating from
individual microphysical processes) contributions for each of the
idealised supercell configurations. It can be seen that the choice
of microphysics scheme can result in thermodynamic differences in
the supercell system equal in magnitude to those arising from CDNC
perturbations: between 500 and 250 hPa, the latent heating
rate in the SUPER-M2500 configuration is almost identical to that
in the SUPER-T250 configuration (solid red and dashed blue lines,
Fig. a). Thus, the
magnitude and sign of the difference in the latent heating rate
between SUPER-M250 and SUPER-T250 (blue solid and dashed lines,
Fig. a) is the same as
that between SUPER-M2500 and SUPER-M250 (red and blue solid
lines), and likewise the magnitude and sign of the difference in
the latent heating rate between SUPER-M2500 and SUPER-T2500 (red
solid and dashed lines) is the same as that between SUPER-T2500
and SUPER-T250 (red and blue dashed lines). In general, the
SUPER-THOM configuration has a much stronger thermodynamic
response to CDNC perturbations than the SUPER-MORR configuration,
with latent heating rates consistently stronger throughout the
atmosphere
(Fig. b). Overall, there
is little evidence of convective invigoration (defined here as
increases in upper tropospheric heating, updraught strengths,
cloud top height and surface precipitation) under increased CDNC
values in either bulk microphysics scheme. Although both schemes
show increased latent heating in the upper troposphere and
decreased heating at mid-levels under polluted conditions
(Fig. b), it has already
been shown that there is no evidence of increased surface
precipitation (Fig. ), and the
upper tropospheric peak in latent heating can be seen to
correspond to an increase in ice (SUPER-M250) or snow (SUPER-T250)
at these levels (Fig. ).
There is no systematic or consistent evidence of increased mean
updraught velocity in the convective cores following the
method of under polluted
conditions (not shown) or in increased cloud top heights of the
convective cores
(Fig. c and d). This may not
be surprising, as it has been suggested that bulk microphysics
schemes are unable by design to produce convective updraught
invigoration effects due to limitations in their representation of
nucleation, sedimentation and the way in which saturation
adjustment limits diffusional growth detailed
in. Indeed, found no latent heating
effect of increased CCN in a bulk scheme used to simulate
idealised deep convection, whereas with a bin scheme increased
latent heating aloft was demonstrated. However,
found that saturation adjustment methods used in bulk schemes
could explain differences in the response of cold pool evolution
and convective dynamics between bin and bulk schemes to aerosol
loading but could not explain large differences in the response
of surface precipitation. Further, some simulations using bulk
schemes have identified invigoration-like effects under aerosol
loading. For example, found evidence of
convective invigoration under increased aerosol loading in a bulk
scheme under weak shear conditions (and suppressed convection
under strong shear), similar to the findings of
who found the same response in a bin scheme.
also found higher overshooting tops and larger sizes of
cumulonimbus in a weak shear environment with increased aerosol
loading. Thus although our results agree with the body of the
literature which does not identify a convective updraught
invigoration effect when bulk microphysics schemes are used, this
is not necessarily attributable to the saturation adjustment
method alone and may also only hold for the particular convective
environment (idealised supercell in strong shear) we consider.
RICO case: mean vertical profiles of hydrometeor mass mixing ratios (MMRs), as
in
Fig. , averaged over the 24 h of the
RICO simulation. (a) RICO-M100 cloudy column domain mean,
(b) RICO-T100 domain mean, (c) RICO-M100 domain mean minus
RICO-T100 domain mean, (d) RICO-M100 condensate mean over each
hydrometeor class, (e) RICO-T100 condensate mean and (f)
RICO-M100 condensate mean minus RICO-T100 condensate mean. Note that because
the rain amounts are very small, especially in M100, panels (a, b) are
shown with a logarithmic horizontal axis.
RICO case: difference in the mean hydrometeor mixing ratio profiles under
polluted and pristine conditions in cloudy columns, as in
Fig. , averaged over the 24 h of the
RICO simulation. (a) RICO-M2500 cloudy column domain mean minus
RICO-M100 domain mean, (b) RICO-T2500 domain mean minus RICO-T100
domain mean, (c) RICO-M2500 condensate mean over each hydrometeor
class minus RICO-M100 condensate mean and (d) RICO-T2500
condensate mean minus RICO-T100 condensate mean.
Total accumulated surface rain (mm) for each of the microphysics simulations,
including a series of sensitivity simulations, for (a) the RICO case
total after 24 h of simulation and (b) the Congo case total over
the period 1–10 August 2007. Note that because the magnitude of the rain
response to CDNC differs so strongly between the configurations in the RICO
case, a logarithmic vertical axis is used in panel (a). The horizontal
dashed line in panel (b) indicates the total precipitation from the
TRMM 2A25 product over the same period.
WRF LES RICO
The results presented in Sect. and
indicate that not only is the way in
which the schemes differ from each other not systematic between
cases of convection, but the difference between the response
of the two schemes to CDNC across types of convection is also not
systematic. The largest difference between the microphysics
schemes in the real-data Congo basin simulations occurs in the
liquid-phase hydrometeor development and response to CDNC. Making
the assumption that the liquid phase is the first to respond to
CDNC perturbations and the perturbation subsequently propagates to
the ice phase, we consider a case of precipitating shallow cumulus
convection to investigate the liquid-phase differences between the
schemes. Note that the “baseline” hydrometeor profiles in
Fig. show data from the
configurations using a prescribed CDNC value of
100 cm-3 (rather than the baseline value of
250 cm-3 used in the Congo basin and idealised
supercell deep convection cases in
Figs. and
), as this is more appropriate
for a pristine marine environment. Even when we restrict our
simulations to the liquid phase, differences in the simulated
hydrometeor classes are evident. The dominant domain-mean
difference between the two schemes in the RICO case is clearly in
the rain profile, with RICO-T100 producing significantly more rain
than RICO-M100. Very little rain is present in the RICO-M100
configuration (Fig. a and d), whilst
the RICO-T100 configuration produces a peak domain-mean rain mass
of about 10-6 kgkg-1
(Fig. b). The liquid cloud profile
is similar in both schemes, with RICO-M100 forming more cloud mass
than RICO-T100 between 805 and 775 hPa in both the
domain mean and hydrometeor-class mean sense
(Fig. c and f).
The response of the hydrometeors to CDNC perturbations also
differs between schemes in the warm-rain RICO case
(Fig. ). In the RICO-MORR
configuration, domain-mean rain and cloud mass both decrease under
polluted conditions, although the rain response is very weak (on
the order of 10-8 kgkg-1) and the dominant
response is a reduction in liquid cloud mass
(Fig. a). In contrast, a reduction in
rain mass is the dominant hydrometeor response under polluted
conditions in the RICO-THOM configuration, and the decrease is
nearly 2 orders of magnitude greater than that in RICO-MORR
(Fig. b). The liquid cloud response to
polluted conditions in RICO-THOM is weaker than the rain response
but still stronger than the cloud response in RICO-MORR. Cloud
mass decreases under polluted conditions between 935 and
825 hPa but increases at higher levels
(Fig. b). Note that once again the
response of the simulated hydrometeors to CDNC perturbations
differs between cases: under polluted conditions, RICO-MORR
exhibits a decrease in cloud and rain mass, while CONGO-MORR
exhibits a decrease in rain mass with little response in the liquid
cloud (Fig. a), and SUPER-MORR shows
almost no liquid-phase response at all
(Fig. a). Likewise, RICO-THOM
exhibits a decrease in rain and an increase in cloud mass under
polluted conditions, while CONGO-THOM exhibits similar behaviour
(Fig. b), but SUPER-THOM shows
a decrease in rain mass with little response in the liquid cloud
(Fig. b). When mean profiles of
each hydrometeor class are considered, the two schemes actually
show similar responses to CDNC (increased upper-level cloud mass
and suppressed rain;
Fig. c and d). This indicates that
the main response to CDNC in this case is not through the
individual microphysical processes but through the absolute
amounts of cloud and rain that are generated.
To illustrate the difference in the strength of response of the
schemes to CDNC, total accumulated surface rain is shown for each
RICO configuration in Fig. a along with
an extra configuration using a “very pristine” CDNC value of
50 cm-3 and a series of sensitivity tests that will be
discussed later. The 50 cm-3 CDNC configuration has
been added because even at a prescribed CDNC value of
100 cm-3 very little rain production occurs in the
RICO-MORR configuration. Warm rain formation differs strongly
between schemes: very low CDNC values are required for the
RICO-MORR configuration to produce any rain, whereas RICO-THOM
produces significantly more rain at all CDNC values
(Fig. a). Even under very pristine
conditions, the RICO-M50 configuration produces an order of
magnitude less rain than RICO-T50
(Fig. a). The different schemes also
respond differently to CDNC perturbations. Rain production in
RICO-MORR (which produces much less rain than RICO-THOM) shuts
down very quickly as CDNC is increased: rain amounts are on the
order of 102 mm at a CDNC value of
50 cm-3, 101 mm at a CDNC value of
100 cm-3 and 10-1 mm at a CDNC value of
250 cm-3; rain production ceases completely at
a CDNC value of 2500 cm-3
(Fig. a). In contrast, rain production
persists for much larger CDNC values in RICO-THOM: rain amounts
are on the order of 103 mm at CDNC values of
50 cm-3, 102 mm at CDNC values of
100 cm-3 and 101 mm at CDNC values of
250 cm-3. While rain amounts are very low at CDNC
values of 2500 cm-3 (on the order of
10-5 mm), rain production has not shut down
completely (Fig. a).
Autoconversion rate as a function of cloud water content for the MORR and
THOM microphysics schemes (solid and dashed lines, respectively) for
super pristine, pristine, moderately polluted and polluted conditions. Also
shown are labelled grey bars showing the mean (solid vertical grey line) and
1 and 2 standard deviations (dashed vertical grey line and end of bar,
respectively) for cloud water content averaged over all prescribed CDNC
configurations for each case (note that the variability in mean cloud water
content with CDNC is significantly less than the variability due to
microphysics scheme).
RICO case: difference in the cloudy column domain-mean vertical profiles of
hydrometeor mass mixing ratios (MMR) between MORR and THOM, as in
Fig. c, averaged over the 24 h of the
RICO simulation for the configurations with the autoconversion treatment
swapped between the microphysics schemes (a) M100 minus M100T,
(b) T100M minus T100 and (c) T100M minus M100T.
Sensitivity tests
showed that the rain rates predicted by different autoconversion formulae in bulk schemes can vary by orders of magnitude. This sensitivity of results is also well highlighted in ; note in particular their reference to . Autoconversion is parameterised differently in the two microphysics schemes used in the current paper. The Thompson scheme follows an adaptation of , while the Morrison scheme follows the method of .
and justify their choice
of an adapted version of the autoconversion
parameterisation through favourable comparison to results from the
bin scheme of . Furthermore, implementation of
in the Thompson scheme begins the
collision–coalescence production of warm rain at almost exactly 14 µm. It is known that raindrop onset begins when the mean
volume radius exceeds a critical value of 13 to 14 µm
. This is one of the
principle reasons the autoconversion scheme was
chosen by rather than
.
While the autoconversion scheme was
initially developed and applied for LES of stratocumulus, other
than varying the prescribed values of cloud droplet number
concentrations we run the microphysics schemes in their baseline
configurations. Thus, although we do not advocate the use of
for non-stratocumulus cases, the Morrison
scheme is frequently used for simulations of deep
convection. Similarly, as the
autoconversion scheme was developed for LES-scale studies, the
authors recognise the potential importance of subgrid cloud
variability at the scales used in the present study. However, we
note that we are running the model and microphysics schemes in the
typical set-up for a convection-permitting model (that is,
neglecting subgrid cloud variability), as one of the main aims of
this study is to highlight uncertainty in commonly used model
configurations which are exactly based on these schemes.
The autoconversion rates as a function of cloud water content for
each of the model configurations are shown in
Fig. . Also shown is the cloud
water content (up to the mean plus 2 standard deviations) of each
configuration . It is immediately clear that the threshold cloud
liquid content for autoconversion in the Morrison scheme (solid
lines) is significantly lower than that in the Thompson scheme
(dashed lines); i.e. rain production can occur at much lower cloud
liquid water contents in Morrison. It is also clear from the mean,
(mean +1 SD) and (mean +2 SD) cloud water content limits
that rain production through autoconversion ought to be possible in
all model configurations. However, despite the higher cloud water
content threshold for autoconversion in the Thompson scheme,
autoconversion rates are much greater once the threshold is
reached, and liquid cloud is converted to rain much faster in
Thompson than in Morrison. From
Fig. , it appears that the
threshold for autoconversion is unlikely to be reached very often
in any of the T2500 cases. In the deep convective cases, rain can
be generated through ice- and mixed-phase processes, but in the RICO
warm-rain case this cannot occur. This explains why, compared to
more pristine conditions, cloud mass increases in RICO-T2500 while
rain mass decreases (Fig. b).
Because Fig. indicates that
the autoconversion threshold may be at least in part responsible
for this response in the RICO-THOM case, we replace the
autoconversion parameterisation in the Morrison scheme with that
from the Thompson scheme and vice versa. We use the notation M100T
to denote the Morrison microphysics scheme (at a CDNC value of
100 cm-3) with Thompson autoconversion that
of and T100M to denote the Thompson scheme with
Morrison autoconversion that
of. Differences in the domain-mean
hydrometeor mixing ratio profiles for each of the
autoconversion swapped configurations in the RICO case are shown in
Fig. . It is immediately clear
that, in the warm-rain configuration, simply swapping the
autoconversion treatment makes the hydrometeor developments of the
microphysics schemes much more like each other. The differences
between the RICO-M100 configuration with the Morrison and Thompson
autoconversion parameterisations
(Fig. a) is quantitatively and
qualitatively very similar to the difference between the RICO-M100
and RICO-T100 configurations
(Fig. c). Likewise, the difference
between the RICO-T100 configuration with the Morrison and Thompson
autoconversion parameterisations
(Fig. b) and finally the
difference between the RICO-T100 configuration with the Morrison
autoconversion parameterisation and the RICO-M100 configuration
with the Thompson autoconversion parameterisation
(Fig. c) are also very similar
to the difference between the RICO-M100 and RICO-T100
configurations (Fig. c).
Similarly, swapping the autoconversion parameterisations between
the microphysics schemes in the RICO cases makes the surface rain
production of the microphysics schemes much more similar. The accumulated surface rainfall in the RICO-M100T
configuration looks much more similar to the surface rainfall in
the RICO-T100 configuration than it does to the RICO-M100
configuration (Fig. a). Rain amounts are
on the order of 102 mm in RICO-M100T and RICO-T100, whereas in
RICO-M100 it is 2 orders of magnitude smaller
(Fig. a). Likewise, the accumulated
surface rainfall in the RICO-T100M configuration is on the order of
101 mm compared to 102 mm in the standard RICO-T100 case
(Fig. a). To further test the importance
of autoconversion in the liquid phase simulations, we first turn
off autoconversion completely in the 100 cm-3 CDNC
simulations, and then allow autoconversion to occur but prevent the
accretion of cloud water by rain. By design, in the absence of ice
processes no precipitation occurs without autoconversion of cloud
water to rain (Fig. a, M100noAUTO and
T100noAUTO). However, in the RICO 100 cm-3 CDNC
liquid-phase configuration, the Thompson scheme can produce surface
rain from autoconversion alone (albeit 2 orders of magnitude less
than when rain can also accrete cloud water
(Fig. a, T100noACCR and T100), showing
that autoconversion acts almost like a “trigger” in this scheme
after which accretion takes over the rain production
process. (Indeed, in nearly all schemes, rain formation from
accretion, once triggered, is orders of magnitude larger than from
autoconversion.) In contrast, zero surface precipitation is
produced in RICO M100noACCR (Fig. a),
showing that in this (liquid-phase only) configuration the Morrison
scheme requires both the autoconversion of cloud droplets to rain
and the accretion of rain by cloud droplets in order to produce
surface precipitation.
Despite the significant effect of autoconversion in the
liquid-phase simulations, changing the autoconversion
parameterisation in the idealised supercell case has very little
effect on the hydrometeor development (results not shown). This is
unsurprising, as ice- and mixed-phase processes will dominate this
shear-driven deep convective environment. However, the Congo basin
configurations show large differences between microphysics schemes
in the partitioning of water into liquid and ice phases (CONGO-THOM
produces much more liquid cloud; CONGO-MORR produces much more
ice). In the CONGO-THOM configurations, the liquid-phase response
to increased CDNC is also very similar to the RICO-THOM response
(increased liquid cloud mass and decreased rain mass;
Fig. b). When the Thompson
autoconversion treatment is implemented in the Morrison scheme,
rain production in the southern half of the domain ceases in
CONGO-M250T, and the liquid phase is instead represented by
low-level cloud with structure similar to the CONGO-T250
configuration (Fig. a
compared to Fig. e). To test if radiative
effects associated with large amounts of anvil ice drive or
contribute to the differences in low cloud, we also set the ice
extinction coefficient to zero in both the longwave and shortwave
radiation schemes in CONGO-M250. However, this has no effect on the
low-cloud characteristics
(Fig. e compared to
Fig. b), and we therefore conclude that
autoconversion of cloud water to rain is the factor dominating the
absence of low-level cloud in the south of the domain in the
CONGO-MORR simulations. In contrast, autoconversion is a less
significant process in the CONGO-T250 configuration. Implementing
the Morrison autoconversion treatment in the Thompson scheme has
very little effect on the hydrometeor structure in the CONGO-T250M
configuration compared to the CONGO-M250 configuration
(Fig. b compared to
Fig. b). As a final test, the autoconversion
process is turned off in both of the microphysics schemes. This
confirms that autoconversion dominates the lack of low cloud in
CONGO-M250: the resulting liquid-phase hydrometeor structure
(Fig. c) is similar to both
CONGO-T250 (Fig. b) and CONGO-M250T
(Fig. a). This also
confirms that autoconversion is much less significant in the
CONGO-THOM configurations: the bulk hydrometeor structure when
autoconversion is turned off in CONGO-T250
(Fig. d) is very similar to
both CONGO-T250 (Fig. b) and
CONGO-T250M (Fig. b).
Congo case: zonal mean vertical sections of hydrometeor classes
(colour contours) from 1 to 10 August 2007, as in Fig. ,
but for the configurations with the autoconversion treatment swapped between
the microphysics schemes (a) CONGO-M250T and (b)
CONGO-T250M; for the configurations with (c) CONGO-M250 with
autoconversion turned off, (d) CONGO-T250 with autoconversion turned
off, (e) CONGO-M250 with the ice extinction coefficient set to zero
in the longwave and shortwave radiation schemes, (f) CONGO-M250 with
the threshold size parameter for conversion of ice to snow reduced to
50 % of its default value, (g) CONGO-M250 with the
threshold size parameter for conversion of ice to snow reduced to
10 % of its default value and (h) CONGO-M250 with the
autoconversion of ice to snow replaced by that used in the Thompson
microphysics scheme. Hydrometeor mass mixing ratios are contoured at
10-6 kgkg-1.
We have shown that the autoconversion process is responsible for
the removal of the large cloud mass at low levels in the model
configuration with the Morrison microphysics scheme. We also see
that this low-level liquid-phase cloud mass forms when we run the
same simulation using the WRF bin microphysics implementation
the SBM part of the Hebrew University Cloud
Model;, although to a lesser extent than in the
Thompson simulations, and the warm cloud produced by the WRF-SBM
produces rain (Figs. S2 and S3 in the Supplement). We therefore
suggest that it is not the Thompson scheme per se which is
responsible for producing the low-level cloud mass, but rather the
larger-scale meteorological conditions in which these
simulations are performed.
A further significant difference between the two schemes in the
Congo simulations is the generation of large amounts of upper-level
ice in CONGO-M250, which is not present in CONGO-T250
(Fig. b and e). In the Thompson scheme, the
fraction of ice mass with a diameter greater than 125 µm
is instantaneously transferred into the snow category
. The same threshold size for cloud ice
autoconversion to snow is used in the Morrison scheme, but the
process is parameterised differently . Because
the Morrison scheme appears to produce large amounts of anvil
cloudiness for the Congo case, which is not seen in the
observations (Fig. ), we perform further
sensitivity tests in which we reduce the threshold size for cloud
ice autoconversion in the Morrison scheme to 50 % of its
original value (Fig. f) and
10 % of its original value
(Fig. g). We then finally
replace the autoconversion of cloud ice to snow in the Morrison
scheme with the parameterisation used in the Thompson scheme
(Fig. h). In all tests, the
upper-level anvil ice is reduced significantly. Using the lowest
value of the threshold size for cloud ice autoconversion reduces
the anvil cloud because almost all of the ice is immediately
converted to snow
(Fig. g). However, using
the Thompson ice autoconversion representation in the Morrison
scheme significantly reduces the amount of cloud ice in the
simulation, and all of the detrained anvil ice is removed
(Fig. h). This suggests
that for the particular Congo simulation we have investigated, the
conversion of cloud ice to snow is the main factor leading to the
significant difference in anvil cloudiness between the two schemes
and is responsible for the difference in upper-level cloud between
the CONGO-M250 simulation and the observations
(Fig. ). Indeed, we note that in equivalent
simulations performed with the WRF-SBM, the same persistent
upper-level ice forms (Fig. S2 in the Supplement). This shows that
differences resulting from conversion of one ice category into
another is a limitation of any scheme, whether bin or bulk, which
uses fixed ice categories. Our results provide further evidence
that the use of discrete ice-phase hydrometeor categories may be
detrimental to the correct simulation of cloud and suggest that
new schemes which do not use such partitioning may give better
results e.g..
Our results show little impact of aerosol on precipitation in the
Congo basin (Figs. b and
), which is also seen when considering
total accumulated surface precipitation
(Fig. b), although CONGO-T2500 exhibits
weak precipitation suppression under polluted conditions. This may
be due to the longer duration of these simulations performed over
a larger domain, allowing the interaction of many cloud systems
rather than considering the lifetime of a single isolated
cloud. However, we also see that although the representation of
autoconversion has a significant effect on the vertical hydrometeor
structure in the CONGO-M250 configurations
(Figs. b,
a and
c), it has a much weaker
effect on total surface precipitation
(Fig. b, M250, T250, M250T, T250M,
M250noAUTO and T250noAUTO). This is perhaps unsurprising, as the
dominant contribution to the accumulated surface precipitation over
the Congo domain will be from ice processes in the convective
region and not from the liquid-phase cloud. Although the lack of
impact of aerosol on precipitation in the Congo simulations may be
due to the use of bulk schemes in this study for the reasons
detailed in , and perhaps a different response
would be seen using a bin scheme e.g., other studies using bulk and bin–bulk schemes have
identified aerosol impacts on precipitation of up to about
15 % e.g.. Indeed, even studies using bin schemes have been
shown to have little impact on total precipitation, although they
induce a shift in rainfall rates . Therefore,
we note again that the choice of microphysics scheme, rather than
aerosol response in either scheme, is the dominant contribution to
uncertainty in the total precipitation.
Discussion and conclusions
This study considered the cloud and precipitation development using
two double-moment bulk microphysics schemes to perform cloud-system-resolving simulations of three
types of convection, two of which were idealised (one deep
convection case with open boundaries and one shallow cumulus case
with periodic boundaries), and one real-data case of deep
convection in the Congo basin using meteorological initial and
boundary conditions. We tested the sensitivity of the simulated
hydrometeors and precipitation to the microphysics scheme and to
CDNC perturbations. The simulations were performed to explore the
uncertainty in cloud and precipitation development and response to
aerosol perturbations in convection-permitting models that can
arise from the microphysics representation. We find that the
variability among the two schemes, including the response to
aerosol, differs widely between these cases. Although previous
studies have found large sensitivity to the choice of microphysics
schemes e.g., we show this in
a consistent set-up by considering different cases with the same
model and same CDNC values and constraining as many other possible
sources of variability as is feasible. Our results show that for
the bulk schemes used in these simulations, aerosol effects are
dominated by the uncertainty in cloud and precipitation development
which arises from the choice of microphysics scheme. This result
was true for multiple cloud types in multiple environmental
conditions.
A key finding is that the difference between the two schemes,
including their response to CDNC, in different environments and
cloud types is not systematic. This could perhaps be related to the
nonmonotonic response to aerosol in different environments found by
(although their study only considered
simulations of idealised supercells with a single bulk scheme and
four environmental soundings). This nonmonotonic response was
attributed to compensatory changes in the microphysical processes
under polluted conditions.
The maximum relative difference in mass mixing ratio between each
hydrometeor class in the M250 and T250 configurations for each case
of convection is summarised in
Table . Not only are the maximum
differences in the domain-mean profiles of the hydrometeor classes
simulated by each microphysics scheme on the order of at least tens
of percent, but it is also clear that both the magnitude and sign
of the difference varies between cases. In some cases, the
magnitude of the difference is huge: most notably in the Congo
basin case, the maximum difference in liquid cloud mass between the
Morrison and Thompson schemes is on the order of
104 kgkg-1 more in Thompson (whereas in the RICO
shallow cumulus case the maximum difference is on the order of
101 kgkg-1 less in Thompson). Likewise, in the RICO
case the maximum difference in rain mass between the Morrison and
Thompson schemes is on the order of 103 kgkg-1 more
in Thompson (whereas in the Congo basin case the maximum difference
is on the order of 101 kgkg-1 less in
Thompson). Even for hydrometeors that have differences of the same
order of magnitude, the sign of the difference can vary between
cases. This result highlights the need for better observational
constraints on mixed-phase and ice cloud microphysics and
hydrometeors, and also perhaps the need for a shift in the
development of microphysics parameterisations away from schemes
which (somewhat arbitrarily) partition hydrometeors into separate
categories. This is also supported by our sensitivity tests of
autoconversion of cloud ice to snow in our Congo simulations.
Maximum relative difference of domain-mean hydrometeor mass mixing ratio profiles for the MORR and THOM schemes. The relative change in the hydrometeor mass mixing ratios are computed in each case for M250 minus T250.
Difference
CONGO
SUPERCELL
RICO
Liquid cloud mass
-10 900 %
-58.3 %
+17.0 %
Ice mass
+98.7 %
+96.9 %
n/a
Rain mass
+82.2 %
-138 %
-3830 %
Snow mass
+40.8 %
-99.8 %
n/a
Graupel mass
+91.6 %
+72.7 %
n/a
n/a = not applicable
Another key finding is that the cloud morphological difference and
the difference in the hydrometeors between different schemes is
significantly larger than that due to CDNC perturbations. Although
we have restricted our study to the comparison of double-moment
bulk microphysics schemes, this result is consistent with
, who found that the difference in convection
between a bulk and a bin scheme was much greater than the
difference within each scheme to varying aerosol
concentrations. Some studies have found a significantly weaker
response to aerosol when using bulk schemes compared to bin
schemes; e.g. . In idealised simulations
of continental deep convection, found that
increases in CCN concentrations led to increased ice mass and total
condensed water mass aloft in both bin and bulk schemes but
increased domain-averaged cumulative surface precipitation
in the bulk scheme compared to a decrease in the bin scheme. This
was because the relative increase in condensate mass
aloft under polluted conditions was found to be much larger in the
simulations performed with bulk microphysics as a result of increased
numbers of smaller cloud particles with slower sedimentation
speeds, thus resulting in reduced surface precipitation. However,
in our idealised supercell simulations we find a similar magnitude
of response to aerosol when using a bin scheme as in
the two bulk schemes which are the focus of this study.
That cloud and precipitation development and their aerosol response
differs across different cloud types in different large-scale
environments is expected. Many studies have shown that aerosol
effects on precipitation depend on the large-scale environment and
cloud type e.g. for reasons related
to differences in different cloud types between the timescale of
increased sedimentation through aerosol loading and subsequent
sublimation and evaporation timescales. Further, several studies of
deep convection have found that the effects of aerosol on deep
convection are much weaker than those of relative humidity
e.g.. found that in idealised simulations
of continental and maritime clouds using bin microphysics the
magnitude and even the sign of aerosol effects on precipitation
depended on relative humidity. found that aerosol
response in idealised simulations of clouds using bin microphysics
and soundings from Houston, Texas strongly depended on relative
humidity with a negligible effect on cloud properties and
precipitation in dry air but more significant effects in humid
air. Conversely, in idealised low-precipitation supercell
simulations with bulk microphysics and dry low-level humidity
performed as part of the study by Kalina et al. (2014), cold pool
area decreased by 84 % and domain-averaged precipitation
was reduced by 50 % under polluted conditions; however, it was
insensitive to polluted conditions when a moist sounding was
used. Thus, assuming that the response in our simulations would
likely be more similar to the results found for bulk microphysics
by , the magnitude of our results in the supercell
case (which uses a moist sounding) may be smaller than it would be
in drier environmental conditions.
In 10-day simulations of deep convection in the Congo basin in
August 2007, we find that both the Morrison and Thompson schemes
have a significant positive bias in cloud and surface precipitation
compared to GERB and TRMM. This may be in part attributable to the
positive moist bias in the Congo basin in the ERA-Interim
reanalysis (used as boundary data for the Congo simulation) when
compared to other reanalyses . Despite the
positive cloud fraction bias in both schemes, we find that the
Thompson scheme compares better than the Morrison scheme against
observed cloud fractions, largely due to the overproduction of
upper-level ice in the Morrison scheme. This is in agreement with
, who found that (despite the two schemes
having different biases at different levels) the Thompson scheme
outperformed the Morrison scheme overall against satellite
observations of cloud in North America due to its more accurate
upper-level cloud distribution, whereas the Morrison scheme had too
much upper-level cloud through the overproduction of ice. This bias is
attributable to differences in the way in which the two schemes
convert cloud ice to snow. However, we also find that despite
a positive surface precipitation bias in both schemes, the Morrison
scheme compares better to observations in this region over this
period. found that differences in accumulated
precipitation produced by warm stratocumulus and warm cumulus
clouds using different microphysics schemes were only on the order
of 10 to 20 %, suggesting that accumulated rain is largely
controlled by large-scale atmospheric properties. However,
differences in accumulated rain in our Congo simulations can be
attributed to differences in the microphysics schemes because all
simulations used the same input and boundary data and therefore are
under the influence of the same large-scale atmospheric
conditions. That one scheme best represents cold cloud compared to
observations but the other scheme better reproduces accumulated
precipitation makes it difficult to conclude that one scheme
outperforms another overall. It also suggests that when setting up
a model configuration for research purposes, one consideration to
distinguish between the use of these two particular schemes may be
whether surface precipitation or radiative effects are more
important to the research question.
We note here that the RRTM LW and Goddard SW radiation schemes used
in these simulations are only coupled to the microphysics through
the hydrometeor masses and not the numbers. This coupling therefore
cannot account for changes in hydrometeor sizes, and thus some
aerosol effects will be missing from these
simulations. Additionally, the microphysics–radiation coupling is
only through cloud water and ice and none of the other frozen
species. This missing aerosol effect may have an especially
important impact in our Congo simulations in which the Morrison
scheme develops and retains significant amounts of upper-level ice,
whereas the Thompson scheme converts nearly all the ice to snow,
which the radiation scheme will not see. This could have
significant radiative flux and feedback impacts
which originate from the use of
somewhat arbitrarily defined ice categories (e.g. if the size
parameter at which cloud ice is converted to snow is changed,
a bulk mass of cloud ice is removed from the radiatively coupled
ice category and moved into the non-radiatively coupled snow
category).
Maximum relative difference in the response of model configurations to polluted conditions. The relative change in the domain-mean rehydrometeor mass mixing ratios are computed in each case for CDNC values of 2500 cm-3 minus 100 cm-3.
Difference
CONGO-MORR
CONGO-THOM
SUPER-MORR
SUPER-THOM
RICO-MORR
RICO-THOM
Liquid cloud mass
-0.59 %
+32.2 %
+146 %
+169 %
-5.21 %
+44.0 %
Ice mass
+12.5 %
-5.61 %
+116 %
+29.7 %
n/a
n/a
Rain mass
-0.67 %
-62.6 %
-93.7 %
-51.6 %
-100 %
-100 %
Snow mass
+1.37 %
+13.8 %
-33.5 %
+109 %
n/a
n/a
Graupel mass
-4.60 %
-29.9 %
-19.1 %
-36.7 %
n/a
n/a
n/a = not applicable
We present the new result that variability in aerosol response due
to the choice of microphysics scheme differs not only between schemes,
but the inter-scheme variability also differs between cases of
convection. The maximum relative difference in the domain-mean
hydrometeor profiles between polluted and pristine CDNC values for
each of the model configurations is summarised in
Table . It is clear that both the
magnitude and the sign of the response of each hydrometeor class to
CDNC differ strongly not only between microphysics schemes, but
also between cases. (Note that Table
shows relative amounts and that the absolute difference in
response to CDNC between each of the schemes and cases can also
vary significantly). Whilst it is not surprising that the different
cases of convection differ in their hydrometeor development and in
their response to polluted conditions, it is worth noting the
magnitude of and variation in the difference in response. A body of
literature uses idealised model configurations to investigate
storm system response to aerosol loading
e.g. and to compare microphysics schemes
e.g.. Our results highlight that the
storm system response in such a model configuration may not be
representative of the response over larger spatiotemporal scales,
supporting similar findings of larger-scale feedbacks and
life-cycle-dependent responses in idealised and real-data studies of
aerosol–convection interactions.
We note that the vertical resolution used in this study is
relatively coarse and that a horizontal grid length of
4 km is at the limit of what may be considered as
“convection-permitting” . However, we use this
grid spacing for consistency with a previous study in which 10 and
4 km grid lengths were shown to be sufficient to reproduce
storm characteristics and aerosol–convection interactions in the
Congo basin . Previous studies have indicated
sensitivity of convection to horizontal grid spacing
e.g. and also that the sensitivity
to grid length can vary with microphysics scheme
, although idealised ensemble studies of
response to aerosol have shown that differences between polluted
and pristine conditions were similar in simulations using
horizontal grid lengths of 4, 2 and 0.5 km, respectively,
and were also relatively robust to domain size
.
An important factor in our set-up is that we use the same values of
prescribed CDNC in all of our cases. Whilst the literature also
shows widely varying response to aerosol, especially between bin
and bulk schemes (in which even the sign of the response may differ),
showed in idealised supercell simulations using
15 CCN concentrations and four environmental soundings that changes in
cold pool characteristics with CCN were nonmonotonic and dependent
on the environmental conditions. Therefore our use of the same CDNC
values in multiple types of convection helps to minimise
uncertainty due to nonmonotonic behaviour. However, considering the
results of , we note that a caveat of the present
study (and indeed of the majority of existing studies) is that the
absolute values of the cloud system and precipitation response to
aerosol identified here may only hold for the CDNC values used in
our study.
We find that the autoconversion representation alone is sufficient
to explain most of the differences between microphysics schemes in
the shallow cumulus case both in terms of their representation of
cloud and precipitation consistent with and
in terms of their response to CDNC. The dominant hydrometeor
difference between the microphysics schemes in the RICO simulations
occurs in the rain – a different result from both the Congo basin
configuration (in which the dominant difference occurs in the liquid
cloud) and the idealised supercell configuration (in which the
dominant difference occurs in the graupel). We also find that
autoconversion of cloud droplets to rain is the mechanism that
prevents the formation (or persistence) of liquid-phase cloud in
the south of the domain in the Congo basin simulations using the
Morrison scheme. This is in agreement with the study of
, who found in idealised supercell simulations
using the Morrison bulk microphysics scheme with a variable shape
parameter for the raindrop size distribution that autoconversion
rates decreased under CCN loading. The importance of autoconversion
representation was shown by , who demonstrated that
the rates predicted by the autoconversion formulae used in bulk
schemes differ by orders of magnitude. Modelling studies and
observations from RICO have found that warm-rain formation can be
explained by the observed aerosol distribution
. In the context of our findings, this suggests
that an accurate description of the autoconversion process in
warm-rain regimes is fundamental not only to a realistic
representation of cloud and precipitation, but also to its response
to varying aerosol concentrations.
We caution that care should be taken when using autoconversion
schemes in regimes other than those for which they were originally
developed, such as the use of the scheme
for deep convective cases. Although not the focus of the present
study, those interested in testing and improving autoconversion
schemes could do so by calculating the mean volume radius and
thereby the height of first raindrop formation through knowledge of
the 13 to 14 µm critical radius for raindrop production
. Similarly, comparison
of results from bulk models to those from bin models
e.g. can also be a valuable tool
for testing schemes. Based on the limited set of cases in our
study, we would not be justified in recommending one of the
autoconversion schemes over the other. Moreover, because there are
so many competing processes besides autoconversion, including
a number of microphysical and dynamical processes, it could be
misleading to claim that one scheme is better than the other
based solely on bulk comparison with observations from a few cases. For
those interested in testing and evaluating the autoconversion
schemes, we suggest that the best approach would be to perform
offline testing based on detailed in situ observations and
calculations, as was done by e.g. , who tested the
autoconversion scheme in such a manner.
Our results (which are shown to hold across multiple cloud types
and types of simulation) have important implications not only for
cloud-resolving simulations, but also for the global modelling
community. Most significant, perhaps, is the radiative impact
which could arise when such major differences occur in the ice
phase. Our Congo simulations illustrate just how large this
uncertainty may be, and our tests using a bin scheme show that this
is not purely an artefact of the bulk microphysics schemes
used. Further, that uncertainties due to the choice of microphysics
scheme dominate any aerosol response within a given scheme has
implications for global modelling studies of aerosol indirect
effects e.g.. Once again, this
highlights the continuing need of our community for tight
observational constraints on cloud and precipitation processes and
their response to aerosol, as well as for ongoing parameterisation
development to allow these processes to be accurately represented
in large domain (or global), long-term simulations.