The nature of ice-nucleating particles affects the radiative 1 properties of tropical convective cloud systems 2 3

14 Convective cloud systems in the maritime tropics play a critical role in global climate, but accurately representing 15 aerosol interactions within these clouds persists as a major challenge for weather and climate modelling. We quantify 16 the effect of ice-nucleating particles (INP) on the radiative properties of a complex Tropical Atlantic deep convective 17 cloud field using a regional model with an advanced double-moment microphysics scheme. Our results show that the 18 domain-mean daylight outgoing radiation varies by up to 18 W m depending on the bioand physico-chemical 19 properties of INP. The key distinction between different INPs is the temperature dependence of ice formation, which 20 alters the vertical distribution of cloud microphysical processes. The controlling effect of the INP temperature 21 dependence is substantial even in the presence of secondary ice production, and the effects of secondary ice 22 formation depend strongly on the nature of the INP. Our results have implications for climate model simulations of 23 tropical clouds and radiation, which currently do not consider a link between INP particle type and ice water content. 24 The results also provide a challenge to the INP measurement community, since we demonstrate that INP 25 concentration measurements are required over the full mixed-phase temperature regime, which covers around 10 26 orders of magnitude in INP concentration. 27 28 https://doi.org/10.5194/acp-2020-571 Preprint. Discussion started: 8 July 2020 c © Author(s) 2020. CC BY 4.0 License.


29
Deep convective clouds are important drivers of local, regional and global climate and weather (Arakawa, 2004;30 Lohmann et al., 2016). They produce substantial precipitation (Arakawa, 2004) and the associated phase changes 31 release latent heat that helps to drive global atmospheric circulation (Fan et al., 2012). Convective clouds have a direct 32 impact on climate through interactions with incoming shortwave and outgoing longwave radiation (Lohmann et al., 33 where is the cloud fraction of simulation r and ∆ is the change in outgoing radiation from cloudy areas only 199 between simulations (s -r). The cloud fraction contribution (∆ ) is calculated using Eq.
(2). 200 Where , is the mean outgoing radiation from cloudy regions in simulation r and , is the mean outgoing 202 radiation from clear sky regions in simulation r and ∆ is the difference in cloud fraction between simulations s and r 203 (s-r). There is interaction between the outgoing radiation from cloudy regions and cloud fraction changes (∆ ) 204 which is calculated in Eq (3). 205 Where ∆ is the change in mean outgoing radiation from clear sky areas between simulations s and r and is 207 the cloud fraction of simulation s. 208 The total outgoing radiation difference between simulations s and r (∆ − ) is therefore as shown in Eq. (4). 209 The interaction term ∆ was found to be negligible and was therefore ignored for the purpose of this paper. When analysing the simulation output, cloudy grid boxes were classed as those containing more than 10 -5 kg kg -1 227 condensed water from cloud droplets, ice crystals, graupel and snow. Rain was not included to ensure analysis did not 228 include areas below cloud base. Other cloud thresholds were tested and found to have no notable effect on the 229 results. For cloud categorisation into low, mid and high clouds, model vertical columns containing cloudy grid boxes 230 were categorised by cloud altitude. Low cloud occurs below 4km, mid cloud between 4 and 9 km and high cloud above 231 two well-defined peaks in cloud base heights (not shown) and roughly correspond to the beginning of the 235 heterogeneous and homogeneous freezing regions, respectively. 236

The observed case
237 MODIS visible images of the 21 st August 2015 are shown in Fig. 2 (a, b) alongside a time series of snapshots of the TOA 238 outgoing longwave radiation in our one of our simulations (c -h). The simulated cloud field has more cloud-free areas 239 than the satellite images but in general produces clouds similar to those shown in the satellite image and in 240 approximately the correct location. Overall the simulations produce a complex and realistic cloud field. Snapshots of 241 the simulated model TOA outgoing shortwave radiation are shown in Fig. A2. 242 In-situ measurements of cloud and aerosol properties were made using the UK FAAM Bae-146 research aircraft, which 243 was flown from Praia, Cape Verde Islands. An extensive suite of in-situ aerosol and cloud particle instruments were 244 operated onboard the aircraft and are described in detail in Lloyd et al. (2019). The aircraft penetrated the growing 245 convective clouds at a range of altitudes from just below the freezing level up to -20°C. In order to show that the 246 model reproduces the observed conditions, the observational data were compared to the conditions in modelled 247 clouds of similar size to those the aircraft flew in (10 -150 km 2 ) where a comparison was thought appropriate. measurements from the aircraft were selected using the same total water content threshold as for the model data 250 (10 -5 kg kg -1 ). Note that observational data only samples clouds along the 1D flight path, while model results include all 251 grid points inside the selected clouds. 252 The vertical wind and cloud droplet and ice number concentrations are shown Fig. A3. The vertical wind speeds from 253 the model and aircraft measurements agree well (Fig. A3a). The aircraft data exhibit less measurements of vertical 254 wind speeds above 10 m s -1 but that is expected since the aircraft was purposefully not flow in very high updraft 255 speeds. The aircraft cloud droplet number concentration (CDNC), measured using a DMT cloud droplet probe, falls 256 predominantly in the regions of parameter space most highly populated by model data when plotted against vertical 257 wind speed (Fig. A3b). Note that the simulated points in Figure A1b represent values of CDNC and updraft speed in all 258 cloudy gridboxes, not just those at cloud base. The updraft speed is collocated with CDNC and therefore does not 259 necessarily represent the updraft speed at which the cloud droplets were activated. The higher CDNC values exhibited 260 in the model data may be due to the higher updraft speeds which were not measured by the aircraft. The observed 261 ICNC was derived from measurements using the DMT Cloud Imaging Probes (CIP-15 and CIP-100) and the SPEC 262 Stereoscopic optical array probe covering a size range from 10 to 6200 µm using the SODA2 processing code to 263 reconstruct ice particle images that are fully contained within the probe sample volume. Because of uncertainties in 264 the optical array probe sample volume for very small images, only ice particle images greater than 100 µm were 265 included. The aircraft ICNC fall almost entirely within the range of the model values (Fig. A3c). 266 267 the presence of SIP. This is consistent with a comparatively small change in TOA radiation when SIP is active relative to 291 when it is inactive ( Fig. 3b and 3c) (we discuss the role of SIP in more detail in Sect. 3.5). 292

Results
The slope of the INP parameterisation is a key determinant of the outgoing radiation. There is a statistically significant 293 correlation between INP parameterisation slope and total TOA outgoing radiation (r 2 = 0.75, p < 0.01, n = 10) (Fig. 3c). 294 Changes in outgoing radiation due to the presence of INP are caused by a combination of changes to the outgoing 295 radiation from cloudy regions, caused by changes in cloud structure and microphysical properties, and changes to 296 domain cloud fraction, whose contributions to the total radiative difference are shown in Fig. 3a (left and centre). In 297 order to appreciate the reasons for these trends, we will now take a closer look at the effect of INP on outgoing 298 radiation from cloudy regions only, domain cloud fraction and cloud type. 299

300
Here we discuss the changes in outgoing radiation from cloudy regions only due to INP parameterisation choice. 301 Daytime outgoing radiation from cloudy regions increases due to INP for all but one INP parameterisation (Fig. 4a). 302 The absolute change in outgoing radiation from cloudy regions is between -0.8 (D10) and +28.1 (A13) W m -2 , and the 303 larger values are a result of large increases in reflected shortwave (up to +37.2 W m -2 ) and relatively moderate 304 https://doi.org/10.5194/acp-2020-571 Preprint. Discussion started: 8 July 2020 c Author(s) 2020. CC BY 4.0 License. decreases in outgoing longwave radiation (up to -11.1 W m -2 ) from cloudy regions. The above absolute changes in 305 outgoing radiation from cloudy regions contribute between -0.7 and +11.4 W m -2 to the domain-mean change in 306 outgoing radiation due to the presence of INP (Fig. 3a, cloudy regions contribution). 307 The enhancement of outgoing radiation from cloudy regions due to INP is caused primarily by increases in cloud 308 condensate relative to the NoINP simulation (Fig. 4b). When INP are included in a simulation, snow and cloud droplet 309 water path are enhanced, causing increases in total cloud condensate, despite decreases (in all except A13) in ice 310 crystal water path. Snow, cloud droplets and ice crystals are the hydrometeors that affect outgoing radiation in CASIM 311 and the combined water path of these three species is significantly positively correlated with cloud shortwave 312 reflectivity (r 2 = 0.62, p < 0.01, n = 11) (Fig. 4c). The mechanism for this INP induced increase in cloud condensate and 313 consequently cloud shortwave reflectivity is as follows: When heterogeneous ice nucleation is active, liquid is 314 consumed in the warmer regions of mixed-phase clouds because of increased heterogeneous ice nucleation ( Fig. 1)  315 and SIP (Fig. A5a). The resultant additional ice crystals in mixed-phase regions facilitate riming causing increases in 316 snow and graupel (Fig. A5c, d), increasing snow water path and reflectivity in mixed-phase and ice clouds. At the same 317 time, the enhanced production of relatively heavy snow and graupel increases precipitation which on evaporation 318 below 4 km, reduces out-of-cloud temperature and increases relative humidity (Fig. A6a, b). This leads to increases in 319 water path in low-level liquid clouds and thus an enhancement in their shortwave reflectivity. 320 However, increases in total cloud condensate alone cannot account for the differences in outgoing radiation from 321 cloudy regions between simulations using different INP parameterisations, which are caused by a combination of 322 cloud microphysical responses. We find that outgoing radiation from cloudy regions is significantly negatively 323 correlated with INP parameterisation slope (r 2 = 0.63, p < 0.01, n = 10) (Fig. 5a), i.e. simulations using a steep INP 324 parameterisation have a higher outgoing radiation from cloudy regions. This result makes sense when we consider the 325 relationships between INP parameterisation slope and a multitude of cloud microphysical properties affecting cloud 326 radiative properties. In particular, a steep INP parameterisation results in a mixed-phase cloud region characterised by 327 a higher ice crystal water path aloft (r 2 = 0.80, p < 0.01, n = 10; Fig. 5b) and higher cloud droplet number 328 concentrations at the bottom of the mixed-phase region (r 2 = 0.89, p < 0.01, n = 10; Fig. 5c) when compared to 329 shallower parameterisations. A steeper INP parameterisation slope allows increased transport of liquid to upper cloud 330 levels due to lower rates of heterogeneous freezing (Fig. 1) and SIP at high temperatures (Fig. A5a). This, combined 331 with higher INP concentrations at low temperatures (Fig. 1), increases ICNC at upper mixed-phase altitudes, as well as 332 https://doi.org/10.5194/acp-2020-571 Preprint. Discussion started: 8 July 2020 c Author(s) 2020. CC BY 4.0 License. 361 Our results show that the INP parameterisation affects the properties and spatial extent of cirrus anvils. We define 362 cirrus anvils to be regions where cloud is present above 9 km only (further details available in Sect. 2.1.4). 2D aerial 363 images of cloud categorisation (Fig. 7a-f) show well-defined regions of anvil cloud (light blue -H) surrounding a large 364 convective system containing clouds at a range of altitudes from <4 km to >9 km. There are clearly differences in the 365 extent and position of cloud categories between simulations (Fig. 7a -f). 366

Effect of INP and INP parameterisation on cirrus anvils
The presence of INP reduces convective anvil extent by between 2.1 and 4.1% of the domain area depending on the 367 choice of INP parameterisation (Fig. 7 g), corresponding to a decrease in anvil cloud of between 22 and 53% relative to 368 the NoINP simulation (not shown). The reduction in anvil extent in the presence of INP is caused by increased liquid 369 consumption at all mixed-phase levels, due to heterogeneous freezing, enhanced SIP and increased graupel and snow 370 production, reducing the availability of cloud droplets for homogeneous freezing (Fig. A5b), reducing ICNC at cloud-371 top, and reducing cloud anvil extent (Fig. 7g). 372 Reductions in anvil extent caused by INP are somewhat offset by the overall increases in cloud fraction across the 373 domain (Fig. 7g). However, it is possible that the effect of INP and INP parameterisation choice on anvil cloud fraction, 374 and the contribution of anvil cloud to overall cloud fraction and radiative changes, would become larger with a longer 375 analysis period. This is because detrained convective anvils can persist longer in the atmosphere than the convective 376 core that creates them (Luo and Rossow, 2004;Mace et al., 2006), but this is beyond the scope of the current study. 377

378
It has been argued that primary ice particle production rate is unimportant for convective cloud properties when We find that the microphysical and radiative properties of the cloud field depend strongly on the properties of the INP 383 even with SIP (Hallett-Mossop process) occurs. Furthermore, the effect of including SIP on daylight domain-mean 384 outgoing radiation varies between -2.0 W m -2 and +6.6 W m -2 (Fig. 3b), showing that SIP has a smaller effect than the 385 The effect of SIP on the radiative properties of the cloud field is dependent on INP parameterisation choice, both in 388 magnitude and sign of change (Fig. 3b). SIP makes the clouds more reflective independent of the chosen 389 parameterisation (Fig. 3b, cloudy regions contribution) due to increases in snow and cloud droplet water path. N12 390 and A13 have the largest overall radiative response to SIP because changes to the radiative forcing from cloudy 391 regions and cloud fraction contributions act to increase outgoing radiation (Fig. 3b). However, the cloud fraction 392 response to SIP is opposite for C86, M92 and D10 meaning the cloudy regions and cloud fraction contributions act in 393 opposite directions, reducing the total radiative forcing. 394 The different response of the domain cloud fraction to the presence of SIP is caused by substantial variation between 395 simulations in the anvil cloud extent (Fig. 7h), from an increase of 10% (+0.9% of the domain area) in N12 to a 396 decrease of 40% (-3.6% of the domain area) in M92 (Fig. 7h). These non-uniform changes in cloud fraction and 397 outgoing radiation can be explained by differences in the response of cloud freezing profiles to SIP due to variations in 398 INP parameterisation slope. For all INP parameterisations, SIP reduces the availability of liquid at higher altitudes. For 399 shallower parameterisations such as M92 this causes a reduction in the amount of cloud droplets reaching the 400 homogeneous freezing regime and thereby reduces ICNC and cloud anvil spatial extent. However, in simulations using 401 a steep parameterisation, almost all available droplets are frozen heterogeneously before they reach the 402 homogeneous regime (see reduced homogeneous ice production rates in N12 and A13 in Fig. A5b). Therefore, in 403 simulations using a steeper parameterisation, such as N12, a reduction in liquid availability due to SIP occurs at the 404 top of the heterogeneous freezing regime, reducing the availability of liquid for riming, causing a reduction in frozen 405 hydrometeor size at high altitudes, a reduction in hydrometeor sedimentation and an increase in anvil extent. accurately representing ice water content in climate models and linking this ice water content to ice-nucleating 423 particle type. 424 Another limitation of the SOCRATES radiation code is its lack of consideration of rain and graupel particles. The effects 425 of these hydrometeors are expected to be less than that of ice, snow and cloud droplets as they precipitate faster and 426 therefore have a shorter lifetime. Furthermore, the effect of graupel on the tropical longwave radiative effect has 427 been found to be negligible and dwarfed by that of snow (Chen et al., 2018). The global radiative effect of rain has also 428 been found to be small in the vast majority of cases even at high temporal and spatial resolution (Hill et al., 2018). The 429 effect of the incorporation of these hydrometeors into radiative transfer parameterisations should however be tested 430 in future studies. 431 The use of both aerosol-dependent (D10, N12, A13) and solely-temperature dependent (C86, M92) parameterisations 432 in this study means that we have examined the radiative sensitivity of a complex cloud field to a larger variety of INP 433 parameterisations used in weather and climate models than if we had exclusively used parameterisations that 434 consider aerosol concentration. However, this experimental design has limitations. For example, due to the lack of 435 aerosol dependence of the C86 and M92 schemes a 'presumed' dust concentration is implicitly present in these two 436 cases and remains uniform throughout the simulation period. The effect of INP parameterisation choice on convective 437 cloud field properties should also be examined with the inclusion of aerosol scavenging but this was beyond the scope 438 of this study.