Shallow marine mixed-phase clouds are important for the Earth's radiative balance, but modelling their formation and dynamics is challenging. These clouds depend on boundary layer turbulence and cloud top radiative cooling, which is related to the cloud phase. The fraction of frozen droplets depends on the availability of suitable ice-nucleating particles (INPs), which initiate droplet freezing. While mineral dust is the dominating INP type in most regions, high-latitude boundary layer clouds can be dependent on local marine INP emissions, which are often related to biogenic sources including phytoplankton. Here we use high resolution large eddy simulations to examine the potential effects of marine emissions on boundary layer INP concentrations and their effects on clouds. Surface emissions have a direct effect on INP concentration in a typical well-mixed boundary layer whereas a steep inversion can block the import of background INPs from the free troposphere. The importance of the marine source depends on the background INP concentration, so that marine INP emissions become more important with lower background INP concentrations. For the INP budget it is also important to account for INP recycling. Finally, with the high-resolution model we show how ice nucleation hotspots and high INP concentrations are focused on updraught regions. Our results show that marine INP emissions contribute directly to the boundary layer INP budget and therefore have an influence on mixed-phase clouds.
Stratocumulus clouds are shallow and thin clouds that cover large parts of the oceans and for this reason they have a significant effect on the radiative balance
Heterogeneous ice formation (or nucleation) means that a solid seed called an ice-nucleating particle (INP) is needed for the ice formation
Models based on the classical nucleation theory (e.g.
An INP in a stochastic CNT-based model means a particle that carries the nucleating substrate but an INP in typical observations is a particle that initiates droplet freezing at set conditions inside the instrument
The most important INPs for shallow boundary layer clouds include dust and biogenic particles
The current view is that marine INPs are emitted as primary particles as a part of sea spray aerosol (SSA). At moderate wind speeds, SSA is produced mainly by bubbles bursting at the sea surface
The details of the marine INPs related to their origin, emission rates and ice nucleation properties are still highly unclear. Nevertheless, there are a few large-scale studies exploring the potential importance of marine INPs on shallow clouds (e.g.
In this study we explored the potential effects of marine INPs on mixed-phase boundary layer clouds by using UCLALES-SALSA, which is a cloud resolving large eddy simulator (LES) coupled with detailed aerosol-cloud-ice microphysics
Previously,
The ISDAC observations are described in detail by
Different cloud types and conditions were seen during ISDAC, of which
We made one modification to the initial temperature and humidity profiles used in the previous case studies. The initial temperature and humidity profiles represented a decoupled marine boundary layer but eventually the boundary layer became coupled in the model simulations
Current simulations are made with the UCLALES-SALSA. This model is based on the commonly used UCLALES
Aerosol, cloud droplet and ice particle chemical composition (here just water, dust and sulfate) and size distributions are described using a sectional approach based on dry particle size bins. Water is the substance that partitions between vapour and condensed phases, dust is the insoluble ice-nucleating material and sulfate is a soluble substance. The model has just one species (dust) with ice nucleation ability, so it is used to describe both background mineral dust and marine biogenic emissions. Although the species is called dust it can be used to represent any ice-nucleating material by adjusting its ice nucleation parameters, such as the contact angle. Details of the ice nucleation scheme are given later in this section. Water and dust physical properties had default values but sulfate density, molecular weight and dissociation factor were set to 1780
Water vapour partitioning is based on diffusion-limited non-equilibrium droplet or ice crystal growth, except that the equilibrium water content is assumed for aerosol when relative humidity (RH) is less than 98 %. For the supersaturated regions, the non-equilibrium droplet growth determines cloud activation, which takes place when aerosol wet size exceeds the critical droplet size. Cloud activation means that the activated aerosol is moved from the aerosol bin to the corresponding cloud bin with matching dry particle size. Cloud (or aerosol) droplet freezing is modelled based on a stochastic (time-dependent) immersion freezing parameterization described below. By default, the freezing is limited to cloud droplets, but the effect of allowing aerosol freezing (interstitial and those outside clouds) will be examined in Sect.
The initial bimodal ammonium bisulfate (called sulfate in SALSA) aerosol size distribution from
Initial aerosol size distribution for the ISDAC case (black line, left axis) and parameterized sea spray aerosol (SSA) emission flux for 6
Here we used an upgraded version of UCLALES-SALSA where size-dependent SSA emissions are parameterized as a function of the domain mean wind speed at the height of 10 m and sea surface temperature (here constant 271.15 K). The parameterization is valid for open ocean, so we ignored the effects of sea ice and nearby coasts in these simulations. For the dry particle size range 0.020–1
Different ice nucleation modes were implemented as described by
Since we are limited to two externally mixed particle populations (one for the soluble sulfate particles and one for the INPs), marine and background INPs need to be described with the same INP population. This means that they will have the same chemical composition (dust–sulfate) and ice nucleation parameters (contact angle). Because we cannot adjust ice nucleation parameters independently, we will keep those fixed and adjust the emissions and initial concentrations of the marine and background INPs, respectively. Although their size distributions could be adjusted, we will use the background and SSA size distributions (Fig.
Our strategy for adjusting the fractions of INPs in the background aerosol and SSA flux is based on the fact that some amount of INPs are needed to maintain the mixed-phase cloud and that marine INP emissions may have a dominant role in certain conditions (not necessarily ISDAC). In the following simulations we will examine the potential effects of marine INPs on mixed-phase clouds with wide range of background INP concentrations. Our previous simulations show that ICNC in the order of 4
All simulations have the same initial background aerosol size distribution (Fig.
Although the ice nucleation scheme described above considers the INP concentration as an adjustable parameter, we can examine if at least the maximum initial background INP concentration is reasonable based on observations from ISDAC and elsewhere.
Comparing our marine INP fluxes with observations or other simulations representing continuous emissions is limited by the fact that the fluxes in our simulations start after the 1h spin-up and then have limited time for an impact. Therefore, our fluxes must be significantly higher than any continuous emissions. Based on the simulated impacts on clouds (shown in the next section), the current INP emissions seem to be reasonably high as they alone can maintain the mixed-phase cloud. On the other hand, the emissions are low enough so that the contribution is small in the presence of high concentrations of background INPs.
As the first step we made eight simulations where marine INP emissions were either on or off and the initial background aerosol INP number concentration had four different values. Marine INP emissions are specified as a fraction of the SSA flux, and here emissions on and off mean fractions 0.005 and 0.0, respectively. Initial background INP number concentrations are specified as a fraction of the initial aerosol, and here the cases are called zero (fraction is 0.0), low (0.00001), medium (0.00005), and high (0.0001). Background INP fractions were selected so that the simulations without marine INP emissions covered the range from an ice-free case up to a cloud that is becoming mostly glaciated. When marine INP emissions are switched on, the fraction of INPs in the SSA is high enough to have an impact on clouds. Simulation time was set to 24 h including a 1 h spin-up for SSA emissions and a 2 h spin-up for ice microphysics. Because most adjustments take place during the first 10 h and the trends are steady after 12 h, we will focus on the first 16 h.
Results from our simulations are shown in Fig.
Time series of cloud base and top heights, ice crystal number concentration, and ice (IWP) and liquid (LWP) water paths (from top to bottom) from the eight model simulations with different background (BKGD) aerosol INP concentrations (zero, low, medium, and high) and marine INP emissions switched on (solid lines) or off (dashed lines).
The largest differences between simulations with marine INP emissions on or off are seen with the lowest background INP concentrations. This means that marine INP emissions become more important with decreasing background INP concentration. In addition, marine INP emissions alone seem to produce the same final mixed-phase cloud state as having the medium or high INP background without marine INP emissions. There are two simulations, both without marine INP emissions, where the INP concentration is so low that the result is a thick almost purely liquid cloud. The other simulations result in an ice number concentration of about 2000
The cloud starts to glaciate (LWP decreases and IWP increases) rapidly when ice crystal number concentration approaches 3000
Cloud development in these simulations depends mainly on ice crystal number concentration, which is related to the availability of INPs, so here we focus on the INP budget. Figure
The three mechanisms (precipitation, subsidence and surface emissions) affecting on INP mass
Precipitation is the main INP removal mechanism, and it can easily exceed production. In fact, the total INP mass is decreasing in all other simulations except the one without background INPs (Fig.
Changes in the column-integrated INP mass from the initial values. The initial values for the four different background INP concentrations (high, medium, low and zero) are 9.6, 4.8, 1.0 and 0.0
Figure
Horizontally averaged profiles of the main processes acting on INP mass concentrations from simulations with (solid lines) and without (dashed lines) marine INP emissions (high background INP concentration). Diffusion, subsidence, and surface flux contribution are multiplied by a factor of 10 for clarity. Corresponding INP mass concentrations in aerosol, cloud, ice and in total are shown in panel
Figure
Subsidence introduces aerosol particles from the free troposphere (positive values at the cloud top) but at the same time depletes cloud and ice species in the cloud (negative values below the cloud top). This is related to the steep gradient in total INP mass concentration (Fig.
The only clear difference between simulations with and without marine INP emissions is seen in the near surface diffusion and surface emissions rates. Marine INP emissions influence only the first model layer, and sub-grid scale diffusion is the main mechanism transporting particles from the first model layer to the layers above where advection dominates. Diffusion reduces concentration differences within the domain just like advection, but diffusion is significantly weaker and limited to the lowest model layers due to the dependency on eddy diffusivity. Diffusion does not cause INP removal to the sea surface because aerosol sedimentation (includes the effect of particle diffusivity) is disabled in these simulations. A test will be conducted in Sect.
The effect of marine INP emissions can be seen in the total INP mass concentration profiles as an increase near the sea surface. Interestingly, aerosol phase INP mass concentration profiles are similar with or without marine INP emissions. They both show a decreasing trend with height above sea surface, which is typically related to a surface source. In this case, however, ice crystal sedimentation and sublimation near the surface is an additional reason. We will focus on this topic in the next section.
Because ice crystal mass and number concentrations are important for the time evolution of the cloud and the processes are related to the INP budget, we will briefly examine the ice budget. Figure
The four mechanisms (subsidence, water vapour sublimation and deposition, nucleation, and precipitation) affecting ice mass
Figure
Horizontally averaged profiles of the main processes affecting on ice mass
Ice nucleation in these simulations is focused on the cloudy region and especially closer to the cloud top. However, ice nucleation is limited to cloud droplets, so aerosol freezing below and above the cloud and the freezing of interstitial aerosol are prohibited. Just like cloud droplets, a fraction of the aerosol contains an INP immersed in supercooled liquid, so the same immersion freezing parameterization can be used to predict the freezing rates. The effect of aerosol freezing is tested in Sect.
In the above it was shown that advection transports INPs to the cloud top where most of the droplet freezing takes place. However, a closer look at 3D data shows that there is significant horizontal variability in the freezing rates and INP mass mixing ratios, and this variability can be best explained by vertical velocity. This is not a surprise knowing the importance of vertical velocity for cloud activation.
Figure
Vertically integrated ice nucleation rate (the number of primary freezing events per second in a column of air above 1
Higher cloud droplet freezing rates are related to updraughts (positive vertical velocity) and marker colour shows that the updraughts have higher INP mass concentrations. The INP concentrations range from less than 5
Here we examine the effects of the microphysical (aerosol and cloud droplet sedimentation, and immersion freezing of unactivated aerosol) and meteorological (decoupled marine boundary layer) model considerations mentioned above. Figure
Sensitivity tests related to
Allowing aerosol and cloud droplet sedimentation (disabled in the default simulations) has two potential effects on INPs. First, aerosol sedimentation could bring INPs from the free troposphere or remove those from the near surface layer by dry deposition. Second, sedimentation could have an impact on vertical distributions. Cloud droplet sedimentation redistributes cloud water, which influences clouds as explained in
Figure
Meteorological conditions are crucially important for clouds but here we focus on the one that has direct relevance for the vertical transport of marine INP emissions, namely decoupled MBL. The original ISDAC simulations were initialized with a decoupled MBL, which reduces vertical mixing and partially isolates the near surface layer from the rest of the boundary layer
In this study we examined the potential effects of marine ice-nucleating particles (INPs) on shallow mixed-phase clouds by using the large eddy simulator UCLALES-SALSA
Our simulations with UCLALES-SALSA support the previous findings about the importance of INP recycling
Prognostic ice microphysics including explicitly modelled ice nucleation is important for the simulations as this enables feedback between INPs and clouds
Efficient INP recycling, feedback between INP emissions and precipitation removal and the fact that marine INPs are emitted directly to the mixed layer mean that modest marine INP emissions can maintain mixed-phase clouds at least in the conditions used in our simulations. Although significant uncertainties are still related to ambient INP emissions, our simulations support the current view
The source code of UCLALES-SALSA version used in this work is available from
TR designed and performed the model simulations. All authors have contributed to developing the UCLALES-SALSA model. TR prepared the manuscript with contributions from all coauthors.
The contact author has declared that neither they nor their co-authors have any competing interests.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors wish to acknowledge CSC – IT Center for Science, Finland, for computational resources.
This research has been supported by the Academy of Finland (grant nos. 322532 and 309127) and the Horizon 2020 (FORCeS (grant no. 821205)).
This paper was edited by Ken Carslaw and reviewed by two anonymous referees.