Cloudy outlook

C&I Issue 3, 2015

Swirling clouds of dust lifted by hot winds blowing across the Sahara desert and drawn high into the atmosphere may appear to have little to with rainfall patterns in countries thousands of miles away. But research is increasingly revealing the key role that mineral dust from the Earth’s surface plays in the formation and behaviour of clouds. The findings are important because clouds are not only a central component of the water cycle, but also play a crucial role in reflecting solar energy and insulating the Earth’s surface – both of which have a big impact on climate.

‘The average person probably imagines that water evaporates into the atmosphere, condenses and forms rain,’ says John Marsham, an atmospheric scientist at the University of Leeds, UK. ‘This is not incorrect, but clouds do not form spontaneously in clean air. There is always particulate matter in the air, such as sea salt, sulphates that arise from the oxidation of sulphur dioxide from industrial activities, dust from deserts, and so on.’

In the absence of any particulate matter in the air, the water vapour would remain as vapour. However, if tiny particles are present, the vapour condenses upon these. As the particles grow larger they provide a bigger surface for more condensation, forming droplets. These in turn can bump into one another and coalesce into larger droplets, which eventually have sufficient mass to fall as rain. In air that is below the freezing point of water, ice condenses on the aerosol particles; again once enough mass is reached they will fall. If they meet warmer air on the way down, they can melt to produce rainfall.

Human activity generates aerosols directly, soot from chimneys, for example, and also modifies natural aerosol sources. For example, increased atmospheric dust comes from deforestation and overgrazing. For these reasons, there is now a global effort to understand how particulate matter in the atmosphere interacts with water vapour to form clouds and influence their behaviour, and how these aerosol interactions feed into long-term climate models. Further into the future, researchers believe this information may also help to improve the accuracy of shorter-term weather forecasting.

By understanding the interaction of aerosols with the atmosphere to produce cloud and precipitation, it may be possible to track and predict the movement of aerosols and in turn, predict precipitation patterns. However, this would require a degree of atmospheric modelling and computation that is currently unachievable.

While the basic theory of nucleation and the formation of clouds around aerosol particles has been around since the 1930s, interest in the field has been reignited over the past 20 years with the realisation of the importance of clouds in determining the Earth’s ‘energy budget’: the difference between the amount of solar energy that reaches the Earth and how much of that absorbed energy is radiated back into space. Clouds play a key role because they can shield the Earth from sunlight, but also act as a blanket of insulation, helping to retain heat – a phenomenon termed radiative forcing.

Broadly speaking, there are two types of aerosol particle: hygroscopic soluble particles such as sulphates that nucleate liquid droplets; and particles that induce ice formation. Most of the rain that falls in the UK, for example, is derived from ice crystals that form within the cloud, which then fall into warmer air and melt to create rain.

‘The ice phase is much less well understood,’ says Marsham. ‘For ice nucleation, people are now trying to understand the real underlying physical chemistry.’

One such researcher is Daniel Cziczo of Massachusetts Institute of Technology in the US, who has investigated nucleation in cirrus clouds, wispy trails of ice crystals that are formed several kilometres above the Earth’s surface and which are important in radiative forcing. Cziczo has flown into cirrus clouds and collected samples of ice to unveil the core nucleation particles. ‘More than half the nucleation centres are made up of mineral dust,’ he says. ‘We have been everywhere from south of the equator to the northern hemisphere and in all the locations we have investigated, mineral dust is a big player.’

While such dust is largely a natural aerosol, ‘land usage by humans may have put up to twice as much dust into the atmosphere as there would have been otherwise,’ Cziczo says. ‘As we change land usage, we change where these particles arise and how many of them there are. For me, it is a fascinating topic because our lack of understanding of how humans have changed the constitution and concentration of particles in the atmosphere, and especially human-induced particles, is something that really limits our understanding of climate change.’

Precisely which particles make good ice nucleation particles and how efficiently they do so is something that concerns Ben Murray, an aerosol scientist at the University of Leeds, UK. ‘Ice nucleation is critical for the properties of clouds, their lifetime, the distribution of ice crystals and the formation of rain droplets,’ Murray says. ‘There is some pretty robust basic theory, which can be used to predict nucleation of liquid droplets, but for ice nucleation this is lacking and we are still in a situation where it is a much more phenomenological situation. The only way we can determine ice nucleation efficiency of an aerosol particle is to put the particle in water and look for the growth of ice crystals.’

This is what Murray has been doing, with surprising results. ‘We use a cold stage instrument where we place a droplet of water on a non-nucleating surface,’ he explains. ‘We suspend our particles of interest in the droplet and gradually cool it while watching it under the microscope. In this way, we can investigate the huge range of dynamic droplet size, from a few microns to a few millimetres, containing just a few particles to millions. We can quantify the probability of that particle nucleating ice normalised to the surface area of the particle.’

Murray and his colleagues set out to unpick the component of mineral dust that might be responsible for its ice nucleation efficiency. ‘A single particle of desert dust is typically made up of a complex mixture of different minerals, which makes things potentially tricky,’ he says. ‘But in a stroke of luck, the geology department here at Leeds happened to have a very good rock store containing relatively pure samples of the various mineral components that you find in desert dust particles.’

In 2013, the team carried out the controlled ice nucleation experiments using powdered samples of each mineral (Nature, doi:10.1038/nature12278). ‘The big surprise that came out is that clay mineral in its relatively pure state is not a very efficient nucleator,’ says Murray. The team found, however, that one particular mineral – a member of the feldspar family – induced the formation of ice crystals with a surprisingly high efficiency. The mineral, a potassium feldspar called microcline, is only a relatively small component of mineral dust – between about 1 and 10%. ‘So people looking at this issue in the past had largely overlooked it,’ says Murray.  ‘While clay minerals appeared in the past to act as quite efficient ice nucleating particles, in fact it seems that it is the small impurity of feldspar that is actually doing the job.’

Murray has also been examining ice nucleation of agricultural soils, where there is growing evidence that biologically derived material could form efficient nucleation centres.

This is something that has also interested Kim Prather of the University of California, San Diego, US. ‘The standard belief is the more ice you have in a cloud, the more likely you will get precipitation out of it,’ Prather says. ‘Our goal is to catch the first stages of ice forming and find out what exactly the chemical constituents are that the ice is forming on.’

In her analysis of ice particles within clouds, Prather has found ‘lots of biological components such as bacteria and molecules associated with microbial life. In fact, many of the species we find in the ice crystals have these biological chemicals and are not simply dust particles’. The presence of biological matter on the particle appears to make the particle a far more efficient ice nucleator, possibly because the structures have a good geometrical ‘match’ to ice crystals, acting as a template for the crystal nucleation.

Meanwhile, a substantial global effort is under way to develop numerical models of the interactions between aerosols, cloud and precipitation and how these can feed into more comprehensive climate models.

Graham Feingold of the US National Oceanic and Atmospheric Administration (NOAA) undertakes such modelling studies. ‘We are interested in how nucleation in liquid water clouds results in very small cloud droplets a few microns in diameter, and how these then grow into drops with sufficient mass to fall as drizzle and rain,’ he says. Here the aerosol particles require a degree of hygroscopicity – but it need not be much. ‘The first thing is that the particle needs to be substantial. If the particle is large enough, it does not need very much hygroscopic material, so size is very important. However, composition is helpful: if the particle is small but sufficiently hygroscopic it will form droplets.’ Such particles might typically consist of sea salt or sulphates.

Feingold cautions, however, against focusing too closely on aerosol composition and concentration. ‘While many people are giving a lot of attention to particle composition, there is a big overlying issue of cloud dynamics, which is even more important. For example, if a strong updraft develops within the cloud mass, the associated expansion results in cooling, which in turn leads to supersaturation of the water vapour. The velocity of the updraft determines the degree of supersaturation, and this affects nucleation processes. In other words, the influence of a particle’s size and its hygroscopicity cannot be viewed in isolation – it must be viewed within the context of the environment in which it sits, and this environment can change depending on the dynamics of the cloud.’

This is just one small confounding aspect of the system, which makes the modellers’ task highly challenging. ‘We used to think that by adding particles, droplets became smaller so the formation of drizzle was suppressed, resulting in less precipitation and a brighter cloud,’ Feingold says. ‘But what we found through a lot of detailed numerical modelling and field and aircraft work is that when you perturb the system, by adding more aerosol, for example, you set off a chain of reactions not all of which go in the same direction – you get internal feedbacks, which can offset one another so the system is able to absorb some of these perturbations.’

So, for example, while more aerosol results in more but smaller droplets, suppressing the formation of drizzle as predicted, in fact the larger surface area to volume ratio of the droplets causes an increase in the rate of evaporation, offsetting the propensity for the cloud simply to become thicker and brighter, as had previously been thought.

‘The biggest challenge we face in understanding how clouds work is in understanding how small changes in environmental conditions, such as temperature, relative humidity, land type and so on, control the cloud system,’ Feingold says. ‘If we can get the cloud dynamics right, we have a chance of looking more clearly at how aerosols influence clouds. We are making progress, but it is painstakingly slow.’

Nicolas Bellouin, a numerical modeller at MACC – Monitoring Air Composition and Climate – based at the UK’s University of Reading, concurs. ‘To separate the effects of meteorology from the effects of aerosols remains a difficult problem,’ he says. ‘It is a very big challenge to address the huge range of scales involved, from the level of aerosol particles, at microns, to the scale of fields of clouds. Any model should be able to span the range, and that is not the case at the moment. There is a lot of activity to increase the resolution of the models but there is a long way to go and we are limited by computer power.’

John Marsham echoes this view: ‘We have learnt a lot over the last decade or two and we have made progress on all scales but this is a surprisingly big field. It has major implications for energy budgets on Earth’s surface and there is a lot of uncertainty involved. There is still a lot of work to do.’

Simon Hadlington is a freelance science writer based in York, UK

Become an SCI Member to receive events discounts

Join SCI