(This blog originally appeared as an article in theWeather magazine of The Weather Club, altered with permission of RMetS)

“Window or aisle madam?” There is only one answer to this question for your average meteorologist/climate scientist. When I have to fly (and I avoid it as much as possible these days), I always opt for the window seat. On long haul to the US I may get to see sea ice, whilst on flights to Niger for a Saharan dust project it is the surprising colour variations of the world’s largest sand desert. On most flights however, it is the clouds that hold my attention. I alternate between thinking how beautiful is the multi-layered scattering of bubbly white water droplets, and despairing of ever being able to represent such an important component of our atmosphere in global climate models.

Clouds affect climate in several ways. Firstly, they are responsible for moving energy and water around our earth-atmosphere system – heat is released inside clouds when water vapour condenses to form water droplets (it’s the droplets we actually see). Secondly some (but not all) clouds lead to rainfall or other precipitation. Finally, clouds interact with both sunlight and the long-wavelength radiation emitted by the earth to space. Different types and thicknesses of clouds do this in different ways; for example high cirrus clouds mainly act like a blanket keeping the energy emitted by the earth from escaping to space. On the other hand, thick stratus clouds reflect lots of sunlight back to space preventing it from reaching the earth’s surface. Clouds change in response to surface and atmospheric temperature changes, changes in humidity and changes in atmospheric aerosol concentration (water vapour has to condense on to these tiny particles in order to form cloud droplets). Thus cloud changes are part of the atmosphere’s response to climate change, as well as being crucial for our everyday weather.

My main problem with clouds is their variability, from minute to minute and from place to place. The processes responsible for creating clouds (the condensation of water vapour onto tiny specks of “dust” (aerosol), the rising and sinking of air parcels and many others) take place on very small space and time scales. Computer models of weather and climate represent the quantities that describe our climate system (for example temperature, pressure, wind, humidity) on a stack of grids laid out across the earth. The equations that predict how these things change with time are then calculated at each of the crossing points of these grids. Any processes or events that happen between the crossing points can only be represented in some average way in these equations, and this can lead to uncertainty in our predictions. Climate models typically have grid points 10s to 100s of kilometres apart.  Even the best computer models for making weather predictions might have a resolution of 1 km (and due to computational expense this is only possible if you are considering a small area of the globe, e.g. the UK and a few days ahead at most). Most cloud processes happen on smaller scales even than this. The method of taking these into account somehow is called “parameterisation”, and considerable research effort is spent determining the best way to do this (see also the excellent PLOS blog about parameterisation and much much more by Tamsin Edwards) . We use field measurements or much more detailed models of the processes happening inside an individual cloud to develop relationships between the key quantities and those that are used in the equations of a weather or climate prediction model. Inevitably however we have to make simplifications or generalisations in some places and this can lead to diversity in how different models represent cloud and other processes and to differences in the resulting predictions.

Despite their variability and complexity, I agree with the founder of the Cloud Appreciation Society,  Gavin Pretor-Pinney, who wrote also in theWeather magazine, “life would be immeasurably poorer without them (clouds)”. My journeys by plane would however be somewhat less emotionally charged!