Dr Rachel Sansom

Profile

I joined ICAS from a background in mathematical physics that got me interested in modeling systems, and a desire to work in environmental sciences. My PhD used Gaussian process emulation to approximate relationships between cloud-controlling factors and cloud properties in low, shallow clouds based on large-eddy simulations. My case studies primarily focused on the large stratoculumus cloud decks that sometimes form off the coast of California, which have a high albedo and hence a high cloud radiative effect. I used response surfaces from GP emulators to visualise how free tropospheric humidity and temperature affect cloud liquid water path and cloud fraction, which showed two distinct regimes of behaviour. One of thick cloud with high cloud fraction where the above-cloud air was warm and moist and one of thin, patchy clouds where the above-cloud air was cool and dry. This follows cloud-top entrainment instability theory, but unlike previous studies we could visually inspect the transition between these two regimes. We also showed that the model’s internal variability can be approximated and accounted for in the emulation method by running initial-condition ensembles.

I then created a perturbed parameter ensemble of stratocumulus-to-cumulus transitions and used GP emulation to analyse the key factors in the transition mechanism. We again found that there were two regimes. In most cases, the lower tropospheric stability is the dominant factor controlling the transition, however, at very low aerosol concentrations the aerosol becomes more dominant, via drizzle depletion, and causes a faster transition. Using response surfaces of a 6-dimensional parameter space allowed us to visualise the joint effects between factors and the nonlinearities of aerosol-cloud interactions.

I’m now using perturbed parameter ensembles and GP emulation as part of two research projects, CLOSURE (grant no. NE/W001713/1) and FORCeS (grant no. 821205). In CLOSURE, I am using emulators to improve approximations of cloud droplet numbers in the UM as well as computational efficiency and in FORCeS, we are looking to constrain aerosol radiative forcing estimates.

Research interests

My research interests focus on aerosol-cloud interactions in low, shallow clouds, particularly in identifying nonlinearities and how they can be represented within models. I have found that perturbed parameter ensembles and Gaussian process emulation are extremely valuable tools in exploring joint interactions between parameters and outputs.

Research groups and institutes

• Institute for Climate and Atmospheric Science
• Atmospheric Chemistry and Aerosols