Understanding Aerosol Impacts on Tropical Land–Sea-Breeze Convection Using a Statistical Emulator Approach

Land-sea breeze circulations are one of the most ubiquitous mesoscale flow regimes in tropical coastal regions. The leading edge of these circulations serves as a persistent boundary layer forcing mechanism for convective initiation and hence for associated aerosol redistribution and transport. However, forecasting land-sea breeze convection is challenging due to uncertainties in the initial conditions, as well as the covariance and interaction of various meteorological and surface parameters. Furthermore, tropical coastal regions are under continuous population growth and increasing anthropogenic activity. As a result, it is important to identify the response of tropical land-sea breeze convection and associated aerosol redistribution via convective processes to different environmental parameters, including those that may be altered by anthropogenic activity in these already populated coastal regions. The goals of this research are, therefore, to (a) identify the relative importance of different environmental parameters and their interactions to the variance in sea breeze convection, and (b) to assess whether the contributions from these multi-dimensional sensitivities are different under varying aerosol conditions.

Our research goals have been achieved through the combined use of idealized cloud-resolving model simulations and an advanced multivariate sensitivity analysis technique. In order to examine the ten-dimensional sensitivity of land-sea breeze convective characteristics, an ensemble of 130 initial conditions for tropical land-sea breeze simulations has been designed by simultaneously perturbing six atmospheric and four surface parameters. Using the Regional Atmospheric Modeling System coupled to an interactive land-surface model, 130 pairs of tropical land-sea breeze simulations have been performed in low- (clean) and high (polluted) aerosol environments. For each convective output of interest, a statistical emulator has been constructed to map the relationship between the individual input environmental parameters and the output responses over the ten-dimensional uncertainty space.

Finally, the environmental parameters that lead to the largest changes in updraft speeds and the redistribution of microphysically active aerosols have been identified through the variance-based sensitivity analysis. The changes in the relative contributions of the environmental parameters to updraft speeds and aerosol redistribution as a function of aerosol loading have been assessed. The results of this analysis will be presented, and the mechanisms responsible in each environment will be discussed.