Investigating the connectivity between emissions of BVOC and rainfall formation in Amazonia using Genetic Programming

A detailed field experiment measuring turbulent properties, trace gases and BVOCs was carried out from April 2014 to January 2015 within and above a central Amazonian rainforest, with the objective of understanding the role of emissions and reactions of BVOCs, formation and transport of aerosols out of the boundary layer on cloud formation and precipitation triggers. Our measurements show that the rainforest emits sufficient reactive hydrocarbons such as isoprene and monoterpenes to provide precursors of secondary organic aerosols and cloud condensation nuclei, thereby possibly influencing the formation of convective precipitation. To further quantify the connectivity between emissions of BVOCs near the surface and rainfall generation, we will use the technique of Genetic Programing (GP), introduced by Koza (1992), based on the concepts of natural selection and genetics. The technique involves finding a mathematical expression that fits a given set of data, and constructing a population of mathematical models from different combinations of variables, constants and operators. Each model (individual) in the population can be considered as a potential solution to the problem. The higher the fitness of an individual, based on a given metric of fidelity to the existing data, the greater the chance of that model being carried over to the next generation. The advantage of the technique is that it can flexibly incorporate multivariate non-linear relations, with the advantage that the numeric solutions obtained are possible to interpret and check for physical consistency. A number of state variables (for example, surface fluxes, meteorological conditions, boundary layer stability conditions, BVOCs and ozone vertical profiles, etc), representing possible influences on BVOC emissions and their interrelations along the way through secondary organic aerosol and CCN formation to rainfall will be used.