Measurement and modelling of volatile organic compounds in a tropical rainforest environment

This thesis describes measurements of volatile organic compounds made at the Danum Valley Global Atmosphere Watch station in Sabah, Malaysia using gas chromatography with flame ionization detection. Measurements of 13 VOCs were made between August 2015 and March 2016. A non-negative matrix factorization analysis is performed on the observational data, which identified three main periods within the dataset. A period influenced by the Indonesian biomass burning season characterized by large increases in most VOCs concentrations; background periods where local conditions dominate; and a period influenced by local anthropogenic events where short lived increases in some species were seen. Biomass burning in the region caused a substantial increase in most VOCs at the measurement site between August and October 2015. Comparison of VOC emission ratios with previous studies indicates that the burning in the region is a combination of peat and forest burning. Combining fire maps with air mass trajectories shows that changes in meteorological conditions, not changes in fire activity, drives the variability in VOC concentrations at the measurement site during the biomass burning period. The background periods were dominated by biogenic emissions. The two VOCs identified as being mainly biogenic in origin were isoprene and propene, with isoprene being the dominant VOC during the background periods. Comparison of propene observations with the GEOS-Chem model shows that the model substantially overestimates propene concentrations. This shows that the current MEGAN propene emissions algorithm based on mid-latitude observations is not appropriate for the tropics. There was good agreement between measured isoprene concentrations and MEGAN derived GEOS-Chem model concentrations. Most previous studies have shown the MEGAN overestimates isoprene emissions. No clear seasonality was seen in isoprene concentrations at the Danum Valley measurement site. This is unlike previous studies in the Amazon, where strong seasonal cycles are observed.