Chemical and physical drivers of the evolution of organic aerosols over forests

Diurnal evolution of organic aerosol over boreal and tropical forests The first research question of this thesis is: how do local surface forcings and large-scale meteorological forcings shape the evolution of organic aerosol over the boreal and tropical forest? This question is dealt with in Chapters 3 and 4 in case studies for the boreal and tropical forest, respectively. To answer this question a modeling tool (MXLCH-SOA) is developed, which represents land surface conditions and dynamical and chemical processes that influence the evolution of organic aerosol (OA) in a balanced way. The novelty of our approach is that it combines the dynamics of a convective boundary layer (BL) with a reduced gas-phase chemistry mechanism and a module for gas/particle-partitioning of semi-volatile organic species. The principles and governing equations of this modeling tool are described in Chapter 2 and in the subsequent chapters the simplified chemical reaction schemes are presented to calculate secondary organic aerosol (SOA) formation from terpenes (Chapter 3 and 4) and from isoprene (Chapter 4). Despite its simplicity, MXLCH-SOA is able to satisfactorily reproduce the main observed characteristics of dynamics, gas-phase chemistry and gas/particle partitioning for the two studied forest ecosystems and it enables us to explain the temporal variability of the concentrations of organic aerosol and its precursors as a function of the various processes. In short, the results show that the diurnal evolution of organic aerosol in a boreal and a tropical forest is the net result of land surface conditions, boundary layer dynamics, chemical transformations and gas/particle partitioning. In the case study for the boreal forest, the entrainment term of the background OA dominates the OA tendency, while in the tropical forest case it is the interplay of several local and large scale processes that shape the diurnal evolution of OA. A sensitivity analysis for the boreal forest case further shows that the OA concentration is sensitive to both volatile organic compound (VOC) emissions and the partitioning of the surface energy budget into a latent and a sensible heat flux. We have identified two regimes, based on which of the two studied land surface drivers dominates: one in which OA is mainly driven by SOA formation from the emitted VOCs and another in which dilution due to entrainment, as driven by the surface energy fluxes, determines the OA concentration. A background OA to fresh SOA ratio is introduced to facilitate the interpretation of this analysis and is used to quantify the contributions of both fresh and background components to the total OA concentration. One main difference between the two case studies is that in the boreal forest entrainment appears to dominate the diurnal cycle, which leads to a decreasing OA concentration during the day, while in the tropical forest the formation of SOA from both isoprene and terpenes leads to increasing OA concentrations during day time. The MXLCH-SOA framework therefore shows the need to represent all these biochemical and physical processes simultaneously in order to understand the diurnal evolution of OA. As the boundary layer dynamics-chemistry system is not a closed system, it is necessary to further study the influence of external forcings on the diurnal evolution of OA, besides the surface forcings. Two types of large-scale meteorological forcings and their effects on OA evolution through their impact on BL dynamics have been studied: subsidence due to the presence of a high pressure system and advection of relatively cool air. In Chapter 3 a theoretical sensitivity analysis is given of OA evolution to subsidence, which is applied to the tropical forest case study in Chapter 4. Subsidence has a rather counter-intuitive effect on OA concentrations: even though it suppresses the growth of the BL and consequently decreases the mixing volume for chemical species, it leads to decreased OA concentrations. The reason for this is that entrainment is strongly enhanced in case of subsidence due to thermodynamic effects, which results in a stronger dilution of OA. This knowledge is applied in the case study for the tropical forest in Chapter 4, since results from a large-scale model show subsiding air motions over the measurement site and surroundings at Borneo. In addition to subsidence, the advection of cool air is needed to reproduce the observed boundary layer dynamics at Borneo: only if subsidence and advection of relatively cool air are accounted for, the observed low BL height can be reconciled with the large observed surface sensible and latent heat flux. This cool air suppresses BL growth and entrainment. Consequently, the aerosol is trapped in a shallower layer, which leads to an increased concentration compared to the case without advection of cooler air. In conclusion, the large-scale meteorological forcings subsidence and advection of cool air have opposing effects on the diurnal evolution of OA, even though both suppress BL growth. These findings show the utility of our method in identifying effects that should be accounted for in large-scale chemistry transport models. The second research question is whether recently discovered pathways of isoprene chemistry are the key to closing the gap between measured and modeled organic aerosol concentrations in tropical forests and other high isoprene environments. To address this issue, several mechanisms which may affect SOA formation from isoprene are implemented in MXLCH-SOA and discussed in Chapter 4. The hydroxyl radical (OH), the main oxidant of isoprene, is thought to be regenerated in the oxidation of isoprene. We find that for the tropical forest case study, we cannot reconcile the modeled concentrations of VOCs, OH and OA with their observed concentrations and fluxes both for cases with and without OH recycling. Therefore, we conclude that the issue of recycling of the OH radical in the oxidation of isoprene has to be solved before its effect on SOA formation can be determined. The formation of SOA from isoprene involves multiple generations of oxidation and due to this complex chemistry there is no single mechanism which can explain SOA formation from isoprene under all conditions. To gain understanding in this issue, we have implemented different pathways through which isoprene SOA is known to form, although we do not explicitly account for the detailed isoprene oxidation chain. A central aspect of this branching approach is whether the isoprene peroxy radical chemistry follows the low- or the high-NOx pathway. We find that the latter channel dominates in our case study. For SOA formed through the high-NOx channel, we further account for the effect of the NO2/NO ratio on SOA yields. In the presented case study this has little effect as this ratio is low, it but could be more important in regions with slower photochemistry or higher emissions of anthropogenic pollution. In the low-NOx regime, isoprene epoxides (IEPOX) are important intermediate gas-phase species in the formation of isoprene SOA. Even though the low-NOx pathway is only a minor one here, the amount of IEPOX SOA formed is likely substantial, although a better understanding of the exact mechanisms for its formation is needed to confirm this. However, as in previous studies we systematically underestimate the organic aerosol concentration in a tropical forest even though we incorporate the state-of-the-art knowledge on isoprene SOA formation in MXLCH-SOA. Nevertheless, we advocate accounting for NOx regime specific chemical pathways when modeling isoprene SOA formation. As this field is rapidly evolving in terms of the development of new measurement techniques and the discovery of chemical mechanisms, we strongly recommend the intensive use of our modeling system to gain further understanding of the diurnal variability of OA and for testing new hypotheses under atmospheric conditions. Satellite observations of cloud droplet concentration over the boreal forest The final objective of this thesis is to understand how aerosols and meteorological factors influence cloud droplet concentration over the boreal forest. This is a first step in translating the process understanding such as addressed in the previous chapters to larger spatio-temporal scales. Since this objective considers different temporal and spatial scales, a different method is applied in Chapter 5 than in the foregoing chapters. Observations of cloud properties by the MODIS instrument onboard the Terra satellite are combined with a model that contains the microphysics and thermodynamics of a single-layered water cloud to obtain a seasonal cycle of cloud droplet number concentrations, averaged over 9 years of data. This seasonal cycle in cloud droplet concentration is compared with aerosol concentrations at the surface and meteorological fields from ECMWF reanalysis. We find that the cloud droplet number concentration is related to the potential temperature gradient in the boundary layer, a measure for the strength of convection, while it shows no clear relationship with the cloud active aerosol concentration at the surface. From this we conclude that the convective transport of the aerosols from the surface to cloud base is the limiting factor for their activation as cloud droplets. However, convection will also influence the formation of clouds from a thermodynamic perspective. Therefore, it is likely that convection, as driven by land surface conditions, regulates both transport of aerosols to cloud base and the height of the cloud base, defined as the height at which water vapor reaches its saturation pressure. To ultimately understand the effect of the boreal forest on cloud properties, the effects of aerosols and thermodynamics should be studied simultaneously.