Understanding aerosol optical properties in fully coupled regional chemistry-meteorology/climate models

Resumo

Atmospheric aerosols are suspended particles in the atmosphere that, among other effects, influence the Earth's atmospheric system by changing the radiative budget. The main mechanisms of aerosols for modifying the atmospheric system are (1) scattering and absorption of solar radiation (the so-called aerosol-radiation interactions, ARI); and (2) altering clouds and precipitation, thereby affecting both radiation and hydrology (the so-called aerosol-cloud interactions, ACI; Boucher et al., 2013). However, the uncertainty of cloud representation and these aerosol effects is much higher than for any other climate-forcing agent. This occurs because physical, chemical and optical aerosol properties are highly variable in spatio-temporal scales due to the short-lived aerosol particles and non-uniform emissions. Aerosol optical properties, which strongly depend on the physical and chemical properties of particles, highly influence ARI and ACI. Hence, an accurate representation of aerosol optical properties is essential to reduce the high uncertainty associated to the former interaction processes. At the same, Europe is one of the most climatically sensitive world regions to climate changes (Giorgi, 2006). Within this target domain, the role of aerosols may be even more crucial over regions as the Mediterranean basin, a crossroad that fuels the mixing of aerosols from different sources (Papadimas et al., 2012). Because of all of that, the main objective of this Thesis is to characterize the representation of aerosol optical properties and its uncertainty by using the fully on-line coupled meteorological-chemistry model, Weather Research and Forecasting model coupled with Chemistry (WRF-Chem; Grell et al., 2005) over Europe. This objective enhances the process-understanding of the representation of aerosol optical properties in on-line coupled models and settles the first step towards an improvement of the representation of ARI and ACI, which finally would lead to a reduction in their uncertainty. In order to achieve the main object established, modelling and observational tools are used to study aerosol properties, especially optical properties, and their influence over the atmospheric chemistry-aerosol-cloud-radiation system. According to this main target, several specific objectives are derived, as described below: " To evaluate the different parameterizations/configurations of WRF-Chem and other models when modelling aerosols and their skill to represent aerosol optical properties. For that purpose, the next steps were: (1) to establish the "best" remote-sensing data available to assess simulations, by evaluating satellite products against Aerosol Robotic Network (AERONET) data; (2) to evaluate the representation of aerosol optical properties by using an ensemble of simulations derived from different European and worldwide initiatives; and (3) to determine whether the inclusion of ARI and ACI, which produce a significant increase of computational time, improves the representation of aerosol optical properties. " To assess the influence of aerosol processes, parameters or modelling approaches in the improvement of the representation of aerosol optical properties by WRF-Chem. For that purpose, it was quantified the sensitivity of aerosol optical properties to: (1) the vertical distribution and some of the most uncertain aerosol processes and parameters; (2) the size distribution of aerosols; and (3) the spatial grid resolution. The results of this Thesis indicate that aerosol optical depth (AOD) is better represented than Ångström exponent (AE), although variability is underestimated. Discrepancies between simulations and observations can be ascribed to errors in the model estimation of the aerosol dry mass, the fraction of particles for a given mass or the water associated with aerosols, as well as a misrepresentation in the aerosol vertical distribution or the influence of emissions and boundary conditions. Moreover, known errors from observations should be born in mind; in this sense, this Dissertation also sheds some light on the skill of different satellites to represent aerosol optical properties. The improvements observed in aerosol optical properties when ARI and ACI are taken account, in particular those related to the vertical distribution of aerosols, justify the inclusion of aerosol radiative feedbacks in the WRF-Chem on-line coupled simulations. The better skills when aerosol interactions are considered also justify the higher time required to run the coupled simulations. Thus, all the simulations carried out in this Thesis hereafter take into account ARI and ACI. In the next step, a test was carried out in order to quantify the sensitivity of aerosol optical properties and vertical distribution to dry deposition, sub-grid vertical convective transport, relative humidity (RH) and wet scavenging. Results highlighted the influence of secondary organic aerosols (SOA) in AOD representation during a biomass burning episode and the dependence of SOA on the sulphate-nitrate-SOA formation by changes in RH, dry deposition or vertical convective transport. By itself, dry deposition also presents a high uncertainty influencing AOD levels. Another well-recognized source of uncertainty in AOD is size distribution. Because of that, a sensitivity test has been designed in order to disentangle which parameter of a log-normal size distribution influences most AOD representation. Results indicate that AOD is highly sensitive to a reduction in the standard deviation of the accumulation mode due to the transfer of particles from the accumulation to the coarse mode. In the same sense, a lower geometric radius for the accumulation mode due to an assumption of smaller particles, and a higher radius for the coarse mode produces also important changes in AOD due to a redistribution of particles thought the total size distribution. As a final experiment, the influence of the resolution in the representation of aerosol optical properties was evaluated. The study revealed that increasing the spatial resolution improves the representation of variables associated with dust, such as AOD. RH representation, to whom AOD is particularly sensitive, also improves. All of these changes are consequence of an improvement in the representation of dynamical processes in regional meteorological/climate models when increasing the horizontal resolution. This Thesis has contributed to the understanding of how aerosol optical properties are implemented in an on-line coupled regional chemistry-meteorological model and the intrinsic limitations in those models. Because of that, further works can lead to address the reduction of the demonstrated uncertainties by a combined approach between observations and modelling. The decrease in the uncertainty of AOD and aerosol representation implies a reduction in the uncertainty associated to the representation of aerosol effects, both for ARI (by AOD) and ACI (by improvement in microphysical properties).