Thesis (M. Sc.)--University of Hong Kong, 2007.

published_or_final_version / Environmental Management / Master / Master of Science in Environmental Management

M.Sc. (Environmental Management) / The use of atmospheric dispersion models to predict ground level pollutants concentrations has been on an increase in South Africa in the last decade. At this stage National Department of Environmental Affairs has published a draft document to provide guidelines on the type or use of models. Most Air Quality Specialists in the country make use of the United States Environmental Protection Agency approved atmospheric dispersion models to conduct air quality investigations. These models were developed in the United States of America after having considered the environmental set up and monitoring capabilities. In light of the above, much of the required input data are not readily available and calculations have been conducted to make up for the shortfall. For domestic emissions, quantifying the emissions factors is proving to be a challenge for modellers. They calculate emissions factors using different data sets from variable sources – sometimes the data are not up to date. This variability could potentially compromise the output of the model. This study aim was to model domestic emissions from an isolated rural township, Leandra, in the Mpumalanga Province – located within a nationally declared Highveld air quality management priority area – for two one month periods – in both the winter – July 2008 – and the summer – October 2008. This was achieved by using a United States Environmental Protection Agency approved AERMOD atmospheric dispersion model. Hourly surface measured meteorology data were obtained from the Langverwacht ambient air quality monitoring station and upper air data from the Irene monitoring station. The data were screened for any suspect values, formatted and then pre-processed by AERMET to be used by AERMOD. The study also investigated and compared the modelled time-series and monitored time-series data. This study calculated the effective emissions rate of 0.3 g PM10 s-1 m-2 by using a combination of monitored hourly PM10 concentrations and dispersion modelling time series data, for a typical Highveld township. Furthermore, the study revealed that, during winter when air is stagnant, Leandra was demonstrably isolated from other emissions sources of strength in the region – i.e. power station and domestic emissions were the dominant emissions sources. Under these circumstances, indoor and outdoor emissions were above the acceptable standards – i.e. they constituted unhealthy ambient air conditions. During summer – with the higher average wind speeds – Leandra was under the influence of industrial sources and the argument of isolation was not valid.

M.Phil. (Energy Studies) / Southern Africa is a major source of regional aerosols and trace gases from biomass burning, and this creates a need for experimental validation and systematics of the magnitude and frequency of aerosol transport episodes affecting the atmosphere of the region. This study links surface measurements of biomass burning atmospheric aerosols and trace gases with air mass trajectory analysis to determine transport pathways for periods of high and low concentrations. The hypothesis of this study is that from chemical signatures of trace gases and aerosols, as well as trajectory analyses, it is possible to identify sources of these emissions from industrial, traffic, marine and biomass burning activities. Consequently, frequencies, durations, intensities and seasonal variations of trace gases can be established. The study aims to interpret the long-term atmospheric monitoring record from a remote monitoring station at Botsalano (North West Province, South Africa) to determine the origin, frequencies, durations, intensities and seasonal occurrences of aerosol/haze episodes influencing the atmosphere of southern Africa. A suite of trace gas analysers and a Differential Mobility Particle Sizer (DMPS®) were used to measure ground level trace gas and aerosol quantities. MATLAB® scripts were used in performing quality assurance and processing to provide a working set of data from which different fire periods could be selected. Fire signatures, based on excess CO above average tropospheric levels and episodes of enhanced particulate matter concentrations in the 10 to 200 nm range, were identified using MATLAB® scripts and Excel®. Altogether 36 plumes were accepted as biomass burning plumes. Twenty-nine fire plumes had weak signals with excess CO ratios ranging between 0.07 and 0.32; seven plumes had strong signals ranging between 0.41 and 0.64. The occurrence of identified biomass burning plumes was high in the dry season from May to October (83%) and low (17%) during the wet season from November to April. Four pathways were identified for the long-range transportation of biomass burning aerosols to the site: easterly, south-westerly, re-circulation and northerly modes, with occurrence frequencies of 39%, 31%, 22% and 8%, respectively. Anti-cyclonic circulation was observed over southern Africa and was evident in the re-circulation and Indian Ocean slow modes. CO and Aitken-mode aerosol number intensities were generally larger for fire emissions arriving in the easterly and south-westerly air masses when compared with those arriving in re-circulation and northerly air masses. Easterly and south-westerly flows were dominated by Aitken-mode aerosol, whereas accumulation mode particles dominated in the re-circulation and northerly modes. Consequently, easterly and south-westerly flows transported emissions from young/fresh fire plumes, with source regions probably close to Botsalano. Re-circulation and northerly flows were responsible for transport of rather aged plumes from more distant regional fires. Based on forward trajectories, this study revealed that the 2006/2007 measurement period exhibited transport features of a La Niña ENSO during which transport of biomass v burning aerosols towards the south in the Indian Ocean slow and Indian Ocean fast modes was most frequent. This study is significant in that it complements earlier studies of regional aerosol transport over the sub-continent and adds to the understanding of the regional scale generation and transport of trace substances through the atmosphere. Furthermore, the study combines a technique for identifying enhanced CO concentrations as a unique identifier of large scale biomass combustion events with the use of the Aitken-mode particle number densities and size distributions. This technique reveals aspects of aerosol growth dynamics through the changing size distributions, thereby adding fresh insights normally not available through conventional particle volume/mass concentrations measurements.

Ph.D. (Geography) / The focus of Aeolian research has mainly been on wind-blown dust from desert and arid areas. Numerous dust emission schemes have been developed over the years aimed at accurately estimating dust emission rates from various soil types and land use surfaces. Limited research has been done on wind-blown dust from smaller area sources – such as mine tailings and ash storage facilities. Lately, the concern about the environmental and health impacts, caused by dust from mine tailings storage facilities and ash disposal sites, has become more prominent, calling for better methods in determining dust emissions and their related impacts. This thesis established a practical approach for wind-blown dust emissions estimation and dispersion modelling from mine waste and ash storage facilities for the purpose of legal compliance assessment. Extensive research on the physics of wind erosion has been done over the past decade, compelling the re-evaluation of previously applied techniques. The latest and most widely applied dust emission schemes are evaluated to determine, through systematic testing of parameterisation and validation, using empirical mine waste and coal ash data, a best-practice prescription for quantifying wind-blown dust emissions and determining effects on a local scale using commercially available dispersion models. The applicability of two dust-flux schemes, (one developed by Marticorena and Bergametti (1995) and the simplified Shao 2004 scheme, as reported in 2011) for the quantification of wind-blown dust emissions, were tested using site specific particle size distribution data, bulk density and moisture content from six gold- and one platinum- tailings storage facilities and from two ash storage facilities. The availability of the required input parameters and the uncertainty associated with these parameters, were tested. The dependency of the Shao et al. (2011) model on plastic pressure (P) and the coefficient cy, both of which are not easily determined, added to the uncertainty of the emission rates. In this study, P and cy were both interpolated using the range limits provided by Shao (2004) for natural soils. By calculating P, using the salt and calcium carbonate content, similar values were obtained. The minimally disturbed dust fraction, as required by the Shao et al. (2011) scheme were derived from particle size distribution analysis but found to be more representative of the fully disturbed particle size faction (𝜂fi) and therefore needed to be corrected to represent the minimally disturbed particle size faction (𝜂mi) through the application of a correction factor, CF𝜂mi. Specific attention was given to the quantification of the threshold friction velocity (u*t) and the threshold velocities (u*), and how these two parameters relate to each under variable wind speed and time durations. This was tested using sub-hourly averaged meteorological data, one set reflected 5-minute intervals and the other 10-minute intervals. Dependent on the frequency and strength of the sub-hourly wind gusts, the resulting dust-flux rates were found to vary significantly when based on hourly averaged wind data in comparison with 5- and 10-minute wind data. Dispersion models are useful tools in air quality management. Whereas ambient monitoring provides actual ambient concentrations for specific pollutants at set locations, atmospheric dispersion models can be used to simulate any number of pollutants and determine the impacts at any location within the modelling domain. These dust-flux schemes of Marticorena and Bergametti (1995) and Shao et al. (2011) have been coupled with the US EPA regulatory Gaussian plume AERMOD dispersion model for the simulation of ground level concentrations resulting from wind-blown dust from mine tailings facilities. For this study, two Case Studies were evaluated; one included two of the gold mine tailings and the second focused on the platinum tailings. Simulated ambient near surface concentrations were validated with ambient monitored data for the same period as used in the model. For the Marticorena and Bergametti (1995) dust-flux scheme, only z0 had to be adjusted to provide a good fit with measured data – whereas the Shao et al. (2011) scheme resulted in significantly higher concentrations, resulting in an over-prediction of the measured data. By applying the correction factor, CF𝜂mi, to the minimally disturbed dust fraction, the predicted concentrations improved considerably. The coupling of the dust-flux schemes with a regulatory Gaussian plume model provided simulated ground level PM10 concentrations in good agreement with measured data. The best correlation was found under conditions of high wind speeds when the prevailing wind was from the direction of the tailings storage facility. This thesis demonstrates that simulated impacts from complex source groups can be performed, within an acceptable range of certainty, using widely applied dust-flux schemes. These dust-flux schemes, developed primarily for large-scale desert and arid areas, have been demonstrated to be applicable also to small-scale sources, of the order of 1 km2, and can be coupled to regularly available dispersion models for impact evaluations of wind-blown dust. The value of this improved approach to the mining and mineral processing industries are substantial, allowing for more accurate health risks and adverse environmental assessments from wind-blown dust from large material storage piles, a source category that has hitherto been difficult to quantify.