Mineral dust has manifold impacts on the Earth system. This includes land degradation at the dust sources, interaction with radiation in the atmosphere and effects on human health and economic activity. While it can be stated that most dust sources are found in arid and semi-arid environments, a general determination of characteristics that make a surface particularly susceptible to wind erosion cannot be given. One dust source type that has gained increasing attention in recent years is alluvial sediments. These sediments are formed and influenced by surface runoff and provide a large amount of fine grained material prone to wind erosion. Alluvial features are abundant in desert regions but are often small in size, for example dry river basins or alluvial fans. Due to their small size and despite their importance, these features are often underestimated or completely disregarded in dust-emission models. In this thesis, the spatio-temporal distribution of active alluvial dust sources is investigated and parameterized for a dust-emission model. For this, an approach to automatically detect alluvial features from two globally available satellite products is developed. These products comprise (1) surface reflectance at visible and near-infrared wavelengths derived from Sentinel-2 or MODIS and (2) HydroSHEDS flow accumulation data based on radar measurements. By combining these two datasets, an alluvial fines map (AFM) is created that shows the distribution of alluvial sediments. The AFM is implemented in a dust-emission model and multi-year model runs are performed for two study regions, one located around the Aïr Massif in the central Sahara, the other one covering western Namibia. Besides the distribution of fine alluvial sediments, another hydrologically influenced source type is analyzed in Namibia, i.e. the Etosha pan, a salt pan that is one of the most important dust sources in southern Africa. Dust activity from Etosha pan exhibits a strong seasonality due to regular flooding of the pan. These inundation events are implemented in the model by creating a monthly water mask from MODIS reflectance data. In the central Saharan study area, a comparison of the simulated dust flux with observed dust source activation frequency (DSAF) derived from the MSG SEVIRI Desert-Dust-RGB product shows that the model is able to reproduce the spatial and seasonal differences in the main activity of the identified sources. This seasonality cannot be reproduced by a control model run, in which the sediment supply by alluvial features is not included explicitly. For the Namibian study area, a model run is performed that includes the monthly water mask for Etosha pan and the AFM for the coastal ephemeral river basins. The simulated period covers 13 years from 2005 to 2017. With an empirical orthogonal function (EOF) analysis, constellations of pressure systems in the southern African region are determined that lead to an increased dust flux from the study area. Especially the Berg wind situation, a unique pressure pattern found in southern Africa with dry and hot continental winds, is identified as an atmospheric circulation pattern that leads to increased dust activity from the Namibian sources. The results highlight how important the consideration of alluvial features is for an accurate simulation of dust fluxes. Due to the global availability of the satellite data, the approach can be implemented in regional, continental or even global studies. Long-term emission fluxes can be used to identify the influence of meteorological patterns on dust emission and can help to estimate dust fluxes under current conditions but also in a changing climate.

Inhalation of respirable dust in coal mines is a serious occupational health hazard which can lead to the development of chronic and irreversible lung diseases, such as Coal Worker's Pneumoconiosis (CWP) and Progressive Massive fibrosis (PMF). After the passage of the Federal Coal Mine Health and Safety Act (CMHSA) in the late 1960's the prevalence of CWP among US coal miners decreased. However, since the late 1990's a resurgence of lung diseases has been reported, particularly in central Appalachia. On the other hand, dust monitoring data suggest that concentrations of respirable coal mine dust (RCMD) and crystalline silica have been on a downward trend. This contradiction has prompted keen interest in detailed characterization of RCMD to shed light on dust constituents-and their sources. Such information might help miners understand where and under what conditions specific sources contribute to RCMD, and how dust controls and monitoring could be enhanced to mitigate the exposure to respirable hazards. Respirable dust particles generated in coal mines are generally associated with three primary sources: the coal strata that is mined and generates mostly coal particles that could contribute for lung diseases, the rock strata that is cut along with the coal and generates most of the respirable silica and silicates, and the rock dust products that are the main source of carbonates which could produce respiratory irritations. Thermogravimetric Analysis (TGA) is one of many analytical tools that might be used for dust characterization. Its primary benefit is that it can be used to apportion the total sample mass into three mass fractions (i.e., coal, carbonates, non-carbonates) which should be roughly associated with the primary dust sources (i.e., coal strata, rock dust products, rock strata) in many coal mines. This thesis consists of two main chapters: Chapter 1, outlines the research motivation, recaps the efforts to establish a standard TGA method for RCMD, and shows results of the validation experiments that were performed in the current work to enable application of the TGA method to a large set of RCMD and laboratory-generated dust samples. In Chapter 2, 46 lab-generated samples from primary dust source materials collected in 15 coal mines, and 129 respirable dust samples from 23 US coal mines are analyzed using the TGA method validated in Chapter 1. Results for both sets of samples are presented and the mine samples are interpreted based on sampling location, mining method and region. Additionally, Chapter 3 summarizes recommendations for future work. / Master of Science / The chronic exposure to dust generated in underground coal operations represents a serious health concern among coal miners that can lead to the development of lung diseases such as Coal Workers Pneumoconiosis (CWP or "black lung). Despite of dust compliance monitoring data that have shown that the concentrations of dust have been declining, since the late 1990's the number of US coal miners diagnosed with lung diseases has been increasing, especially in central Appalachia. This contradiction has prompted keen interest in detailed characterization of respirable coal mine dust (RCMD) to shed light on dust constituents-and their sources. Such information might help miners understand where and under what conditions specific sources contribute to RCMD, and how dust controls and monitoring could be enhanced to mitigate the exposure to respirable hazards. Thermogravimetric Analysis (TGA) has been proposed as an alternative approach for dust characterization. Its primary benefit is that it can be used to apportion the total sample mass into three mass fractions (i.e., coal, carbonates, non-carbonates) which should be roughly associated with the primary dust sources (i.e., coal strata, rock dust products, rock strata) in many coal mines. This thesis consists of two main chapters: Chapter 1, outlines the research motivation, recaps the efforts to establish a standard TGA method for RCMD, and shows results of the validation experiments that were performed in the current work to enable application of the TGA method to a large set of RCMD and laboratory-generated dust samples. In Chapter 2, 46 lab-generated samples from primary dust source materials collected in 15 coal mines, and 129 respirable dust samples from 23 US coal mines are analyzed using the TGA method validated in Chapter 1. Results for both sets of samples are presented and the mine samples are interpreted based on sampling location, mining method and region. Additionally, Chapter 3 summarizes recommendations for future work.