Aeolian dust emission dynamics across spatial scales: landforms, controls and characteristics
Doctoral Thesis
2018
Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
University of Cape Town
Faculty
License
Series
Abstract
Variable erodibility (surface characteristics) and erosivity factors (meteorological conditions) result in dust emission dynamics being complex in both space and time. Accounting for localscale surface variability is critical to our understanding of dust emitting processes. This study identifies mineral dust using remote sensing, establishes emission thresholds through field measurements and identifies particle chemistry for major dust sources in the Central Namib Desert. Examining over 2000 Landsat images over a period from 1972 to 2016, identified 40 days of visually detectable dust, originating from sub-km scale point sources. The observations suggest that dust sources can be identified at the landform scales which particularly include ephemeral river valleys and saline pan surfaces. These persist throughout the 25-year record; however, a gradual shift in source point clusters is noted through time, which can be tentatively attributed to anthropogenic modification of the hydrological systems. A PI-SWERL (Portable In-Situ Wind ERosion Lab) wind tunnel was used to measure the emission potential of the Landsat derived targets. The most emissive sources were paleostockpiles of alluvial silt deposits and associated degraded nebkhas within the Kuiseb River Delta. These had a geometric mean emission flux of 0.076 mg m-2 s -1. In comparison, the active channel had a geometric mean emission flux of 0.008 mg m-2 s -1, undisturbed desert pavement 0.007 mg m-2 s -1, pan surfaces 0.001 mg m-2 s -1 and wadis within the gravel plains 0.030 mg m-2 s -1. The emission thresholds were augmented with site-specific field measurements such gravel cover (%), moisture content (%), particle size (µm), elemental composition (%) and shear and compressive strength (kg cm-2). A Boosted Regression Tree (BRT) machine-learning algorithm identified the most important surface and sediment characteristics determining dust emission from the measured surfaces. The model explained 70.8% of the deviance in the measured dust flux with the top predictor variables and their relative importance (%) as follows: gravel cover, 16%; moisture content, 14%; kurtosis, 13%; very coarse silt, 13%; very fine sand, 11%; fine sand, 8%; compressive strength, 7%, calcium, 7% and magnesium, 6%. Such an analysis can be used to identify critical thresholds for dust emission and standardise testing protocols. Linking landforms with such emission measurements allow for the assessment of two existing dust emission schemes: the Preferential Dust Scheme (PDS; Bullard et al. 2011) and the Sediment Supply Map (SSM; Parajuli et al. 2017). Although these schemes represent a major advance in our representation of dust emission source areas and erodibility, this study shows that these schemes still need to be improved to accurately depict dust emission potential. For the PDS this would include producing a global rasterised output with quantified dust emission potential and for the SSM, a more accurate classification of the highly emissive geomorphic units. Landsat source point sediments were subjected to physical and geochemical analyses and compared to samples obtained from passive collectors such as the Big Spring Number Eight (BSNE) and active PI-SWERL exhaust emissions, using an auto-SEM (QEMSCAN). This provided individual particle mineralogy (>2 µm resolution) for a total of approximately 10000 to 60000 particles per sample which enabled a comparison of particle size, shape and mineralogy. The samples consist of a mixture of minerals reflecting the varied metamorphic geology and consists predominantly of feldspar, quartz, mica, other aluminosilicates such as the alteration products epidote and chlorite and low to medium grade metamorphics such as amphibole and pyroxene, iron oxihydroxides, titanium minerals, carbonates and clay minerals.
Description
Reference:
von Holdt, J. 2018. Aeolian dust emission dynamics across spatial scales: landforms, controls and characteristics. University of Cape Town.