ADVANCING METHODS FOR APPORTIONING THE SOURCES OF SEDIMENT IN RIVERS: COMBINING SPECTROSCOPY AND STABLE ISOTOPES WITH BAYESIAN MIXING MODELS

Sediment fingerprinting is a commonly employed technique for estimating sediment contributions from various eroding terrestrial sources to fluvial sediment load via a mixing model approach. However, there remain significant shortcomings in sediment fingerprinting practice, specifically relating to difficulties in producing high-temporal resolution apportionment estimates, inconsistencies in mixing model uncertainty representation, and a lack of attention given to organic matter provenance. Addressing these deficiencies, a combined X-ray fluorescence spectroscopy (XRFS) and diffuse reflectance infra-red Fourier transform spectroscopy (DRIFTS) approach is developed to rapidly, accurately and non-destructively analyse suspended particulate matter (SPM) geochemistry directly from sediment covered quartz fibre filter (QFF) papers at masses as low as 3 mg. An improved Bayesian source apportionment mixing model is then developed which allows for full characterisation of spatial geochemical variability, instrument precision and residual error, to yield a realistic and coherent assessment of the uncertainties associated with sediment fingerprinting estimates. Lastly, a novel application of a coupled molecular and δ2H and δ13C compound-specific isotope analysis (CSIA) of leaf wax n-alkane biomarkers is conducted to demonstrate the apportionment of plant-specific organic matter contributions to streambed sediments. Employing these developments in conjunction with automatic water samplers, high-temporal resolution SPM source apportionment estimates are derived throughout the progression of numerous storm events in a lowland agricultural catchment, revealing significant temporal variability in SPM provenance at 60- and 120-min resolution. Lower resolution, weekly, baseflow sampling is also performed, revealing distinct seasonal cycles in SPM geochemistry and sediment source apportionment over a 23-month period. Collectively, the developments presented in this thesis significantly advance sediment fingerprinting research by enabling organic and inorganic fluvial sediment fractions to be quantitatively apportioned at both low- and high-temporal resolution within realistic levels of uncertainty, thereby enhancing our understanding of sediment dynamics under a range of instream hydrological conditions.

Name dc:type.qualificationname

phd

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

University of East Anglia

Year dc:date.issued

2015

Author dc:creator

• Cooper, Richard

Language dc:language

en