Geochemical Fingerprinting - Application of Geochemical Tracers to Fluvial Sediment - page 18

to the deposit has been correctly assessed within the errors inherent in the statistical
analysis. However, if the intent is determine the relative amount of sediment eroded
from the two source areas, the results will be biased such that the model will overes-
timate the amount of material eroded from the closest source. Differences in particle
transport rates produced by varying particle sizes can lead to similarly biased results.
In the case where elemental concentrations are used as geochemical fingerprints,
there is also an assumption that the elements exhibit conservative behavior. That is,
the elements selected as a fingerprint move with the sediment and are not lost from
the system. This follows because inverse/mixing models represent a form of mass
balance analysis. Thus, elements that tend to be mobile within aquatic systems and
possess lower affinities for particulate matter generally serve as poor fingerprints.
2.3 Methodological Approach
While the specific methods used in geochemical fingerprinting varies from one inves-
tigator to the next, the general approach involves the completion of five key steps
(after Zhang et al. 2012 ): (1) delineation and characterization of sediment sources
within the catchment, (2) determination of the fingerprinting properties that most
effectively identify and discriminate between sediments of the defined sources, (3)
collection and characterization of river sediment, selected on the basis of the time-
frame under consideration, (4) determination of sediment provenance using numeri-
cal modeling procedures, and (5) assessment of the uncertainty inherent in the mod-
eling results. These steps are discussed in detail below.
2.3.1 Source Delineation
The first step in any fingerprinting analysis is to define the primary sediment sources
within the catchment that may be of interest. Historically, sediment sources have
been subdivided into two main categories: upland (hillslope) sediments, and channel
bed and banks sediments (Fig. 2.3 a). Both types of sediment may be eroded and
transported to the water body by one or more geomorphic processes.
For fingerprinting analyses, upland sources are often subdivided further on the
basis of the spatial extent and location of mapped geological units (Collins et al.
1997a ; Walling et al. 1999 ; Douglas et al. 2003 ; Miller et al. 2005 ), soil types (Miller
et al. 2013 ), land-use/land-cover categories (Collins 1995 ; Walling and Woodward
1995 ; Russell et al. 2001 ; Miller et al. 2013 ), or contributing tributary areas (Klages
and Hsieh 1975 ; Collins et al. 1997b , 2009 , 2010a ) (Fig. 2.3 b).
This spatially defined source approach is plagued by several problems. First, soil
erosion is not only a function of soil type, land-use, or the underlying geology, but
varies as a function of factors such as topography and process. Agricultural pastures,
for example, may be eroded in steep upland areas by sheet and rill processes and on
low-relief floodplains adjacent to the channel by advancing headcuts associated with