Geochemical Fingerprinting - Application of Geochemical Tracers to Fluvial Sediment - page 33

sediments during flood events (e.g., Peart and Walling 1986 ; Walling and Wood-
ward 1992 , 1995 ; Walling et al. 1993 ; Wood 1978 ; Collins et al. 1997a , b , 1998 ,
2001 ). Collins et al. ( 2001 ), for example, collected 65 samples from 13 floods dur-
ing two wet periods between 1997 and 1999 within the upper Kaleya catchment of
southern Zambia. The objective was to determine the predominant contributions of
sediment from four sources including bush grazed lands, commercially cultivated
lands, communally cultivated lands, and channel banks and gullies. As is now com-
mon, the collected sediment samples were analyzed for a range of properties that
respond to different environmental controls, an approach that is expected to lead
to better source area sediment discrimination because the individual tracers will
exhibit a higher degree of independence (Walling et al. 1993 ; Collins et al. 1997b ).
Collins et al. ( 2001 ) chose to analyze sediments for pyrophosphate-dithionite
extractable trace metals (Al, Fe, Mn), acid extractable metals and metalloids (As. As,
Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb, Sb, Sn, Sr, Zn), base cations (C, K, Mg, Na), organic
matter (C, N), radionuclides ( 137 Cs

226 Ra) and total P. Source contributions
were determined using the unmixingmodel described by Collins et al. ( 1997a ), which
included correction factors for grain size and organic matter as well as a weighting
factor to account for differing degrees of analytical precision between the tracers.
They found that surface soils in communally cultivated lands were the predomi-
nant sediment source. However, there were minor variations in source contributions
between the samples collected during an individual flood (Fig. 2.7 a). More impor-
tantly, because the samples were collected during different discharge and sediment
load conditions, attempts to decipher source contributions during an entire flood (or
longer time periods) needed to consider the sediment load at the time the suspended
sediment samples were collected. By considering load, the contributions associated
with samples characterized by higher sediment loads are given more weight than
samples characterized by lower sediment loads. Mathematically, a load-weighted
mean source contribution ( P sw ) can be calculated for any source(s) for any given
time period (Walling et al. 1999 ) using the following equation:

210 Pb ex ,

,

n

P sx L x
L t

P sw = (2.13)

s = 1

where L x is the instantaneous suspended sediment load at the time of sample col-
lection, L t is the sum of the instantaneous sediment loads for all samples collected
during the time period of interest (e.g., a flood or season), and P sx is the percent-
age contribution from a specific sediment source, s , to the sediment sample, x . When
applied to an individual flood, the weighted mean source contribution provides infor-
mation on the source of the sediments transported past the monitored site during the
event. In the case of the upper Kaleya catchment, Collins et al. ( 2001 ) found that in
addition to the noted intra-flood variability in source contributions, inter-flood and
seasonal variations in contributions occurred (Fig. 2.7 b).
The use of suspended sediment samples to determine the primary sources of
sediment to a channel has decreased in recent years as it is plagued by several
difficulties, including (1) the fact that estimates of sediment source within suspended