Research Article 2014 WRR Burdekin sediment budget | Page 6

Water Resources Research
10.1002/2013WR014386
3. Methodology
3.1. Suspended Sediment Sample Collection
River water samples were collected from existing streamflow gauge locations draining the five major sub-
catchments of the Burdekin River (Upper Burdekin, Cape, Belyando, and Suttor Rivers, as well as the Bowen
River to represent the otherwise ungauged Lower Burdekin subcatchment), the outflow of the Burdekin
Falls Dam and the end-of-river freshwater discharge point during streamflow events over five consecutive
water years (1 Oct to 30 Sept; 2005/2006–2009/2010). Site details, locations, and data history for each site
are presented in Table 1 and shown in Figure 1; time series plots of streamflow hydrographs and concentra-
tion data are provided in supporting information FS01. Surface water ‘‘grab’’ samples (top 0.5 m of water column)
were collected at these sites during flood conditions with a bucket and rope. Where possible, samples were col-
lected over the rising, peak and falling stages of the streamflow hydrograph over multiple streamflow events that
occurred each wet season. Samples were collected from the center of the channel flow where possible, and were
well mixed with a stirring rod before being subsampled into prerinsed 1 L polypropylene bottles. Samples were
kept on ice prior to laboratory refrigerated storage and subsequent analysis. These water samples were used to
measure total suspended solids (TSS) concentrations and to calculate fine suspended sediment loads for the
streamflow conditions at each site for each year sampled. We only examined the washload fractions because the
delivery of fine sediments to the GBR is the focus of this study [see Bainbridge et al., 2012].
To increase the spatial density of data, a network of trained landholders was established to collect water sam-
ples at ungauged minor tributaries, many of which become inaccessible to external visitors during floods.
Twenty-four sites were established, located as close to the bottom of each tributary catchment area as possible,
at sites safely accessible to the landholder during floods (see Figure 1). Between 2004 and 2011, volunteers col-
lected 460 water samples from the 24 sites over rising, peak and falling stages of streamflow events (supporting
information Table A1). Samples collected by the volunteer network were kept refrigerated until analyzed.
3.2. Laboratory Analysis
3.2.1. Total Suspended Solids Analysis
TSS analysis was performed at the TropWATER Laboratory, James Cook University (JCU), Townsville and at the
Queensland Department of Science, Information Technology, Innovation, and the Arts (DSITIA) laboratory in
Brisbane using standard techniques. TSS (in mg L 21 ) was measured gravimetrically by weighing the fraction
remaining on a preweighed Whatman GF/C filter (nominally 1.2 lm pore size), dried at 103–105  C for 24 h,
after vacuum filtration of a measured volume of sample (Method 2540D) [American Public Health Association,
2005]. We note there is a tendency for this method to underestimate the ‘‘true’’ suspended sediment concen-
tration (SSC) particularly where abundant (i.e., > 25%) sand particles are present [see Gray et al., 2000].
3.2.2. Sediment Particle-Size Analysis
A subset of water samples collected from the rising, peak and falling stages of the flood hydrograph for
each of the gauged sampling sites were selected for particle-size analysis. Samples were selected from
four of the study water years (2005/2006–2008/2009) where available and include a total of 274 sam-
ples. See Table 1 for site specific sample numbers and water years represented. These samples were
processed from either an additional 1 L bottle collected during streamflow events, or a subsample of
the original water sample. Particle-size distributions for the water samples were determined using a Mal-
vern Mastersizer 2000, a laser diffraction particle-size analyser with a lens range of 0.02–2000 mm. The
parameterization methodology of Sperazza et al. [2004] was applied, and all data presented are the
mean of three measurement runs. Sediments were classified as one of three size classes based on the
Udden-Wentworth sediment grain size scale [Leeder, 1982]: (1) clay (< 3.9 mm); (2) very fine and fine silt
(3.9–15.6 mm; hereafter referred to as fine silt); and (3) coarse silt and sand (15.6–2000 mm; hereafter
referred to as coarse sediment).
3.3. Sediment Load Calculations
Streamflow and corresponding TSS data from each of the gauged locations were entered into a regres-
sion style ‘‘Loads Regression Estimator’’ (LRE) model developed by Kuhnert et al. [2012] to predict sus-
pended sediment loads (in tonnes) with estimates of error for each subcatchment site and each water
year. The LRE uses a generalized additive model (GAM) to incorporate key hydrological processes con-
sisting of:
BAINBRIDGE ET AL.
C 2014. American Geophysical Union. All Rights Reserved.
V
9072