TRACING SUSPENDED SEDIMENT SOURCES

In order to identify the source of suspended sediment transported by rivers in the WRIScS study area, the ‘fingerprinting’ approach has been used.
The fingerprinting method involves collecting suspended sediment from a river at routine monitoring sites (photo right) and comparing its characteristics to those of surface soil and sediment samples collected from potential contributing areas or sources within the watershed. A detailed description of the technique is described in Walling et al. (1999) and Collins et al. (1997).

 

FIELD SAMPLING

The following samples were collected for use in the fingerprinting study:

  1. Surface soil samples were collected from each of the expected land use types, namely, citrus, banana and forested areas. The samples were collected from a representative range of site conditions within each of the land-use types, including soil, slope, position within the delivery pathway and at different distances from trees. Samples were also collected from the river banks at different points along the river at locations where bank erosion was likely to contribute to the suspended sediment load.
  2. Large volume storm-runoff suspended sediment samples were collected from each of the sites where river monitoring was undertaken. The samples were collected to cover different suspended sediment concentrations, rising and falling limbs of events and a range of conditions through the rainy season.

The majority of suspended sediment has a particle-size <63mm . Also it is likely that not all of the material on the hill-slopes (or source areas) will become eroded or transported as suspended sediment in the river . To attempt to minimise differences due to selective transport, only the <63mm fraction of the soil sample was used for fingerprinting.  

Samples were analysed for particle-size, carbon, nitrogen, phosphorus and radionuclides (Caesium-137 and unsupported Lead-210) to provide 'fingerprint' parameters

 

Description of the Mixing Model

The fingerprinting technique works by measuring a range of tracer properties for the source material and comparing the results with equivalent data measured from the suspended sediment collected from the basin outlet.

The tracer properties used in this study were selected on the basis of over fifteen years of experience in using the fingerprinting technique at the University of Exeter, UK. These properties are selected to provide statistically significant differences between each of the land-use sources and are incorporated in a ‘composite fingerprint’.

The basis of the technique is to use the tracer properties in a multivariate mixing model to estimate the relative contribution of the potential sediment sources to a particular suspended sediment sample. A linear equation for each of the trace properties in the composite fingerprint relates the concentration measured in the suspended sediment sample to that in the mixture which represents the sum of the contributions from the different source groups.

The mixing model therefore consists of a series of linear equations for each of the tracer properties. This set of linear equations is solved using the least squares method with the relative contribution of the different sources to the suspended sediment sample established by minimising the sum of the squares of the residuals for each of the trace properties used. Thus the model is represented by the following equation:

where Res is the sum of the squares of the residuals for the n tracer properties, Cssi is the concentration of tracer property i in the suspended sediment sample, Csi is the mean concentration of the tracer property in the source group s and Ps is the relative proportion from source groups.

The constraints on the linear model are:

  • contributions must lie within the range 0 to 1

  • the sum of the contributions from all sources is 1

It has been shown that the concentration of tracer properties in sediment are related to the specific surface area of the sediment sample and consequently when comparing sediment and source material samples, as in this technique, it is important to take into account differences that may result from different particle size distributions. The concentrations of each of the tracer properties were therefore corrected according to the ratio of the specific surface area of the suspended sediment sample to the specific surface area of the source material sample.

The discharge and suspended sediment concentration and thus the suspended sediment load will vary between the samples collected. In averaging the results from the mixing model runs for a series of suspended sediment samples this needs to be accounted for. The results are therefore weighted according to the instantaneous suspended sediment load at the time of sampling according to the following equation:

Psw is the load-weighted relative contribution from source group s, Lx is the instantaneous suspended sediment load for suspended sediment sample x, Lt is the sum of the instantaneous loads (Lx) for all samples from that sampling site and Psx is the relative contribution from source group s for sediment sample x.