2 LITERATURE REVIEW 2.1 Introduction
2.4 Advantages and limitations of suspended sediment monitoring methods
hydrograph, through as many flood events as possible. Monitoring through the highest flows in the study period allows the establishment of a robust site-specific relationship between SSC and water level, and the estimation of peak discharge sediment loads for higher flows with longer (e.g. >2 year) return periods that may not have been experienced within the study period. Such high discharges are required for example in planning scenarios, or model calibrations.
Placing monitoring points upstream of confluences allows comparison between catchments and hence insight into the effects of catchment characteristics on SSC, such as land cover or gullied areas. At reach scale, the representivity of point sampling needs to be calibrated by width and depth integrated sampling at a range of discharges, to accommodate in
channel SSC variation.
2.4 Advantages and limitations of suspended sediment monitoring
Manual turbidity and visual clarity measurements can be added to this categorisation.
Manual and automatic sampling
Historically, most SS sampling campaigns relied on manual sampling (Ballantine 2015), whereby water samples were collected by hand for later SSC analysis or turbidity measurement, and/or where hand-held equipment such as turbidity meters, Secchi disks or clarity tubes were used to measure in-channel turbidity or visual clarity.
Manual samples can be taken on a fixed or flexible (e.g. flood-responsive) schedule by a relatively unskilled operator, and do not require a power source or fixed housing. The equipment used can be relatively low cost and unsophisticated, comprising either open vessels for grab sampling or simple isokinetic samplers for depth-integrated sampling through the water column. Operators can adjust their position at the sampling site in response to flow and bank conditions, and point or depth-integrated samples can be taken without the operator entering the water. No fixed power supply or housing is required, removing these limitations from sample site selection. Consistency of results relies heavily on operator training and compliance, albeit at a relatively low technical level (Bannatyne et al 2017). The training and administration of personnel is a significant commitment which is ongoing throughout a manual sampling programme (Bannatyne et al 2017), but which is likely to produce acceptably consistent and accurate SS data (See Chapter 6).
However, unless the person responsible for sampling remains permanently near the sampling point and is thus available to notice and monitor flood flows, SS loads and yields may be under-estimated (Horowitz 2013). Access to and proximity of the sampling site for the operator is therefore a limitation on sample site selection. Further, safety considerations at night and during dangerous weather and/or flow conditions potentially limit flood flow sampling. Due to this requirement for near-continual but still limited operator presence, the manual sampling method has been described as labour intensive, expensive, inconvenient, difficult and hazardous (Kuhnle 2013) not only in comparison with contemporary instrumented approaches, but also due to the cost of laboratory sample analysis (Ballantine et al. 2015). This “first world” view may however be less pertinent in developing countries such as South Africa where employment is scarce and labour costs are low, and where involvement in a sampling programme provides direct financial and social benefits in terms of job creation and poverty alleviation to individuals and their communities (Bannatyne et al 2017).
Grab samples that are only representative of near-surface SSC provide good data in homogenous conditions, e.g. where flows are uniform and where silt-sized particles predominate (Gordon et al. 2004). Isokinetic depth-integrated sampling allows a more representative sample to be taken throughout channel depth, although it under-represents
the small area of suspended bed sediments below the vessel inlet point (Gordon et al.
2004). Accuracy depends on the operator using the correct transit rate to prevent over-filling the vessel during sampling (Gordon et al. 2004).
Passive (and pump) point and rising-stage samplers (i.e. automatic samplers) comprise medium-cost installed equipment that, like manual sampling, extract a water sample either from a single point, or at stages from the rising water column, including close to the river bed. They are commonly used in large-scale, long-term monitoring applications by agencies such as the United States Geological Survey (USGS) (Kuhnle 2013), but are not in widespread use in South Africa. They can be set to sample at specific times or to be triggered by rising flows, and offer 24 hour sampling potential. The greater sample volumes required to accurately determine load and yield at low SSC can more easily be accommodated with automatic than with manual sampling (where large vessel sizes become unwieldy). However, low flow/low SSC is a less important component of load and yield estimations than the high flow/high SSC which can be accurately sampled with smaller vessels (Gordon et al 2004).
Whilst automatic samplers reduce to an extent the dependency on operator presence, they still require regular emptying, servicing and maintenance (Kuhnle 2013). Sample times are not always known for rising stage samplers, pump samplers require a power supply, and both rising-stage and pump samplers require a secure housing or installation point, being vulnerable to theft, vandalism, and damage by floods and large debris. This limits their use and placement due to high associated costs. Further, the samples taken at “rising stages”
throughout a flood event are always extracted from the water surface area and are thus not representative of SS throughout the full depth of such flows (Gordon et al 2004), particularly the sand-sized fraction. Similarly, pumped samples are always taken from a fixed point above the bed and may not be representative of the upper water column during high flows.
In common with manual sampling, representivity must be achieved by correlating point or depth-integrated automatically-collected samples with periodic width- and depth-integrated sampling of the channel cross-section (Gordon et al 2004), as relationships at different discharges between such samples and cross-sectional SSC are not constant (Kuhnle 2013).
Manually monitoring turbidity and visual clarity as a surrogate for SSC can be done in
channel or on extracted samples. The same advantages and disadvantages pertain as with manual sampling, with the added benefit that laboratory analysis is not required (Ballantine et al 2015), but the additional constraint that robust relationships between turbidity, visual clarity and SSC can be difficult to accurately establish (Ballantine, 2015). Further, turbidity meters typically do not cope with very high sediment conditions, requiring sampling and SSC analysis still to be undertaken (Bannatyne et al 2017).
Installed sensors and probes
Wren et al (2000) and Kuhnle (2013) concur on many points regarding the benefits and limitations of installed sensors and probes for the continuous acquisition of SS data. Fixed instruments offer consistent and accurate results from 24-hour monitoring, and avoid reliance on and training/ administration of operators. Although they may require skilled installation, set-up, and maintenance, they require little ongoing support once this is done, beyond occasional download of stored data if not continuously sent electronically or via the cellular network to off-site storage.
A range of sensors and probes are manufactured which use either optical turbidity as a surrogate for SS, or measure SS by acoustic or optical backscatter, or laser diffraction.
These instruments range from sophisticated to highly sophisticated, with associated vulnerability to malfunction. All require secure placement and a permanent power supply, and all are thus vulnerable to theft, vandalism and flood damage. Most are relatively expensive capital items, some extremely so, which in combination with the foregoing factors limits their use and placement due to cost, typically reducing the spatial coverage of a programme (Wren et al, 2000; Kuhnle 2013). Most require periodic servicing and maintenance due to e.g. biofouling of sensors.
Additionally, few such instruments are manufactured and/or repaired in South Africa which experiences a weak currency value against those of the typically European or American manufacturers All this leads at best to lengthy delays in supply/ repair, and at worst to unaffordability, or non-replacement if the instrument is damaged or lost. Further, their purchase conveys no benefit to the local community or economy (Bannatyne et al 2017).
Being fixed at a point in the channel, most such instruments require calibration at least against local conditions and particle sizes, and typically against manual width and depth integrated channel sampling. This implies that limited manual sampling must still be included in monitoring programme design and implementation (Wren et al. 2000; Gordon et al. 2004;
Kuhnle 2013).
Conclusions
The choice of sampling method has profound implications for the design and implementation of a SS sampling programme. Choice of method is subject to a range of spatiotemporal and socio-economic limitations, and offers a reciprocal range of benefits. “Plug and play”
installed instrumentation has the attraction of avoiding the significant training and administrative burden associated with manual sampling methods, cost being the significant drawback. Manual sampling requires a high level of support, yet in the South African context offers a relatively affordable method of acquiring acceptable SS data that has tangible benefits not only to the research programme, but also to the communities within which such
research is being undertaken. The efficacy of the CT-based approach to direct SS sampling is the focus of this thesis.