Introduction

Engineers have for a long time recognised the need to measure catchment erosion rates and this is frequently done by monitoring how much sediment is transported in the stream or river which drains it. These sediment data are required during the feasibility stages of all irrigation or water resources projects and the values obtained may have a major influence on the final design. Repeated sediment data collection is also a part of the appraisal of many ongoing projects. Each new requirement has given rise to the development of a different monitoring technique. This set of twenty-four slides shows examples of some of these techniques in use throughout the developing world and covers the conditions likely to be associated with various catchment sizes.

Sediment Monitoring

It has been estimated that a total of 14km³ of weathered material is carried by rivers to the sea each year. This is equivalent to an average of 0.1mm being lost from the whole of the earth’s land surface. (More graphically, this equates to roughly 1cm being lost form the whole of Nepal or more than 2 metres from the land surface of Brunei.)

These high sediment loads in rivers and canals create many problems for the engineer who is trying to implement techniques to control and use the water. Siltation in a river channel will reduce its discharge capacity and may result in flooding, accretion in a reservoir will result in a reduction in storage capacity and therefore its useful life; sediment deposition in an irrigation scheme may modify the flow network and thus give rise to problems with water management.

As part of its research into problems associated with irrigation and water resources, the International Development Group (IDG) of HR Wallingford has been involved in studies of catchment erosion and the sediment processes. The data collection requirements have been such that a number of sediment sampling systems have been developed and many of the slides included here show examples of these systems in operation. The number of slides chosen from any particular country does not necessarily correspond to the magnitude of the problem in that country but rather just reflects the areas where our research has been conducted.

Throughout the developing (and developed) world, man continues to make dramatic changes to the local environment. These changes, be they from forest clearance, intensive agriculture or communication and urban development, have invariably resulted in increased run-off and soil vulnerability and hence, increased soil erosion. It has been calculated that in the developed catchments of Asia, the erosion rate increases by 50% every decade.

In the developed world, a large amount of research has been undertaken into techniques to predict erosion rates. These techniques are well suited to temperate conditions. However, in tropical latitudes, the problems are exacerbated by the fact that the soils are generally more susceptible to erosion and are also subjected to more intense rainfall. The measurement and quantification of these sedimentation rates is very necessary if a better understanding of the sediment processes is to be achieved and improved design solutions are to be found.

However, differences in measurement and processing techniques developed to suit individual sites have caused many problems to those engineers and researchers who work on erosion and sediment yield projects. There is a growing feeling amongst these workers that standardisation of measurement techniques would be very valuable since it would mean that all the data collected from sites around the world would be directly comparable.

Not least amongst the problems is that of deciding which scale of measurement is required and how effectively and accurately it can be carried out. There is then the problem of how to predict sediment yield at one scale from information obtained at another scale. Once fine silt is picked up by the surface run off it remains in suspension until it leaves the catchment; however, the larger sand sizes may be deposited and re-worked many times. It is therefore not possible to apply a simple factor based on catchment size to scale up from one data set to another. This has led to the concept of a sediment delivery ratio defined as the ratio of sediment yield at the catchment outlet to the gross sediment yield at the field level.

This concept in turn is fraught with problems, one of the major ones being to decide on which parameter the sediment delivery ratio depends. There is evidence from the USA that sediment delivery ratio decreases with increasing catchment size. However, the general applicability of this technique to tropical areas has yet to be verified.

A range of numerical models, of varying complexity do exist which seek to provide solutions where there is little data available, but they tend to be site specific and data hungry. There is therefore an increasing need to try and improve the understanding of these processes and to provide a reliable national and international database on which they can operate.

Soil loss and sediment yield can be studied at several different scales, each of which has relevance to different aspects of the erosion problem. They all require different monitoring techniques.

Plot Studies

At the smallest size are Plot Studies which were originally proposed to compare erosion rates for different crop cover and soil types. They are typically rectangular plots, about 20 metres long by 2 or 3 metres wide. The top and both sides are bunded to reduce the ingress of material eroded from the surrounding land which would distort and measurements. These bunds can be made from soil, brick, concrete or plastic sheeting. At the bottom of the plot some device for collecting the run off water and suspended solids is constructed. These devices range from a simple funnel leading into a (clean) oil drum, to proportional flow splitters and concrete storage tanks.

By varying only one parameter between plots, this technique is very useful for comparing the erosivity of different soils or the protection given to one soil type by a range of crops. However, as discussed above, results obtained cannot be simply ‘scaled up’ to evaluate likely erosion from a whole catchment; there are many other effects which are evident at the larger scale that cannot be determined from plot data. Care must also be taken to ensure that the scale of the parameter under test is small enough for the experiment to be valid; there is no point, for example, in trying to identify the effects of tree spacing if you only get one or two trees to the plot!

If the data obtained from one series of tests are to be compared with those collected from another, then the researcher will also need to have information on the associated rainfall. For most plot studies, daily rainfall totals will be sufficient and these may be obtained from a standard chart recorder sited in accordance with international standards.

Sub-Catchment

Next up in size is the small Sub-catchment. This typically has an area in the order of 20 – 30 ha and is characterised by having one natural stream outlet where a gauging structure may be constructed. This scale of measurement is used to gain information about sediment yield, crop yields, runoff rates and nutrient loss. Sub-catchments may also provide valuable data on the effects of various catchment management practices.

Where possible, catchments should be selected such that variations in the parameter under study have a random distribution over the whole catchment area. In order to gain an insight into the erosion mechanisms present in a large catchment, a number of smaller sub-catchments can be chosen and these monitored to determine individual characteristics.

The minimum measurement programme at this scale should include rainfall intensity runoff and sediment loss. Depending on the research objectives, secondary parameters such as soil moisture, crop cover and crop yield can also be measured.

To calculate rainfall intensity, the researched needs to know the volume falling for given times throughout the storm, therefore at least one recording rain gauge is required. In some cases information on spacial variability of rainfall is also required which will entail the installation of extra rain gauges.

Data on run off and sediment yield will be collected at the outlet of the catchment; this is why it was said earlier that the catchment must have a well defined outlet channel in which a rating structure (eg one in which the discharge is directly related to the depth of flow) can be constructed. Because the requirement includes the monitoring of sediment, extreme care must be taken to ensure that the design of structure chosen will allow all the sediment to pass through while still maintaining a known discharge rating throughout a wide range of flows. The IDG has found that for most sites, a modified ‘H’ flume (designed by USDA Handbook No 224) has the best characteristics. Where high sediment concentrations are encountered, the upstream bed slope can be increased to a maximum of 8% in order to keep it self-flushing. A modification to the flume drop structure results in a turbulent zone downstream of the flume providing an ideal location from which to pump water/sediment samples throughout each storm, for later analysis.

An automatic pump sampler has been developed to carry out this work. An extended drive pump is mounted in the drop basis between two wing walls, so that sampling can occur for most storms, even when discharges are low. Discrete water/sediment samples are pumped into a maximum of 25 numbered, half-litre bottles at pre-programmed time intervals; the whole sampling sequence being triggered when the water reaches a pre-determined level in the flume. The bottles must be checked daily and any samples collected so that a laboratory determination of sediment concentration – usually by vacuum filtration, can be carried out. If these data are to be converted into sediment flux (gm/m²/s), the instantaneous discharge at the time each sample is collected must be known. This is done either by making a ‘tick’ mark on the water level trace, or by accurately recording the sample time, in an automatic logger.

The sample time interval may be controlled by any suitable device from the most simple cams to field micro computers and, to avoid any restriction on possible locations, the whole kit should run on 12 volt, rechargeable batteries.

Large Catchments

Large catchment studies are necessary to provide information on macro-scale projects, such as the effects of basin management or to assess the capacity and life of a possible reservoir. At this scale, the researcher will need to monitor river discharges and sediment loads issuing from a catchment of more than 10 km², and often greater than 200²km. In these cases construction of a flow gauging structure is unlikely to be practical due to cost, head, levels of sediment and discharge. Measurements of sediment yield from small catchments, cannot be directly extrapolated to large catchments, since the effect of the sediment delivery ratio phenomenon is not easily quantifiable and there is no reliable means of estimating delivery ratios from climatic and catchment characteristics other than those previously studied. A different set of techniques have therefore been devised to cover this situation and again we have to look at the two main parameters, stream discharge and sediment flux.

There are a number of tried and tested methods for measuring river discharges (such as International Standards Organisation ISO 748). In essence they all demand a very careful selection of the monitoring site. Ideally, it should be on a straight and stable reach with a ‘control’ section to provide a fixed relationship between discharge and stage (water level). In practice, this is often difficult to find and so frequent velocity and cross-section measurements must be carried out for the duration of the sediment study. Velocity measurements can be carried out independently from the sediment sampling or in conjunction with sampling, generally by current metering. Where necessary, velocity data can be collected by using a current meter suspended from a boat throughout the whole range of flows, however safety of the boat crew must always be paramount. An extrapolation of the measured flows based on theoretical work may therefore be necessary to obtain a relationship at the higher values. Once the stage/discharge relationship has been established, a simple record of the variation of water level (stage) with time will provide the require discharge data.

The sediment monitoring technique employed is another form of the pump sampling technique. Since the samples are to be collected from a natural river, the means whereby a representative sample (or set of samples) may be obtained, must be considered very carefully. Other research studies involving sediment monitoring that have been undertaken by IDG have clearly shown the need to consider bed material load and wash load separately when collecting and analysing samples. This is because different hydraulic factors govern their transport rate in natural rivers.

Wash load can be defined as the movement of fine sediment particles, usually taken to be less than 53 microns (63 x 10-³mm), which are permanently carried in suspension and are not found in appreciable quantities in the bed material. Their discharge is primarily dependent on the rate of sediment supply and is a function of the factors which govern gross erosion rates such as rainfall, relief, soil types, vegetative cover and land use. An important characteristic of wash load is that, apart from small random fluctuations, its concentration in the flow at any given location and time is uniform at all points in the cross-section. Wash load discharge can thus be readily calculated as the product of concentration at a point and the mean discharge at the time of sampling. Wash load is often also referred to as ‘silt load’.

Bed material load consists of coarser particles (greater than 63 microns) which are present in the bed. Some of these are temporarily supported I the flow by turbulent eddies but from time to time settle back on the bed, whilst others move in almost permanent contact with the bed. Their transport is primarily dependent on the carrying capacity of the flow, and is governed by hydraulic parameters, in particular the flow velocity. In contrast to wash load particles, suspended bed material particles are not uniformly distributed through the depth, concentration normally decreasing with height above the bed.

Measurements are thus required at a number of discrete points in the vertical in order to accurately define the sediment concentration profile. The product of concentration and velocity profiles at a vertical enables sediment flux profiles to be developed. The integral of these flux profiles over the depth gives the sediment discharge per unit width. Bed load is often referred to as ‘sand’ load.

Because suspended bed material load transport is governed by hydraulic parameters, a form of ‘sediment rating curve’ can be developed from a relatively short period of data collection. Provided that this rating curve covers the full range of flow conditions, it can be used to predict suspended bed material load from flow data alone. Wash load is not dependent on hydraulic parameters and therefore, if it is to be estimated reliably, concentration should ideally be monitored continuously.

To convert these data into total sediment discharge, it is necessary to understand the distribution of sediment within the flow. Slide number nine shows three graphs. On the left graph are plotted the vertical distributions of various sediment sizes from which it will be seen that the finer sizes have a virtually constant concentration with depth while at the other extreme, the sand is concentrated near to the bed. When these are multiplied by the measured velocity distribution as shown in the middle graph, the sediment flux, the right graph, for each particle size fraction is derived. The integration of the sediment fluxes with time provides the sediment yield for the site. In practice this process has been simplified so that only two sediment fractions are considered. Those greater and less than (or equal to) 63 microns.

Equipment and measurement procedures employed for monitoring wash and bed load

Pumping techniques allow sediment samples to be collected from various positions both horizontally and vertically distributed within the flow. From this data the variation of sediment concentration in this section can be studied and a more accurate figure of the sediment discharge derived.

The following guidelines apply to all the pump sampling techniques developed at HR Wallingford :

1. The pump must be able to pass sediment without changing its grain size distribution and without retaining any of the sample
   
2. The line velocity must be of the order of 1 m/s. For portable pumps this tends to result in a maximum pipe diameter in the order of 15mm (5/8 inches).
3. The pump suction head can never be greater than 9 metres and the suction pipe length is limited to a maximum of 30m. For pipe runs of more than about 10m, the plastic pipe will have to be reinforced.
4. When using sinker weights the size used is dependent on the velocity of flow and depth to be sampled.
5. All systems rely on the sampling nozzle(s) being properly aligned and positioned at known point(s), both horizontally and vertically in the flow.

For sampling in rivers where the bed is well defined and stable two methods are applicable. If a bridge is present near the selected sample site, then a simple crane can be used. A sinker weight must be attached below the nozzle to stop the assembly from being swept downstream. Generally the pump is on the bridge deck and the samples sucked up to the bridge to a maximum of 8 – 9 m.

Secondly, where no bridge exists, a mast fixed to the river bed with a series of nozzles mounted up its length can be employed. This has the advantage that the nozzle position is always fixed relative to the channel cross-section. The pump is generally kept on the river bank and the samples sucked from the mast to the bank. This system has a major disadvantage that floating trash may get caught on the mast and nozzles and, particularly during high flows, the field staff will not be able to reach the structure to clear it away. This build up of trash can cause excessive forces on the mast and local scour, either of which may result in the mast failure. If the river under investigation is subject to high trash loads or scour, this method is not advised.

Where no bridge exists, or problems of a meandering river channel, scour or trash are likely to be encountered, it is best to employ a cable way system. Here are the limiting factors :

1. the width of the river, since very large anchor piers will be required to hold the tension resulting from long spans, and most important,
2. the suction length of the sample line which is limited to about 30m.

The catenary sag in the supporting cable can lead to problems of an ‘up hill struggle’ to retrieve the sampling equipment and, during high flows, the cable may touch the water surface! In spite of the problems mentioned above, this method has the advantage that samples can be collected from almost any vertical or horizontal position across the flow.

It is recommended that an aerofoil section be put on the submersed part of the cable from which the sinkerweight and nozzle is hung; this can reduce drag on the support cable and suction line by as much as a factor of 10.

The suspension, mast or cableway techniques described above may also be used at control structures or within constrained channels, such as irrigation schemes.

In weir sluice channels or on wide weir crests, fixed masts with fixed nozzles are appropriate, on the assumptions that trash is not a problem and that suction line distances are not too great. Within irrigation channels, particularly lined channels, the fixed nozzle/mast system is also applicable. In the case of tunnels or culverts, the sediment flux may be measured at the outlet by using cranes similar in design to those used for sampling from a bridge.

Cranes mounted on the stern of boats have been used successfully, although maintaining the boat ‘on station’ can be a highly skilled operation. Safety considerations may mean that information cannot be obtained during periods of high discharge.

Sample analysis

The small, half-litre samples collected by the automatic pump samplers can only be subjected to the most simple gravimetric analysis. However, the size of sample obtained by the other techniques is only limited by problems of handling, which can be reduced by performing a sand/silt split on site.

Once a steady flow through the pump has been established, the suspended sediment is split into ‘sand’ and ‘silt’ samples by passing it through a 63 micron sieve. The ‘sand’ sample is that collected on the sieve and is nominally collected from a large volume of water, up to 50 litres. The ‘silt’ sample is collected from, nominally a 1 litre sample, after passing it through the sieve. In both cases the volume sampled is partially dependent on the concentration present in the river and, in the case of silt, the capacity of the filtration equipment employed to extract the silt particular from the water before oven drying.

Reservoirs

A further way in which an estimate of catchment-wide, mean sediment yield may be obtained is by running a series of comparative surveys of the closing reservoir every few years after impoundment. The accuracy of this method is dependent on the number of range lines surveyed and the method used to calculate the settle sediment volume.

In the above notes, it is stated that "in many tropical and sub-tropical developing countries, soil erosion is a serious threat to agricultural production". If the extent of this problem is to be quantified and the efficacy of erosion control monitored, the engineers and researchers must standardise their data collection techniques.

Studies carried out by the International Development Group of HR Wallingford have been directed toward prediction techniques and developing new methodologies to allow more reliable predictions of soil loss and sediment yield to be made.

This research is funded by the British Government’s Department for International Development.