Sediment as a physical pollutant
Sediment as a chemical pollutant
Key processes: Precipitation and runoff
Key concepts
Measurement and prediction of sediment loss
Recommendations
Although agriculture contributes to a wide range of water quality problems, anthropogenic erosion and sedimentation is a global issue that tends to be primarily associated with agriculture. While there are no global figures, it is probable that agriculture, in the broadest context, is responsible for much of the global sediment supply to rivers, lakes, estuaries and finally into the world's oceans.
Pollution by sediment has two major dimensions.
One is the PHYSICAL DIMENSION - top soil loss and land degradation by gullying and sheet erosion and which leads both to excessive levels of turbidity in receiving waters, and to off-site ecological and physical impacts from deposition in river and lake beds.The other is a CHEMICAL DIMENSION - the silt and clay fraction (<63m m fraction), is a primary carrier of adsorbed chemicals, especially phosphorus, chlorinated pesticides and most metals, which are transported by sediment into the aquatic system.
Erosion is also a net cost to agriculture insofar as loss of top soil represents an economic loss through loss of productive land by erosion of top soil, and a loss of nutrients and organic matter that must be replaced by fertilizer at considerable cost to the farmer in order to maintain soil productivity. The reader is referred to Roose (FAO, 1994a) for a detailed analysis of the social, economic and physical consequences of erosion of agricultural land and of measures that should be taken to control erosion under different types of land use, especially in developing countries. Whereas Roose is mainly concerned with the impact of erosion on agriculture, this publication is primarily concerned with agricultural erosion from the perspective of its impacts on downstream water quality.
Control of agricultural pollution usually begins, therefore, with measures to control erosion and sediment runoff. Therefore, this chapter deals with the principal mechanisms which govern erosion processes, and those measures which can be taken to control erosion. Processes discussed here also apply to fertilizer and pesticide runoff presented in the following chapters.
Global estimates of erosion and sediment transport in major rivers of the world vary widely, reflecting the difficulty in obtaining reliable values for sediment concentration and discharge in many countries, the assumptions that are made by different researchers, and the opposing effects of accelerated erosion due to human activities (deforestation, poor agricultural practices, road construction, etc.) relative to sediment storage by dam construction. Milliman and Syvitski (1992) estimate global sediment load to oceans in the mid-20th century at 20 thousand million t/yr, of which about 30% comes from rivers of southern Asia (including the Yangtze and Yellow Rivers of China). Significantly, they believe that almost 50% of the global total comes from erosion associated with high relief on islands of Oceania - a phenomenon which has been underestimated in previous estimates of global sediment production. While erosion on mountainous islands and in upland areas of continental rivers reflects natural topographic influences, Milliman and Syvitski suggest that human influences in Oceania and southern Asia cause disproportionately high sediment loads in these regions.
Sediment, as a physical pollutant, impacts receiving waters in the following principal ways:
· High levels of turbidity limit penetration of sunlight into the water column, thereby limiting or prohibiting growth of algae and rooted aquatic plants. In spawning rivers, gravel beds are blanketed with fine sediment which inhibits or prevents spawning of fish. In either case, the consequence is disruption of the aquatic ecosystem by destruction of habitat. Notwithstanding these undesirable effects, the hypertrophic (nutrient rich) status of many shallow lakes, especially in developing countries, would give rise to immense growth of algae and rooted plants were it not for the limiting effect of light extinction due to high turbidity. In this sense, high turbidity can be "beneficial" in highly eutrophic lakes; nevertheless, many countries recognise that this situation is undesirable for both aesthetic and economic reasons and are seeking means to reduce both turbidity and nutrient levels. Box 4 presents the impact of sediment on coral reefs.· High levels of sedimentation in rivers leads to physical disruption of the hydraulic characteristics of the channel. This can have serious impacts on navigation through reduction in depth of the channel, and can lead to increased flooding because of reductions in capacity of the river channel to efficiently route water through the drainage basin. For example, calculations by the UFRGS (1991) of erosion and sediment transport in the Sao Francisco River Basin, a large drainage system in eastern Brazil, demonstrate that the central portion of the river basin is now dominated by sediment deposition. This has resulted in serious disruption of river transportation, and clogs hydraulic facilities that have been built to provide irrigation water from the main river channel. The sediment largely originates from rapidly eroding sub-basins due to poor agricultural practices.
|
BOX 4: SEDIMENT AND DESTRUCTION OF CORAL REEFS Sediment has been identified as a major cause of decline and destruction of coral reefs, world wide. Experts (M. Risk, pers. comm., 1995) estimate that percentages of reefs affected by siltation are: Central America - 100% Studies of coral reefs in the Australia indicate that terrestrial particulate organic carbon can be transported off-shore over distances of 110 km to reef locations (Risk et al., 1994). Sediment is largely produced by agricultural activities and from erosion of deforested lands. Sediment production from intensive logging of the island of Madagascar have killed the fringing reefs. Observations from space described the transition of Madagascar from an island of green in a sea of blue, to an island of brown in a sea of red (sediment). |
The role of sediment in chemical pollution is tied both to the particle size of sediment, and to the amount of particulate organic carbon associated with the sediment. The chemically active fraction of sediment is usually cited as that portion which is smaller than 63 m m (silt + clay) fraction. For phosphorus and metals, particle size is of primary importance due to the large surface area of very small particles. Phosphorus and metals tend to be highly attracted to ionic exchange sites that are associated with clay particles and with the iron and manganese coatings that commonly occur on these small particles. Many of the persistent, bioaccumulating and toxic organic contaminants, especially chlorinated compounds including many pesticides, are strongly associated with sediment and especially with the organic carbon that is transported as part of the sediment load in rivers. Measurement of phosphorus transport in North America and Europe indicate that as much as 90% of the total phosphorus flux in rivers can be in association with suspended sediment.
The affinity for particulate matter by an organic chemical is described by its octanol-water partitioning coefficient (KOW). This partitioning coefficient is well known for most organic chemicals and is the basis for predicting the environmental fate of organic chemicals (see Chapter 4). Chemicals with low values of KOW are readily soluble, whereas those with high values of KOW are described as "hydrophobic" and tend to be associated with particulates. Chlorinated compounds such as DDT and other chlorinated pesticides are very hydrophobic and are not, therefore, easily analysed in water samples due to the very low solubility of the chemical. For organic chemicals, the most important component of the sediment load appears to be the particulate organic carbon fraction which is transported as part of the sediment. Scientists have further refined the partitioning coefficient to describe the association with the organic carbon fraction (KOC).
Another important variable is the concentration of sediment, especially the <63 m m fraction, in the water column. Even those chemicals that are highly hydrophobic will be found in trace levels in soluble form. Where the suspended load is very small (say, less than 25 mg/l), the amount of water is so large relative to the amount of sediment that the bulk of the load of the chemical may be in the soluble fraction. This becomes an important issue in the monitoring of hydrophobic chemicals as noted in Table 17.
Unlike phosphorus and metals, the transport and fate of sediment-associated organic chemicals is complicated by microbial degradation that occurs during sediment transport in rivers and in deposited sediment. Nevertheless, the role of sediment in the transport and fate of agricultural chemicals, both for nutrients, metals, and pesticides is well known and must be taken into account when monitoring for these chemicals, and when applying models as a means of determining optimal management strategies at the field and watershed level. For this reason, models using the "fugacity" concept (uses the partitioning characteristics [Chapter 4] of chemicals as a basis for determining the environmental compartment - air, sediment, water, biota - in which the chemical is primarily found) has proven effective in predicting the environmental pathways and fate of contaminants (Mackay and Paterson, 1991).
FIGURE 4 Schematic diagram showing the major processes that link rainfall and runoff
· Conclusion: The role of sediment as a chemical pollutant is a function of the chemical load that is carried by sediments.
Organic chemicals associated with sediment enter into the food chain in a variety of ways. Sediment is directly ingested by fish however, more commonly, fine sediment (especially the carbon fraction) is the food supply for benthic (bottom dwelling) organisms which, in turn, are the food source for high organisms. Ultimately, toxic compounds bioaccumulate in fish and other top predators. In this way, pesticides that are transported off the land as part of the runoff and erosion process, accumulate in top predators including man.
The major characteristic of non-point source pollution is that the primary transfer mechanisms from land to water are driven by those hydrological processes that lead to runoff of nutrients, sediment, and pesticides. This is important, not only to understand the nature of agricultural pollution, but also because modelling of hydrological processes is the primary mechanism by which agriculturalists estimate and predict agricultural runoff and aquatic impacts. Except where agricultural chemicals are dumped directly into watercourses, almost all other non-point source control techniques in agriculture involve control or modification of runoff processes through various land and animal (manure) management techniques.
In large parts of the world, precipitation is in the form of rain. However, in those areas where precipitation is in the form of snow, the science becomes more complex. Nevertheless, control measures, whether for areas subject to rain or snow can be easily summarized. Therefore, for the purpose of this publication, focus will be on the relationships between rainfall and runoff.
While the practice of hydrology can be quite theoretical, the principal concepts are easily understood (Figure 4).
Rainfall: The primary controlling factor is the rate (intensity) of rainfall. This controls the amount of water available at the ground surface, and is closely related to measures of energy that are used in many mathematical formulations to calculate soil detachment by rain drops. Soil detachment makes soil particles available for sediment runoff.
Soil Permeability: Permeability is a physical characteristic of a soil and is a measure of the ability of the soil to pass water, under saturated conditions, through the natural voids that exist in the soil. Permeability is a function of soil texture, mineral and organic composition, etc.. In contrast, "porosity" is the measure of the amount of void space in a soil; however, permeability refers to the extent to which the porosity is made up of interconnecting voids that allow water to pass through the soil. As an example, styrofoam is highly porous but impermeable, whereas a sponge is both porous and permeable.
Infiltration: Infiltration rate, the rate at which surface water passes into the soil (cm/hr), is one of the most common terms in hydrologic equations for calculating surface runoff. Infiltration is not identical to permeability; it is mainly controlled by capillary forces in the soil which, in turn, reflect the prevailing conditions of soil moisture, soil texture, degree of surface compaction, etc. Infiltration will vary between and within rainfall events, depending upon factors such as antecedent soil moisture, nature of vegetation, etc. In general, infiltration rate begins at a high value during a precipitation event, and decreases to a small value when the soil has become saturated.
Surface runoff: This is the amount of water available at the surface after all losses have been accounted for. Losses include evapotranspiration by plants, water that is stored in surface depressions caused by irregularity in the soil surface, and water that infiltrates into the soil. The interaction between infiltration rate and precipitation rate mainly governs the amount of surface runoff. Intense rainstorms tend to produce much surface runoff because the rate of precipitation greatly exceeds the infiltration rate. Similarly, in areas of monsoonal rain and tropical storms, the length and intensity of precipitation frequently exceeds infiltration capacity. Destruction of protective surface vegetation and compaction of the soil, especially in tropical environments, leads to major erosional phenomena due to the amount of surface runoff (Figure 5). Except for nitrogen which is usually found in groundwater in agricultural areas, surface runoff is the primary contributor of agricultural chemicals, animal wastes, and sediment to river channels.
Interflow: (sometimes called "throughflow") Because soil horizons have different levels of permeability not all water in the soil will move downward into the groundwater. The residual water in the soil will move along the soil horizons, parallel to the ground surface. Interflow usually emerges near the bottom of slopes and in valley bottoms. Therefore, identification of these hydrologically active zones is an important part of agricultural non-point source control measures. Interflow is the mechanism which has also been linked to soil piping, a potentially destructive characteristic in some soils by which shallow "pipes" form naturally in the soil and are enlarged by interflow to the point where they collapse causing gullies in the agricultural surface.
FIGURE 5 Massive gully erosion in agricultural areas of southern Brazil
FIGURE 6 Relationship between drainage area and sediment delivery ratio (Source: USDA, 1983)
Groundwater: Groundwater is supplied by water which passes through the soil horizons into the parent material and/or bedrock underlying the soil. Groundwater tends to flow towards rivers channels where it emerges and supports stream flow during periods of little or no rain.
This component of stream flow is called "base flow". The chemistry of base flow reflects the soil and bedrock geochemistry, plus any agrochemicals that have been leached into the groundwater.
Snowmelt: The phenomenon of snowmelt greatly complicates prediction of agricultural pollution using conventional hydrologic models. Snowmelt, by itself, is not normally a major producer of surface runoff. However, the combination of spring rain and snowmelt on frozen or thawing soils can produce serious erosional problems. Snowmelt tends to contribute greatly to agricultural non-point source pollution by carrying to adjacent streams the animal wastes, sludges, and other wastes that were spread on frozen agricultural soils during the winter period. Correct management of animal wastes in regions of frozen ground has major beneficial effects on water quality.
The sediment delivery ratio (SDR) is commonly used in erosion and transport studies to describe the extent to which eroded soil (sediment) is stored within the basin. The SDR is defined as:
SDR =
where yield is determined from reservoir sedimentation or from a sediment monitoring station, and gross erosion is estimated using an estimation techniques such as the Universal Soil Loss Equation.
The SDR is always less than 1.0 as illustrated in Figure 6, indicating that soil that is eroded at the field level tends not to travel far before it is deposited. Indeed, sediment storage in rills on fields, at field margins and at the foot of slopes is large. Storage also occurs in river channels (bed and overbank deposition), in wetlands, and in reservoirs and lakes. The SDR is highly variable, however the concept is one of the most important in the understanding of erosion and sedimentation processes and how these operate in time and space (see, for example. Walling, 1983).
The concept of the sediment enrichment ratio (SER) is quite important in understanding the impact and economic cost of chemical loss from fields. The process of surface erosion tends to be selective towards fine particles. Consequently, the particle size characteristics of material eroded at source (at the plot level) is progressively changed towards finer particles through deposition of the coarser fraction (e.g. sand-size material). Because of the chemically enriched nature of fine particles due to the large surface area of clay-size sediment, the concentration of chemicals that are associated with sediment (phosphorus, metals, organic nitrogen, hydrophobic pesticides) increases as the impoverished sand-size fraction is lost during down-field transport resulting in an increasing proportion of the chemically enriched fine (silt-clay) fraction.
TABLE 7: Agricultural non-point source models (Compiled from: Beasley and Huggins, 1981; Knisel, 1980; Lane and Nearings, 1989; Novotny and Olem, 1994; Young et al., 1986; Abbott et al., 1986)
|
NAME |
APPLICATION |
TIME SCALE |
SPATIAL SCALE |
|
A. Low to medium data needs | |||
|
Unit area loads (statistical prediction) |
Sediment loss |
Long-term averages |
10's to 100's km2 |
|
NOTE: Statistical models use aggregated data for comparable conditions. Predictive power is low but can be useful for screening purposes or where no field data are available; or where the spatial scale is so large that field data are uneconomical. | |||
|
USLE (Universal Soil Loss Equation) |
Average soil loss for specific crops, etc. |
Annual |
Plot/field |
|
RUSLE/MUSLE |
Average soil loss for specific crops, etc. |
Annual |
Plot/field |
|
NOTE: Empirical USLE-type models have been applied to large area analysis, using remote sensing data, etc. for regional estimates of soil loss (e.g. Brazil). USLE-type models are often incorporated into more detailed hydrological models below. | |||
|
B. Data intensive modelling (process-oriented) |
| ||
|
ACTMO (Agricultural Chemical Transport Model) |
Hydrologic processes |
Event, continuous |
Field |
|
AGNPS (Agricultural Non-point Source Pollution) |
Hydrology, erosion, N, P and pesticides |
Event, daily, continuous |
Grid cell, field scale |
|
ANSWERS (Areal Non-point Source Watershed Environment Response Simulation) |
Hydrology, erosion, N P and pesticides |
Single storm |
Grid cell |
|
CREAMS (Chemical, Runoff and Erosion from Agric. Management Systems) |
Hydrology, erosion, N, P and pesticides |
Daily, continuous |
Field scale |
|
EPIC (Erosion-Productivity Impact Calculator) |
Hydrology, erosion, nutrient cycling, crop and soil management and economics |
Event, daily, continuous |
Field scale |
|
HPSF (Hydrologic Simulation Program-Fortran) |
Hydrology, water quality for conventional and toxic organic pollutants |
Event, daily, continuous |
Watershed |
|
SHE (Système Hydrologique Européen) |
Hydrology, with water quality modules |
Event, daily, continuous |
Watershed |
|
SWAM (Small Watershed Model) |
Hydrologic processes, sediment, nutrients and pesticides |
Daily, continuous |
Watershed |
|
SWAT (Soil and Water Assessment Tool |
Hydrologic processes, sediment, nutrients and pesticides |
Event, daily, continuous |
Simultaneous simulation for hundreds of sub-basins |
|
SWRRB (Simulator for Water Resources in Rural Basins) |
Water balance and hydrologic processes and sedimentation |
Event, daily, continuous |
Watershed |
|
WEPP (Water Erosion Prediction Project) |
Hydrologic processes, sediment processes |
Single storm, daily, continuous |
Hillslope, watershed, grid cell |
The Sediment Enrichment Ratio (SER) is defined as:
SER =
Sediment chemistry is measured at some point downslope, e.g., at the edge of a field or in adjacent streams.
The importance of the enrichment ratio lies in the fact that there is proportionally more fine-grained sediment transported than coarse-grained sediment during surface erosion. Therefore, the sediment being transported has a finer texture than the source soil material. Because of the affinity of soil nutrients for fine sediment, this proportionally larger loss of fine material means that there is net impoverishment of the soil. As discussed in Chapter 3 (Fertilizers), this usually constitutes "mining" of the natural nutrition of the soil (often referred to as "natural capital") and which may never be replaced by the addition of fertilizer. The cost to the farmer is therefore two-fold: loss of productivity due to loss of natural nutrition in the soil; and economic cost of fertilizer which is added in the attempt to compensate for this loss.
Agriculturalists worldwide have spent much time and resources attempting to find reliable methods of predicting erosion and sediment-associated chemical runoff under different conditions of crop type, tillage practices, etc. Consequently, there is a large number of models that have been developed for the prediction of agricultural non-point source runoff of sediment, nutrients and pesticides. Many of the models permit gaming with alternative choices of land management, crop type, and fertilizer and pesticide application rates. Because all models (except unit load models) require hydrometric input and many use a sediment sub-component, it is appropriate to integrate these into a single table (Table 7), together with their principal characteristics.
In general, there are three types of models based on input data needs.
1. Simple screening models, such as the unit load approach, attempt to provide approximate answers about the likely magnitude of sediment and chemical runoff. This approach is a statistical methodology in which data on runoff of sediment, nutrients and pesticides on a unit area basis (e.g. tonnes of sediment per hectare) are collated from many studies to reflect similarities in crop type, soil and physiographic characteristics. Unlike the other types of models which are largely focused on prediction and improvement of agricultural management at the farm level, the unit load approach is mainly focused on impacts of agriculture on downstream water quality and without consideration of alternative farm management practices.
Despite the unreliability and large margins of error (refer to Tables 8 and 14 and related text, this approach has been widely used as a cost-effective means for providing first-approximation answers for agricultural areas for which there are no data. The methodology was originally developed by McElroy et al. (1976) who collated a major database on unit loads. This approach was further developed into a US-EPA Screening Procedure by Mills et al. (1985) and which remains the most comprehensive document on the subject. Unit load data reflect conditions in the United States; application of these data to other climatic and physiographic environments should be avoided. Nevertheless, it is an approach that may be worth considering for development in other parts of the world.
2. Simple empirical relationships: the widely used and respected Universal Soil Loss Equation (USLE) of Wischmeier (1976) has had remarkable success at the plot level and has been incorporated into many of the complex models of Table 7. The USLE is designed as a field management tool and provides aggregated information at the storm, seasonal or annual level. Wischmeier (1976) reported that the average prediction error for annual soil loss was 12%; larger errors are to be expected for single storm events. The USLE is one way to determine erosion potential for input into the denominator of the Sediment Delivery Ratio. The USLE is detailed here because of its success and because the same type of approach has been used in Africa (Elwell and Stocking, 1982) and elsewhere (e.g. Modified USLE in Brazil by Chaves, 1991).
The USLE is calculated as:
A = R. K. LS. C. P
where:
|
|
A |
= |
Calculated soil loss in t/ha for a given storm or period. |
|
|
R |
= |
Rainfall energy factor |
|
|
K |
= |
Soil erodibility factor |
|
|
LS |
= |
Slope-length factor |
|
|
C |
= |
Cropping management factor (vegetative cover) |
|
|
P |
= |
Erosion-control practice factor |
Each of the factors can be calculated or estimated using field data (as in the case for R and LS) and from tables or nomograms for all other factors. Novotny and Olem (1994) provide an excellent commentary on this and other methods for estimating or modelling erosion. The USLE is designed for rainfall only and does not handle snowmelt or rainfall on frozen ground. The USLE requires calibration data from standard plot experiments which are widely available in North America and, more limitedly, from other parts of the world.
At the international level, the simplicity and effectiveness of the soil loss approach prompted an important series of experiments in Zimbabwe in the 1950s and 1960s with the primary objective of determining losses of nitrogen, phosphorus and organic carbon from the natural fertility of the soil. According to Stocking (FAO, 1986) who exhaustively analysed this database, it represented (at the time of his report) the "best such database in any developing or tropical country". This work led to the development of the SLEMSA model (Soil Loss Evaluation Model for Southern Africa) for conditions in Southern Africa (Elwell and Stocking, 1982). The value of this approach has further led FAO (1985) to develop an international network, "Network on Erosion-Induced Loss in Soil Productivity", with research partners in Africa, South America and Asia (as of 1995 - internal FAO communication).
FIGURE 7 Erosion measurement plots in the Negev Desert, Israel
Although Wischmeier (1976) has cautioned against extending soil loss models beyond field loss studies, these models are intuitively attractive for predicting erosion over large areas. It should be noted that, due to transport losses (Sediment Delivery Ratio noted above), such erosion estimates apply only to total erosion at source and do not reflect sediment loads (or sediment yield) measured at downstream locations. Such estimating techniques would, if calibrated so that the errors are known, have useful application as a screening tool for the estimation of erosion potential under conditions of similar crop, soil and topographic factors over large areas. Internationally, there appears to be little systematic information on calibration; however the Network on Erosion-Induced Loss in Soil Productivity (FAO, 1991) may eventually provide suitable information.
The most extreme example of use of the soil loss approach for large area estimation is in Brazil where the UFRGS (1991) and Carvalho (1988) used large area maps and satellite data to estimate several of the parameters of the USLE for application at regional scales. The intent was to provide generalized estimates of regional erosion potential for the entire country. While this approach has wide margins of error, it represents a method of screening for major change in erosion potential arising from combinations of agricultural land use, climate, and topography and merits further consideration, especially where plot calibration and in-river sediment monitoring data are available.
Together with the need to further develop screening level models, is the need to generate improved field data on erosion and sediment loss. Hudson (FAO, 1993b) has presented a wide range of simple field measurement techniques that are particularly useful in developing countries.
TABLE 8: Selected values for sediment loss
|
Location |
Land use |
Soil loss (t/ha/yr) |
Comments |
|
Italy |
Wheat Maize Pasture |
5.614 18.767 2.224 |
Average of 7 years' plot data from central Italy. Source: Zanchi, C. 1988. The cropping pattern and its role in determining erosion risk: experimental plot results from the Mugello valley (central Italy). In: Sediment Budgets. M.P. Bordas and D.E. Walling (eds.). IAHS Publication No. 174. Int. Assoc. Hydrol. Sci., Wallingford, UK. |
|
Philippines |
Reforested and agricultural |
22-39.7 |
Sediment yield from nested catchments from 18.8 to 2041 km2 in Luzon. Source: White, S. 1988. Sediment yield and availability for two reservoir basins in central Luzon, Philippines. In: Bordas and Walling (see above). |
|
Morocco |
Arid, grazing |
25.0-59.0 |
Range calculated from sedimentation in three reservoirs having catchment areas from 107-780 km2. (Source: Lahlou, A. 1988. The silting of Moroccan dams. In: Bordas and Walling (see above). |
|
Kenya |
Semi-arid grazing |
79.5 |
Average in 1986 for seven sub-basins within a total area of 0.3 km2. Source: Sutherland, R.A. and Bryan, R.B. 1988. In: Bordas and Walling (see above). |
|
Bolivia |
Andean arid, semi-arid |
5.21-51.8 |
Four basins <1000 km2 in headwaters area. Source: Guyot, J.L. et al. 1988. Exportation de matière en suspension des Andes. In: Bordas and Walling (see above). |
|
United Kingdom |
Agriculture |
1.9 (net) |
Fishpool Farm, UK, area, <1 km2. Source: Walling, D.E. and Quine, T.A. 1992. The use of caesium-137 measurements in soil erosion surveys. In: Erosion and Sediment Transport Monitoring Programmes in River Basins. J. Bogen, D.E. Walling and T. Day (eds). IAHS Publication No. 210. Int. Assoc. Hydrol. Sci., Wallingford, UK. |
|
Lesotho |
Agriculture |
7.8 (net) |
Field measurements near Ha Sofonio, Lesource. Source: See above. |
TABLE 9: Increases in sediment yield caused by land use change (Walling and Webb, 1983; Ostry, 1982)
|
|
Land use change |
Increase in sediment yield |
|
Rajasthan, India |
Overgrazing |
× 4-18 |
|
Utah, USA |
Overgrazing of rangeland |
× 10-100 |
|
Oklahoma, USA |
Overgrazing and cultivation |
× 50-100 |
|
|
Cultivation |
× 5-32 |
|
Texas, USA |
Forest clearance and cultivation |
× 340 |
|
N. California, USA |
Conversion of steep forest to grassland |
× 5-25 |
|
Mississippi, USA |
Forest clearance and cultivation |
× 10-100 |
|
S, Brazil |
Forest clearance and cultivation |
× 4500 |
|
Westland, N. Zealand |
Clearfelling |
× 8 |
|
Oregon, USA |
Clearfelling forest |
× 39 |
|
Ontario, Canada |
Conversion to agriculture |
× 14 |
3. There is a wide variety of deterministic and stochastic models that attempt to simulate the physics of the erosion process. The data requirements for calibration and verification are extremely large. While such models may have certain advantages, especially in terms of the level of detail in which one can simulate alternative farm practices, these are generally unsuitable for developing countries due to their data requirements and the observation that management judgements for farm-level decisions can almost always be made on the basis of more generalized data combined with experience and common sense.
Sediment yield, usually expressed as tonnes per unit area of the basin per year, is the amount of sediment measured at some point in the basin divided by the basin area. It always less than the total erosion due to sediment storage during transport, and is highly variable because of measurement difficulty, the temporal variability of hydrological processes, and changes in land management practices in the basin from one year to the next.
The values of Table 8 demonstrate not only the wide range of values in different climatic and topographic situations, but also the problem of interpreting sediment data because of spatial and time scales. As noted above, the Sediment Delivery Ratio subsumes a wide range of storage processes that operate at the field, sub-basin and basin scales. Sediment yield values are also highly variable in time, with smaller basins subject to greater variability than larger basins. Consequently, published values for sediment yield (t/ha/yr) must be interpreted with great caution. The Italian data in Table 8 are the only plot data and would appear much smaller if they were based on sediment measurements taken at the sub-basin scale. The magnitude of change in sediment yield arising from changes in agricultural land use is shown in Table 9.
Estimates of sediment yield have important economic consequences. In many developing countries the database with which to estimate reservoir life is very limited. According to White (1988) examples of predicted sediment yield in Asia tend to be between two and sixteen times lower than actual measured rates, with the consequence that actual reservoir life is greatly reduced. This is also a problem in northern Africa (Lahlou, pers. comm.). Partly this arises from the use of unreliable prediction techniques and the use of short-term data which usually fail to account for occasional but severe episodes of erosion (e.g. major storm events), and from increased land pressure after construction of the reservoir. White reports that the Magat Dam (Philippines) was constructed on the basis of a planned sediment yield of 20 t/ha/yr when, in reality, the yield was shown to be 38 t/ha/yr - cutting the useful life of the reservoir in half! In the Republic of South Africa, off-site damage (reservoir sedimentation, water treatment, etc.) due to soil erosion was estimated in 1989 to total US$ 37.6 million annually (Braune and Looser, 1989). Because agriculture is a major contributor to sediment yield it is essential that national agricultural organizations account for off-site costs.
In all aspects of monitoring, modelling, or prediction of non-point source management, the scale factor is poorly understood by many practitioners. The scale factor not only pertains to the cost of collecting data for model calibration, but it is also vital to the ability to extrapolate useful management principles that will apply to larger areas. As an example, during the 1970s it was assumed by many modellers who were engaged in the non-point source management programme of the North American Great Lakes, that extrapolation of cause-effect relationships from small-area studies into larger parts of the Great Lakes basin was feasible. It was found, however, that extrapolation became quite uncertain as basin area increased, for a variety of reasons.
TABLE 10: Influence of spatial scale on basin assessment (Ongley, 1987)
|
(A) SMALL AREAS (few hectare to several km2). · Detailed measurement of one land use or land management practice is possible. · Data collected at catchment mouths do not necessarily represent target land use because of the influence of single phenomenon (such as gullies) or random events (specific action of one farmer, for example). · Sediment data are more closely related to erosion at this scale than at larger scales. · Expensive due to the number of monitoring sites necessary. · Difficult to relate sediment and quality data to downstream receiving waters due to rapid information dilution (increasing noise) in downstream direction. · The physics of catchment behaviour can be modelled in deterministic form. · This scale is effective for site-specific management interventions but is too small for interventions involving general questions of land use. · Data reveal synoptic scale effects (i.e. can resolve short-term scale effects within small areas). (B) MEDIUM AREAS (tens to several hundreds km2). · Capable of representing combinations of similar land uses and/or physical information. · Is a scale for which land use policies pertaining to diffuse sources can be effective and for which effectiveness can be assessed. · Physico-chemical data do not relate directly to erosion or source but more to what is transported (information loss problem). · Not a useful scale for erosion process studies, but can be used to assess overall erosional contribution to diffuse source chemistry. · This scale inhibits dominance of site data by a specific phenomenon (e.g. one gully); this scale homogenizes phenomena so that their effects can be modelled by stochastic rather than deterministic processes. · Deterministic models have very large data requirements at this scale. · Is useful for determining impact of land and land-use practices on water quantity/quality, i.e. upstream impact on downstream sites. · Data reveal seasonally variable effects. (C) LARGE AREAS (> several hundred km2). · Too large to understand influence of land use or land management on downstream water quality, or role of natural physiographic units (except for macro-phenomena). · Provides regional estimates of sediment transport but not of erosion. · Can estimate influence on downstream receiving water bodies for physical and chemical variables. · Can be functionally linked to medium areas but not to small areas. · Related to behaviour of trunk rivers. · Spatial elements are best modelled by stochastic models (excluding hydrological modelling). · Not a useful scale for evaluating management options nor for evaluating the a posteriori effectiveness of management actions in upstream areas. · Data reveal seasonally variable effects but may be substantially influenced by long-term lag phenomena. |
The problem of extrapolation has not been definitively addressed in regards to levels of uncertainty. For example, although it is known that control of agricultural erosion is essential to mitigate nutrient (especially phosphorus) loss to receiving watercourses, it is unclear over what distance phosphorus runoff from agriculture has an immediate impact. The question then is, does erosion control in far upstream areas have an immediate benefit for a receiving water that may lie hundreds of kilometers downstream? Alternatively, one may adopt the position that over longer periods of time (say, scores of years) large storms will ensure that most of the eroded sediment is, in fact, transported through the river system and, therefore, one can assume a 1:1 relationship between upstream sediment production and downstream sediment transport. Research by Meade and Trimble (1974) and others shows that over longer time periods, sediment is mobilized differentially along the river basin. Contemporary sediment mobilization in the middle reaches of piedmont rivers of the United States represented sediment that was eroded several decades earlier but which had been deposited during transport into long-term storage as valley floor deposits. Table 10 presents the influence of spatial scale on basin assessment.
The essential point is that time and space scales must be recognized when preparing management plans for erosion control. While the near-term physical benefits of erosion control are likely to be felt quickly in receiving waters, sediment-associated contaminants that are stored in the river basin may take decades to finally be transported out of the basin, notwithstanding the resources spent on upstream erosion control.
These recommendations reflect two very different scales which, in turn, reflect two different types of issues. At the small scale are the recommendations which apply at the farm level and which reflect actions to be taken by the individual farmer after considering the economic costs of his alternatives. At the large scale (river basin scale) are those issues that tend to reflect policy and investment needs of states. This includes issues such as determining the net contribution of agriculture to river pollution relative to other types of polluting sources.
1. Internalizing costs
Although the control of erosion at source in rain-fed agriculture is a major factor in improving downstream water quality and associated ecological impacts, implementation of control measures will only be successful if the farmer can determine that it is in his economic interest to undertake such measures. Therefore, the economic benefits such as maintenance of soil fertility, reduced energy consumption in minimum till situations, etc., relative to economic costs of excessive fertilizer usage and loss of productivity by "mining" of soil capital, must be clearly demonstrated. This implies that agricultural agencies must use a holistic approach to the economics of farming practices.
2. Screening and estimating tools
There is an urgent need to develop simple and robust screening models for use in developing countries to determine the potential for erosion and soil loss at source (field level). The models must contain the ability to game with crop and land management alternatives. Data requirements must be relatively small and easily accessible, require no real-time calibration, and aggregate results on a seasonal or annual basis. Further calibration of the USLE (and its derivatives) and SLEMSA models are encouraged. These models must be developed from the perspective of assisting in farm-level decisions, and not from the perspective of further elaboration of the physics of erosion and sediment transport.
At the large scale, screening methods are needed to develop policy options for erosion control and land use practices at the basin scale. Soil loss models require adequate calibration so that the errors inherent in large scale use, using satellite and large scale map information, are known. Use of sediment loading data from river monitoring sites is an inadequate alternative due to the large sediment losses that occur during sediment transport from field to river.
3. Erosion control
There are no unique solutions to erosion control. Control measures depend very much on the economic situation of the farmer, the degree of importance placed on sediment erosion by environmental authorities, availability of capital, and the state of development of the country. The following control measures are those classified and recommended by the US-EPA (1993). These categories are used in many parts of the world, including developing countries. These techniques also have beneficial effects for conservation of nitrogen and phosphorus in the soil.
|
· CONSERVATION COVER |
Establish and maintain perennial vegetative cover to protect soil and -water resources on land retired from agricultural production. |
|
· CONSERVATION CROPPING |
A sequence of crops designed to provide adequate organic residue for maintenance of soil tilth. This practice reduces erosion by increasing organic matter. It may also disrupt disease, insect and -weed reproduction cycles thereby reducing the need for pesticides. This may include grasses and legumes planted in rotation. |
|
· CONSERVATION TILLAGE |
Also known as reduced tillage, this is a planting system that maintains at least 30% of the soil surface covered by residue after planting. Erosion is reduced by providing soil cover. Runoff is reduced and infiltration into groundwater is increased. No-till, common in North America, is a conservation tillage practice. |
|
· CONTOUR FARMING |
Ploughing, planting, and other management practices that are carried out along land contours, thereby reducing erosion and runoff. |
|
· COVER AND GREEN MANURE CROP |
A crop of close-growing grasses, legumes, or small grain grown primarily for seasonal protection and soil improvement. Usually it is grown for 1 year or less. |
|
· CRITICAL AREA PLANTING |
Planting vegetation, such as trees, shrubs, vines, grasses or legumes, on highly erodible or eroding areas. |
|
· CROP RESIDUE USE |
Using plant residues to protect cultivated fields during critical erosion periods. |
|
· DELAYED SEEDBED PREPARATION |
Any cropping system in which all crop residue is maintained on the soil surface until shortly before the succeeding crop is planted. This reduces the period that the soil is susceptible to erosion. |
|
· DIVERSIONS |
Channels constructed across the slope with a supporting ridge on the lower side. By controlling downslope runoff, erosion is reduced and infiltration into the groundwater is enhanced. |
|
· FIELD BORDERS AND FILTER STRIPS |
A strip of perennial herbaceous vegetation along the edge of fields. This slows runoff and traps coarser sediment. This is not generally effective, however, for fine sediment and associated pollutants. |
|
· GRASSED WATERWAYS |
A natural or constructed channel that is vegetated and is graded and shaped so as to inhibit channel erosion. The vegetation will also serve to trap sediment that is washed in from adjacent fields. |
|
· SEDIMENT BASINS |
Basins constructed to collect and store sediment during runoff events. Also known as detention ponds. Sediment is deposited from runoff during impoundment in the sediment basin. |
|
· STRIP CROPPING |
Growing crops in a systematic arrangement of strips or bands across the general slope (not on the contour) to reduce water erosion. Crops are arranged to that a strip of grass or close-growing crop is alternated with a clean-tilled crop or fallow. |
|
· TERRACING |
Terraces are constructed earthen embankments that retard runoff and reduce erosion by breaking the slope into numerous flat surfaces separated by slopes that are protected with permanent vegetation or which are constructed from stone, etc. Terracing is carried out on very steep slopes, and on long gentle slopes where terraces are very broad. |
Note however, that in tropical areas a number of these measures may produce situations that enhance the breeding of disease vectors as a result of ponding of water or reducing of water velocity in waterways.
TABLE 11: Annualized cost estimates for selected erosion management practices in the USA
|
Practice |
Rank |
|
Grassed filter strips |
1 (least cost) |
|
Cover crops |
2 |
|
Strip-cropping |
3 |
|
Conservation tillage |
4 |
|
Reforestation of crop and pasture land |
5 |
|
Diversions |
6 |
|
Permanent vegetation on critical areas |
7 |
|
Terraces |
8 |
|
Sediment ponds and structures |
9 (most cost) |
Data from Chesapeake Bay (USA) installations indicate the following ranking of costs of erosion control measures (Table 11). Cost factors and rankings will vary greatly, especially in parts of the world where labour costs are less and where the economic benefit is factored into the ranking system. For example, grassed filter strips may have no economic benefit and are replaced with alternative practices such as herbaceous cover mixed with fruit trees.
Erosion in tropical areas has a unique set of problems. Marginal agricultural practices such as slash and burn on highly erodible tropical soils and tillage on steep tropical slopes lead to highly unstable situations which erode quickly during rainy seasons. Similarly, deforestation in tropical lands, either for agriculture or for timber, tends to leave a highly erodible land surface. In many tropical countries erosion from deforested areas is having a devastating influence on coastal zones including destruction of coral reefs far offshore. Poor land management practices such as overgrazing, especially on hill lands, always leads to serious erosion problems which are difficult or impossible to remedy due to the scale of damage and cost of reconstructing hill-sides. Erosion control measures for semi-arid areas are extensively described by Hudson (FAO, 1987).
While recommendations to control these abuses are self-evident, the fundamental causes lie often in national economic goals that are incompatible with environmental and water quality objectives, and in social policies that do little to contain destructive marginal agricultural practices.
