The measurement,estimation and monitoring of soil erosion by runoff at the field scale: Challenges and possibilities with particular reference to Britain

Abstract

Soil erosion is widely acknowledged as a global problem but attempts to measure and estimate its significance are frustrated by our inability to develop reliable, cheap and easy methods of assessment. The limitations of qualitative methods such as GLASOD, errors and inaccuracies inherent in modelling based on small-scale plot experiments, and problems with ¹³⁷Cs approaches, mean that alternative strategies are required. For runoff-related erosion on arable land we propose the use of a well-tried estimation technique: volumetric measurement of rills, gullies and fans. Amounts of wash and interrill erosion can also be estimated. This approach allows for the estimation of erosion rates at the field scale, rather than relying on extrapolations from plot-based data. Measurements are based on sampling the population of rills and gullies and can be adapted to the aims of the project for ‘broad-brush’ or detailed data. Monitoring of large areas to produce regional assessments of erosion risk is frequently required and volumetric estimates provide these data. Thus predictions of the extent, frequency and amounts of erosion can be made and the vulnerability of particular crops becomes clear.

Keywords

Soil erosion arable land erosion by runoff measurement of erosion erosion monitoring

I Introduction

In this paper we refer to visible soil erosion, the presence of rills and gullies within a field, and their associated deposits. The transfer downslope of soil due to tillage may be included in some estimates of soil erosion (Van Oost et al., 2007) but this process is not considered here. When we first started to assess erosion it was the visible evidence of erosion that initiated our interest and the features we saw in the field, soil held up by field boundaries or deeper valley floor colluvial deposits could be as well explained by visible erosion processes as by tillage (Evans, 1992a).

There are two aspects to assessing soil erosion: assessing it at a site (plot, field) and assessing it across a landscape, catchment or larger areas. Plot experiments to study the processes driving soil erosion (Laflen and Moldenhauer, 2003) have been the main way of assessing erosion at a site by measuring the mass of soil eroded from a plot under varying measured rainfall, soil and crop conditions. Such an approach is open to criticism (Boardman, 2006; Evans, 1993a; Evans et al., 2016). Models to assess erosion across the landscape are derived from those plot experiments (Panagos et al., 2015), often with the aid of Geographical Information Systems (GIS) to enable extrapolation (Borelli et al., 2017). Such assessments have been important in bringing soil erosion to the fore in protecting soil health (ELD Initiative, 2015; FAO, 2015; Montanerella, 2015; evidence incorporated into e.g. Defra, 2018).

From such erosion model assessments soil erosion is widely acknowledged as an environmental problem of global significance. Emphasis is placed on ‘accelerated erosion’ resulting from human exploitation of the landscape mainly by agriculture giving rise to rates of erosion 10- to 100-fold in excess of background ‘natural’ levels, which approximate rates of soil formation (Montgomery, 2007). The widespread non-sustainable use of soils means that their major functions, including that of providing a basis for food production, is seriously threatened in the medium term. Estimates vary depending on assumptions about rates of soil formation, rates of degradation, including erosion, developments in crop science and population pressures on the productive capacity of the land, including the pressures of urban sealing. Many future scenarios also depend on wise use of the land, which has not always been the case in the past (Boardman et al., 2003).

However, we do need to attain a perspective on erosion. Is erosion as serious as many think? Possibly, but we still do not know. Soil erosion is often equated with soil degradation, but the two topics do need to be better defined and distinguished (Evans, 2002a). It is likely the best information on erosion worldwide is still GLASOD (Global Assessment of Soil Degradation; Oldeman et al., 1991), ‘most modelling approaches have so far failed to come up with more useful assessments’ (FAO, 2015: 44). GLASOD pulled together many researchers’ qualitative knowledge on the topic (Oldeman et al., 1991). However, there is evidence that GLASOD is a very imperfect predictive tool (Sonneveld and Dent, 2009). Van Camp et al. (2004) are particularly critical and suggest that the GLASOD approach should be abandoned. Comparison with maps and data from Spain show that assessment of water erosion was poor (Sanchez et al., 2001). A more recent global survey, the Status of the World’s Soil Resources (FAO, 2015) quotes soil erosion rates taken from sources that use ‘a relatively simple modelling approach combining information on soil type, land use, topography and climate’ (FAO, 2015: 101 and Figure 6.1: 102, based on Van Oost et al., 2007) or ‘a summary by Den Biggelaar et al. (2003) suggest that global rates of erosion are between 12 to 15 tonnes ha^-1 yr^-1’ using evidence from plot-level and GIS data (FAO, 2015: 175). A major problem with such an approach is that it assumes erosion occurs across the landscape. Obvious visible erosion such as rills and gullies and their deposits do not occur, in our experience, across the whole landscape but are in particular fields (Boardman, 1998: Figure 1; Evans, 2002b, 2017; Fischer et al., 2017; Prasuhn, 2011). Erosion assessed at plot or field scale should not therefore be extrapolated across the whole landscape, as this will overestimate the problem (Evans et al., 2017; Garcia-Ruiz et al., 2015). It seems likely that if there is very little visual evidence of erosion in farmers’ fields, such as wash erosion (Evans, 2017), amounts eroded are small, much less than those predicted by models.

Figure 1.

Location of traverses at Whiteway Bottom, South Downs, autumn 1987.

The need for estimates of rates of erosion seems self-evident. At the very least, if we do not need to have exact figures, we need to know where the problems lie so that mitigation measures can be put in place. Govers et al. (2017: 47) argued that ‘simple visual observations on the presence of rills and gullies or wind deflation areas are clear indications that the implementation of conservation measures is necessary’. This might suggest that a qualitative assessment is sufficient. However, they have previously argued that erosion rates should be ‘reliably’ quantified in order to assess the necessity for conservation measures (2017: 47). They rightly then go on to point out the difficulties of obtaining reliable erosion rates.

In certain areas of the world, erosion rates threaten crop yields. Although the relationship between erosion and yield may have been overstated (see Bakker et al., 2004), it remains a threat in areas of thin soils and heavy usage. For issues of off-site damage by runoff and muddy floods there is no clear correlation between high erosion rates and damage (Boardman et al., 2019). However, there is ample evidence for damage to both fields and off-the-site recipients of eroded sediment as a result of severe erosion (e.g. Boardman, 1988; Boardman et al., 1994, 1996, 2009; Cerdan et al., 2002; Edwards and Owens, 1991; Evans, 1996a; Evans and Boardman, 2003; Harold and Edwards, 1972; Hjelmfelt et al., 1986; Larson et al., 1997).

A major reason for making field-scale estimates of erosion as advocated here are the limitations or misuse of alternative approaches. These are well documented and not the focus of this paper. However, we need briefly to justify our field-scale approach.

The limitations of a modelling approach are exemplified in the GCTE soil erosion model comparison exercise using common data sets and a series of frequently used models at the field scale (Boardman and Favis-Mortlock, 1998). A similar exercise evaluated erosion models at the catchment scale. A major source of concern was the need to calibrate models and the finding that ‘it is not sure that the model will have a good predictive quality if the event lies outside the range of calibration events’ (Jetten et al., 1999: 538). However, ‘models are often improperly calibrated i.e. model parameters are set to values that are not appropriate for the location under consideration’ (Govers et al., 2017: 47).

Estimates of sediment yield in rivers are not equivalent to erosion on hillslopes. This is because of storage of sediments between field and watercourse and non-field sources, such as erosion of river banks. Within-catchment budgeting of sediment sources and sinks has been extensively investigated; for example, Trimble (1983). River yields are also significantly affected by storage in dams (Syvitski and Kettner, 2011). Deposition in dams and retention ponds is another approach to estimation of sediment yields. Both approaches are subject to errors (Verstraeten and Poesen, 2002).

Experimental plots, typically 22 m × 2 m, are frequently used to measure erosion rates over wide areas. Such extrapolations can lead to over-estimates (Boardman, 1998). Plot measurements are a useful tool in studies of rill and interrill erosion and in comparisons of relative differences between land uses and agricultural practices (Cerdan et al., 2010). Most erosion models are based on the results of plot experiments particularly those models derived from the very widely used Universal Soil Loss Equation (USLE) (Laflen and Moldenhauer, 2003). Indeed, in the USA it is the primary method of assessing soil erosion (NRC, 1986). Models have rarely been tested against real-world field-scale data (Evans and Brazier, 2005), and their results do not compare well, and applications of model rates may result in spurious output (Evans and Boardman, 2016a, 2016b) greatly overestimating the severity of erosion.

Great hope was placed on the use of ¹³⁷Cs as a tool for mapping and quantifying amounts of erosion and deposition (Walling and Quine, 1990, 1991). Unfortunately, the basic assumptions behind the technique have been questioned (Parsons and Foster, 2011). Comparison of field-scale measurement at many sites in the UK with ¹³⁷Cs predictions show gross disparities, with consistent over-prediction of erosion rates by ¹³⁷Cs (Evans et al., 2017).

Zachar (1982: 247) considered the results of field measurement to be ‘important aids in the assessment of splash and rill erosion under given conditions, because they cannot be simulated under laboratory conditions, nor can they be derived by theoretical analysis, no matter how well established’. What we refer to as ‘wash’ (see Discussion) may be what Zachar considered could not be simulated.

In this paper we address the assessment of rates of water erosion at the field scale. We confine our attention to cultivated land and do not consider erosion on grazing land. One of us (RE) accepted early in his work on erosion that volumetric assessments of erosion were the easiest and best way to tackle the problem of assessing rates (Evans and Boardman, 1994). Furthermore, when comparing volume of soil eroded (320 m³) from a gully compared to volume deposited (304 m³) comprising what appeared to be a full suite of deposits from sands and gravels to organic rich silts, 95% was accounted for in deposits (Evans and Boardman, 1994). From this work it was considered that volumetric assessments of water erosion were an acceptable way to assess erosion and rates of erosion, and that the rates were of the right order of magnitude. When discussions were held on how erosion should be assessed if a national monitoring scheme was established (Evans, 1988), it was agreed that volumetric measurement of erosion within identified eroding fields should be taken, rather than a plot and USLE approach. Boardman used the field-based approach to locate eroded fields extensively and has volumetrically assessed erosion in those localities (Boardman 1983, 1988, 1990, 1991, 2003; Boardman et al., 1996, 2009). In Britain, such assessments (Evans et al., 2016) have allowed erosion to be placed in perspective. German and Swiss researchers have also assessed and monitored erosion based on visual and volumetric measurements of water erosion (Fischer et al., 2017; Prasuhn, 2011). In our view, although such assessments are comparatively rare in comparison with the use of model assessments of water erosion, they give much more realistic estimates of the extent of water erosion and erosion rates.

The need for field-scale measurements of erosion is compelling: a recent comparison of sediment yields with ‘measured and predicted hillslope erosion rates’ completely ignored field-scale measurements and preferred erosion rates from experimental plots and model outputs (Vanmaercke et al., 2012: 586).

In this paper we present and review a methodology for the estimation of erosion rates at the field scale based on estimates of rill and gully volumes, including ephemeral gully volumes, that is, a gully that can be ploughed out (ISSS, 1996). We discuss rates of wash and interrill erosion and errors inherent in the technique. We give examples of its application. We then move from the scale of the individual field to that of the landscape and the need for surveys and monitoring programmes of different spatial and temporal scales. Methods of assessment of erosion at the field scale are highly relevant to the efficacy of monitoring schemes at the regional scale. We focus on erosion by water (runoff), and on arable land, these being the most frequently occurring situations in temperate landscapes within the domain of land degradation.

II Field-scale measurement of soil erosion

The approach we propose is to base estimates of erosion at the field scale on volumetric measurement of rills and gullies and ephemeral gullies. The lengths of rills are measured by tape, from maps, from air photographs or by pacing. Cross-sectional areas are measured by sampling (e.g. every 10 m along the channel). The sampling interval depends on the purpose of the survey and the amount of detail and accuracy required. We illustrate this principle in the examples in Section III. Estimation of volume of soil deposited is an alternative or ancillary approach: fan areas and sampled depths provide these data. Losses of soil from the field must be noted or taken into account. This approach has been used by many researchers and we give selected examples in Tables 1 and 2. However, few workers have considered the issues of errors, reproducibility and implications for regional monitoring schemes (Evans and Boardman, 1994) and we address this issue in Section IV.

Table 1.

Selected reports of field-scale estimates of erosion rates.

Country	Reference	Notes
Australia	Murphy and Frewin (1993)	Winter erosion in one field; rill volumes estimated
Belgium	Govers (1991)	86 fields monitored for winter erosion for three years; volumes of rills estimated
	Vandaele and Poesen (1995)	Monthly visits to all fields in two small catchments; rill and ephemeral gully volumes estimated
France	Auzet et al. (1993)	20 small catchments monitored for one winter; rill volumes estimated
	Ludwig et al. (1995)	35 small catchments monitored for three years; rill and ephemeral gully volumes estimated
Germany	Steinhoff-Knopp and Burkhard (2018)	Twice a year surveys of 86 fields for 17 years; rill and ephemeral gully volumes estimated
Sweden	Alstrom and Akerman (1992)	Winter erosion over three years 935 fields surveyed; ephemeral gullies and rills volume estimated
Switzerland	Ledermann et al. (2010)	One-year period, three study areas: volumes of rills and ephemeral gullies estimated*
	Prasuhn (2011)	203 fields monitored for 10 years; volumes of rills and ephemeral gullies estimated*

*Also estimated sheet erosion (wash).

Table 2.

Selected reports of field-scale estimates of erosion rates: UK examples.

Area	Reference	Notes
Cambridgeshire	Evans and Morgan (1974)	Storm in May 1973. Erosion in spring-sown barley and sugar beet; volume of deposits measured.
North Norfolk	Evans and Nortcliff (1978)	Winter erosion; large field; estimate of rill and gully volumes
West Midlands	Reed (1979)	1967–76 survey of three parishes; some volumetric data quoted
West Sussex	Boardman (1983)	One field; measurement of large fans
Somerset	Colbourne and Staines (1985)	Nov–June erosion; 40 fields; volumes of rills and ‘traces’ measured
17 Localities in lowland Britain	Evans (1993b); Evans et al. (2016)	Erosion assessed in >1700 fields
South Downs	Boardman (1988)	Winter erosion; one-year period; c.100 fields rill/gully volumes estimated
South Downs	Boardman (2003)	Review of 10-year monitoring of c.100 fields

The justification for this approach is that in almost all cases of erosion, rills, ephemeral gullies and (more rarely) gullies account for most of the soil lost in an eroding field (e.g. Govers and Poesen, 1988). Furthermore, plot-based studies do not consider gully erosion or rills and gullies in valley-bottoms or topographic depressions and therefore underestimate erosion rates in many landscapes. A further justification is that field measurement can be rapid – again depending on the amount of detail required. In our examples in Section III, we have indicated the time taken for the fieldwork. Ground-level photographs of rills and gullies may be used as a record of length and size; subsequent analysis shows them to be a reliable means of estimating soil losses (Watson and Evans, 1991).

The comparison of amounts of soil lost from rills and that deposited within the field allows estimates to be made of losses from the field. The losses are texture-dependent with higher values on fine and lower on coarse-grained soils (Evans, 2002b, 2006; Evans and Boardman, 1994).

Simple measurements of the volume of deposited soil, together with data on particle size, allow us to estimate soil losses from an erosion site. Thus, at Albourne, a comparison of uneroded soil profiles and the particle size of deposited soil suggests that ≥145 t of fine material passed through the fans and was lost from the site (Boardman, 1983).

In the processing and presentation of field-measured data there are several challenges. Erosion data are usually related to the area of which it is representative. The choices are the area which provides runoff to create the erosion feature (often referred to as the catchment); this can be as small as the area contributing runoff to a tractor wheeling, or that part of the field directly contributing runoff, or the area of the field in which erosion occurs, or the wider landscape (Evans, 2002b, 2017), often arbitrarily defined as a sampling transect, a soil landscape or an administrative region. Table 3 is an example of the contrasts in amounts eroded related to the size of the area selected in wet (1983) and drier (1984) years, and reflects not only erosion severity in those years but also the extent of erosion. In wetter years a greater number of fields erode (Evans, 1996a: Fig. 8.1; Evans and Brazier, 2005: Fig. 3).The rate for the monitored transect gives a mean value for all non-urban land in that monitored transect.

Table 3.

Mean amounts eroded within fields and transects in Shropshire in 1983 and 1984 (from Evans and Boardman, 1994).

	Amount eroded 1983 m³ha^-1	Amount eroded 1984 m³ha^-1
Area within field directly affected by erosion	69.41	40.99
Contributing area within field	16.22	3.57
Field	4.06	1.41
Monitored transect	0.63	0.05

Erosion rates are generally expressed as a mean rate. However, many fields in a locality will not show visible signs of erosion, and of the eroded fields some will have eroded far more severely than others. The distribution of erosion rates is remarkably skewed (Boardman and Favis-Mortlock, 1999; Evans, 1998, 2006); median erosion rates are far lower than mean rates. Median rates more realistically reflect the severity of erosion in a soil landscape.

III Examples of estimation of erosion

As noted above, RE accepted that although estimates of amounts eroded in channels and deposited in fans varied somewhat, they were within an acceptable range (c.0.5–c.2.0) and gave a good indication of the magnitude of the erosion event (Evans and Boardman, 1994). At that time, in the early 1980s, it was not known how long it would take to carry out volumetric assessments, but it was considered that even rapidly gathered information on the ground (i.e. in the field), was probably more realistic than modelled estimates for which at that time there was no validation by field assessments. Here we give further justification for field-scale assessments of erosion and indicate that the time taken to carry out the measurements is not prohibitive.

1 Whiteway Bottom, South Downs

Estimates of erosion on around 100 fields were made in autumn 1987 on the eastern South Downs (Boardman 1988, 1998). In all cases, estimates were based on volumetric measurements of rills, ephemeral gullies and gullies and occasionally of fan volumes. As part of the survey, more detailed measurements were made in Whiteway Bottom (GR TQ 395056), mainly on fields prepared for winter cereals (Figure 1). Major erosion events were on 7 and 20 October, with minor events on a further 11 days until 11 November. Over 60 mm of rain fell in the area on 7 October. Detailed surveys were made of the relevant fields on 15, 18, 22, 24, 30 October and 4 November. This constitutes about four working days. Details of the traverses are shown in Table 4 and totals for four most severely eroded fields are given. Table 4 shows considerable variation in rates of erosion when considering small areas within fields. The traverses include both valley side and valley bottom (ephemeral gully) losses of soil. They do not include wash and interrill losses. In total, they indicate that >2000 t of soil was lost from the four fields. The purpose of the surveys was to provide evidence for compensation claims resulting from damage to Breaky Bottom vineyard and farmhouse from muddy floods originating in Whiteway Bottom. The surveys and analyses resulted in several consultancy reports and a major academic paper (Boardman, 1988).

Table 4.

Erosion rates on four fields at Whiteway Bottom, South Downs, October 1987.

Field	Traverse: vs/vb	Land use	Area (ha)	Soil loss (m³)	Rate (m³ ha^-1)
1	1–2 vs	wc	2.56	16	6
	4–5 vs	wc	2.04	69	34
	6–7 vs	wc	4.28	403	95
	8–9 vs	wc	2.32	34	15
	E–F vs	wc	2.08	426	205
	S–T vb	wc	0.07	29	444
Field 1 total			13.35	979	73
2	G–H vs	wc	1.44	68	47
	J–I vs	wc	1.08	78	73
	K–L vs	wc	2.00	133	66
	C–D vs	wc	0.92	73	80
	M–N vs	wc	1.10	17	15
	A–B vs	wc	1.00	84	84
	O–P vb	wc	0.09	33	384
	R–Q vb	wc	0.09	41	461
Field 2 total			7.72	527	68
6	3–4 vs	cult	2.40	74	31
	18–19 vs	cult	1.80	69	39
	20–21 vs	cult	1.72	9	6
	U–V vb	cult	0.11	170	1541
	10–11 vb	cult	0.09	30	350
	w–x vb	cult	0.13	109	838
Field 6 total			15.2*	461	30
7	12–13 vs	cult	2.68	17	6
	16–17 vs	cult	2.60	1	0.5
	14–15 vb	cult	0.15	162	1083
Field 7 total			5.28*	181	34

*Parts of these fields were not affected by erosion.

cult: cultivated; ws: prepared for winter cereals; vs: valley side; vb: valley bottom.

2 Fairlight Place, near Hastings, East Sussex

Serious erosion occurred on an 8.08 ha field at Fairlight on 17 June 1987 (GR TQ 847115). The relative relief of the field was 38 m, slope length 350 m and mean gradient was 6^o. The field, recently planted with maize, was visited on two days before the end of June, estimates were made, and data collected (Figure 2). Rainfall records from Fairlight Cove, 3 km east of the field, show a fall of 18 mm on 17 June. The purpose of the survey was to investigate erosion on maize since, at the time, there was little UK data on what was already perceived as a growing problem. The site was on Curtisden association soils (Jarvis et al., 1984), which were classified as being at small to moderate risk of erosion, with locally higher risk (Evans, 1990a). The soil is dominated by silt and fine sand. The surface had been ring rolled and mean bulk densities of 1.22 gm cm^-3 were recorded for the upper 5 cm of soil. Three sets of wheelings were evident on the field with a post-drilling set largely lost to the parallel 0.75 m spaced rills (Figure 2). Crusts ≤4 mm thick had formed with fine sand and silt washed into coarser material.

Figure 2.

Erosion on maize field, Fairlight Place, Hastings.

Cross-sections of rills were measured on four traverses of the field at 30, 80, 130 and 204 m (Table 5). The overall estimated mean soil loss for the field was 21.3 m³ ha^-1. Erosion rates were higher on slope convexities. The estimate of erosion along the lower convexity (Figure 2), based on measurements of 19 rills in an area of 150 m², was 98 m³ ha^-1.

Table 5.

Estimates of erosion at Fairlight Place.

Traverse	No rills per traverse	Soil loss per sector (m³)
304 m*	60	23.8
204 m	91	33.4
130 m	108	48.7
80 m	141	27.7
30 m	132	38.8
Total		172.4

*Estimate for sector 304 m is based on assumption of rills being 20% smaller than in previous traverse.

IV How reproducible are the estimates of erosion?

As noted above, earlier work suggested that estimates made by different researchers gave acceptable results. Here, we confirm that approach by a re-examination of previously published data.

Estimates of erosion at Whiteway Bottom have already been described. Figure 1 shows the location of traverses along which rill cross-sections were measured. The estimates, along one traverse, with a two-week interval between measurements, show a 0.5% difference (Table 6).

Table 6.

Field 1. Whiteway Bottom comparison of traverse E–F (see Figure 1).

	Length of traverse (m)	Rill length (m)	Area(ha)	Nos rills	Av. rill spacing (m)	Total soil loss (m³)	Rate (m³ ha^-1)
16 Oct 1987	160	130	2.08	89	1.8	426.5	205
30 Oct 1987	160	130	2.08	104	1.5	424.8	204

The number of traverses is clearly subject to the time and resources available and also the purpose of the survey as to how much detail and accuracy is required. In many situations, it is reasonable to undertake one traverse across the mid-point of the eroded slope and estimate total erosion based on the mean rill length. To test this approach, rill-volume data were collected for soil loss on five plots with different treatments on the South Downs in 1985–86. The plots were ∼180 x 6 m (Robinson and Boardman, 1988). Table 7 shows estimates of soil loss from one mid-point traverse (column G) compared with estimates from traverses at 10 m intervals along the plots (column H). Figure 3 shows a statistically significant correlation between a mid-point traverse estimate and a more detailed set of measurements with the latter set giving a 25% lower estimate of erosion This approach is particularly appropriate where parallel or sub-parallel rills are found (e.g. Figure 2).

Table 7.

Estimation of erosion rate from mid-point of slope traverse.

Plot number	APlot area (ha)	BMax. rill length on plot (m)	CMid-point of rills (m)	DSample point (nearest sample to mid-point)	EX-section of rills at sample point (m²)	FEstimated soil loss (B x E)(m³)	GEstimated soil loss based on mid-point traverse(F/A)(m³ ha^-1)	HEstimated soil loss for plot based on all traverses(m³ ha^-1)	% difference
1	0.144	65	32.5	32	0.0008	0.052	0.36	0.24	+50
2	0.108	180	90.0	90	0.0042	0.756	7.00	6.98	+0.3
3	0.126	165	82.5	85	0.0172	2.838	22.52	15.12	+48.7
4	0.162	175	87.5	85	0.0431	7.543	46.56	32.39	+43.7
5	0.220	175	87.5	85	0.0193	3.378	15.35	18.75	-18.1
Totals							91.79	73.48	+24.9

Figure 3.

Soil loss from a comparison of one mid-point and five traverses on an experimental plot.

V Discussion

1 Rates of erosion

Rates of rill and gully erosion measured in the field can be related to factors noted in the field (crop type, topsoil texture) and from maps (location-valley floor, depression or slope; length of slope, slope gradient, contributing area; soil type) and rainfall measured at the nearest location, to explain the occurrence and severity of erosion (Evans, 1990b, 1992b, 1992c, 1993b, 1995, 1996a, 1996b, 1998, 2002b, 2005, 2006). As would be expected from plot studies and the factors incorporated into the USLE, rainfall amount, soil type and crop type controlled the severity and extent of erosion within a landscape from year to year.

Rates of erosion assessed in the field by different researchers in different countries vary in what seems a logical way according to climate (i.e. higher where more rainfall; soil type and higher where more erodible soils) (Evans, 2002b) and crop type – higher where crops (e.g. maize) are more vulnerable (than e.g. cereal) to erosion (Evans, 1995). Results from field-scale monitoring in Switzerland (Prasuhn, 2011) and England compare favourably with regard to occurrence and rates of erosion, and differences can be explained by differences in proportions of crops grown (Evans, 2013). Generally, erosion measured in farmers’ fields was considerably less than on plots. Rates modelled using plot data were higher still. Distribution of rates measured on plots and in the field are similar, both being markedly positively skewed (Evans, 1995).

2 Wash

A problem when assessing erosion in the field is how to account for what is termed in much of the erosion literature ‘interrill erosion’, especially literature dealing with plot experiments and the USLE or its revised versions. Here we use the term ‘wash’ (i.e. the flow of water across the surface of the land which will carry soil fines – silt, clay and organic particles, dislodged by rain drops or runoff of low velocity over saturated unstable topsoils) in a way similar to that defined by Govers and Poesen (1988). Wash takes place across a surface leaving little or no visible evidence of its action. For example, it can take place across a grass field/setaside/grass margin, as well as over a bare soil surface and in tractor wheelings –‘tramlines’ (Evans, 2013: Figs 1 and 2; Evans et al., 2017: Fig 2c). The clearest signs are splashed sand grains (fines washed away) and ‘flow lines’ of particles of fine sand, often with dark coloured organic particles. Interrill erosion generally shows as microrills and micro-deposition fans as well as very shallow braids. In our terminology these features would be classed as rill erosion. Rills are often closely spaced and splashed grains are easily transported into them. We recognise that most researchers have not distinguished between these processes and have lumped them together under terms such as ‘wash’ or ‘interrill’ erosion. There does seem general agreement that rates of wash or interrill erosion are far lower than those associated with rilling and gullying. This is confirmed by literature studies; for example, Zachar (1982: 145) considered ‘interrill erosion to represent 10–30% of rill erosion’, and it accounted for 22% of erosion in an intensely monitored field in Belgium (Govers and Poesen, 1988) and ∼25% in fields in Switzerland (Prasuhn, 2011). However, proportions vary with time and spatial factors. Evans (1990b: 216), in reviewing various field and plot assessments in the UK, suggested that ‘wash on most fields under arable crops will transport <0.3 m³ ha^-1 yr^-1’, and reiterates (Evans, 2006) that assumption when discussing sediment delivery to rivers in Britain, namely that wash transports 0.1–0.3 t ha^-1 yr^-1 to rivers. In a catchment in Norfolk, England, wash occurs across the landscape regardless of soil type or crop cover (Evans, 2017). Estimates of erosion in Britain, and presumably elsewhere, need to factor in wash. However, because wash occurs across the landscape (Evans, 2012, 2017) it is important to take into account that not all sources of sediment are farmers’ fields: roads, tracks and eroding river banks are also important, as they will be elsewhere, and wash from cultivated land will be one of several sources of sediment reaching rivers (e.g. Biddulph et al., 2017; Collins et al., 2010, 2013). However, because wash might occur in every field, models might be more successful predictive tools for wash erosion (Evans et al., 2017).

3 Monitoring

There is a need to put erosion into perspective. It seems to be assumed that the same intensity of erosion occurs across the landscape, but it does not. As we have noted above, visible channel erosion and its deposits are not seen in every field. By quoting a mean erosion rate across, for example, a monitored landscape (e.g. Table 3), which takes into account the number of fields showing visible erosion as well as those without visible erosion, the importance of erosion is put into perspective. This is further shown in a German study. Fischer et al. (2017) reported a very strong relationship between two ways of assessing erosion in farmers’ fields in Germany as assessed by ranking visual estimates based on land use and visible erosion from aerial photographs and rates estimated by using USLE factors. As expected, high assessments compared well with high rankings and low ones with low ones. Although some fields were predicted to have high mean rates of erosion (Table 2; >7.2 t ha^-1, ≤42 t ha^-1), in ascending order of severity: canola (oilseed rape), beet, maize, soy bean and hops, these crops cover only 2440 (30.1%) of 7942 monitored fields, of which 2035 fields were drilled to maize. The remaining 5502 fields have predicted rates ≤1.4 t ha^-1, of which cereals (mean rates <0.6 t ha^-1) covered 1983 fields (25.0%). It is noteworthy that relativities between mean erosion rates are similar to those assessed in the field (Evans, 2013). Grass fields (mean predicted rate 0.3 t ha^-1) covered a further 3463 (43.6%) fields. For the whole number of fields assessed the mean rate of monitored fields was 2.7 t ha^-1, a less impressive figure and, assuming a bulk density of 1.3, equivalent to a surface lowering of 0.2 mm yr^-1.

Assessments from individual fields cannot be applied across the landscape or catchment. Other than in the German example (Fischer et al., 2017), few researchers have attempted to assess soil erosion across a large area of land on a field scale, that is, how much erosion is there across a landscape, how extensive is it, and how often does it occur in the same field? Most models incorporated into GIS (Borelli et al., 2017; Panagos et al., 2015) appear to assume unvalidated high rates of erosion across landscapes (Evans and Boardman, 2016a, 2016b). Thus, if the emphasis is always on erosion rates, and their impact on crop productivity, the problem of soil erosion may be overstated. Severe erosion in a field might occur in one year, and that is often the incident the policy-maker will focus upon, but rill and gully erosion may not occur extensively every year, nor in the same field every year. Thus, 1987 was a notably severe year for erosion in parts of England (Boardman, 1988; Evans, 1996a) and perspective is needed to put such years in context. In Britain, in the early 1980s, after much discussion it was agreed that rather than use the USLE to assess erosion, a field-based scheme should be conducted It was only after a field-based monitoring scheme was put in place that the size of the erosion problem was properly assessed and such a field-based scheme allowed modelled results to be assessed (Evans and Boardman, 2016a, 2016b; Evans and Brazier, 2005; Evans et al., 2017).

Although it is important to target monitoring at those areas that are most vulnerable to erosion (e.g. Fischer et al., 2017), it is best if a variety of soil landscapes is monitored to obtain the full range of rates and the occurrence (extent and frequency) of erosion from year to year on different soils under a range of crops, with different rainfall patterns each year. The monitored localities should be visited at those times when the land is most at risk of erosion, that is, when soils are bare and especially when topsoils are saturated (in winter in Britain) and erosive rainfalls (>10 mm in Britain) occur. Visits should be made shortly after erosive rains have fallen (Evans, 2017) but often that is not possible and visits should then be made at the end of the season when erosion is most likely to happen; for example, in Britain, at the end of winter to monitor erosion in autumn-drilled crops and in early summer for spring-sown crops. If, for reasons of economy, only one visit can be made as was the case in Britain for the initial monitoring scheme, the visit could be made in late summer/early autumn when erosion features are still apparent in harvested autumn-sown fields as well as in spring-sown crops.

To ascertain where to monitor is a crucial question. In England and Wales that was achieved by ad hoc mapping of eroded fields (Evans, 1996a; Evans and Cook, 1986) which identified where erosion was most likely to occur. Thus, by the early 1980s it was known that erosion occurred widely in England, but the size of the problem was not known; was it a serious problem? That is why the National Monitoring Scheme was set up. It was already considered that the USLE approach was inadequate (Evans, 1980a, 1980b) for how realistic were its predictions, as the model was not validated, and the amount of data needed to run the model was extensive and needed to be collected, which would take much time, effort and expense. The 1:250,000 soil map of England and Wales was published in 1983 (SSEW, 1983). This, in association with the known distribution of eroded fields and advice from field staff of the Soil Survey and the (then) Agricultural and Advisory Development Service was used as the base for selecting 17 localities to be monitored. The monitoring scheme confirmed the ad hoc distribution of eroded fields and the risk of erosion of lowland soil associations was assessed (Evans, 1990a). Later work, carried out by an independent consultancy, confirmed that the associations considered most at risk were correctly identified (Evans et al., 2016). Other monitoring schemes were initiated (Evans, 2005), one on a statistically sound basis recording erosion at National Soil Inventory Points. Although over three years 257 points were visited at the appropriate times of year, the number of eroded fields located did not allow anything more useful than a mean rate of soil erosion for England and Wales and did not allow erosion rates and extent to be related to soil landscapes or crops grown in fields within those landscapes. Such a scheme was considered too expensive and time-consuming, The original monitoring scheme, regardless of its drawbacks, did provide good quality data, and at moderate cost:

Over the 5 year period 1982–86 the number of (eroded) fields examined each year ranged from 135 in 1984 to 560 in 1983. On average it took 5.1 (range 4–7) working days to interpret the aerial photographs and 23.3 days (16–32.25) days to carry out field work. To calculate the amounts of soil eroded in fields and to analyse the data has taken so far a further 212 days, 59.8% of the total time spent, excluding writing-up time. (Evans, 1992b: 85)

A similar project design could be used elsewhere.

Measuring/estimating amounts of soil eroded in the field can be time-consuming and costly; a technique to measure volumes eroded in gullies similar to that described here is about 10 times less expensive than using a sophisticated LiDAR technique (Castillo et al., 2012). Such studies not only require expensive equipment but may also need fixed points in a field to be imaged repeatedly (Wells et al., 2016). Such techniques are good for studying erosion processes and obtaining more accurate measures of volumes moved but at a cost. Also, not many eroded fields can be assessed within a short period of time before the erosion features are obliterated by cultivation. Such assessments also need to include the time and travel costs to locate the visibly eroded fields within a landscape, as well as carry out the field measurements. They are not appropriate to monitor what is happening across a soil landscape as there may be too many fields to be assessed within a short period.

We argue that the extent of visibly eroding fields in a landscape is important when estimating rates of erosion across that landscape. There has to be a trade-off between accuracy (within ± c.30%) and cost of assessment. This is presumably one of the reasons why it is often decided to use modelled erosion rates rather than estimate amounts eroded in the field. For once the model is set up it can be applied, using GIS, across a landscape: but does this reflect reality? We argue it does not. However, using the techniques described here once results have been reported from many fields in many locations within a similar climate zone and the results corroborate earlier findings (e.g. Evans et al., 2016) those estimates might be used as substitute values and only the obviously visibly eroding fields need to be located. Fields that are obviously eroding are easy to locate in the field (Watson and Evans, 2007), or on aerial photographs (Boardman, 2016; Evans, 2005; Fischer et al., 2017) and transects across landscapes are easily carried out. For instance, that was how much of the ad hoc data on the distribution of eroded fields was assembled (Evans and Cook, 1986) prior to setting up the National Monitoring Scheme in England and Wales. Furthermore, what would now be called citizen science (e.g. members of Friends of the Earth (Evans and MacLaren, 1994; Evans, 1996a: 97–103)), the Soil Association (Evans, 1987) can also be very useful in confirming by their field work in different parts of the country that soil associations considered to be at risk in one location are at risk where found in other parts of England.

VI Conclusion

It is relatively easy to locate channel-eroded fields. It is more difficult and time-consuming to assess erosion rates. In England, erosion has been mostly assessed (measured, estimated and monitored) at the field scale: the assessments accord well with each other (Evans et al., 2016). We argue that assessment of erosion in the field is the best way of assessing erosion. Field assessment and extrapolation of those data seem to be a better alternative than assessment and extrapolation from unvalidated models.

To the estimate of volume/mass of soil eroded by rills and gullies an amount needs to be added to account for wash erosion. This will be a small amount. Wash takes place across the landscape and in terms of soil delivered to water courses may overall be more important than more severe erosion from fields covering only a small part of the landscape or occurring far less frequently.

Field-scale assessment of erosion is a satisfactory technique, therefore, for identifying the severity (or not) of erosion in fields and how erosion severity and incidence varies by soil type, crop type and the weather from year to year. It can also give an indication of land use changes over time as agricultural policy changes and hence how a landscape’s susceptibility to erosion will change. The impacts of the changes in land use brought about by changes in agricultural policy can only be obtained by long-term monitoring, such as in the Sompting catchment on the Sussex South Downs in southern England monitored over two decades (Evans, 2010) and more.

By mapping rill and gully eroded fields, the extent of erosion in landscapes is put into context and perspective. Thus, the question ‘Is erosion a problem of arable land in Britain?’ can be answered. It is in some soil landscapes, but is rarely extensive and severe and over the short-term is not a problem with regard to thinning of the soil and associated loss of crop productivity.

Over the short term, it likely runoff carrying soil particles and other diffuse pollutants, not soil erosion, is the problem (Boardman et al., 2019; Evans, 2009, 2017). These pollute water courses and reservoirs, which supply water to citizens. Treating the water to make it potable is costly.

Policy-makers believe what they read (Defra, 2018) and make policy accordingly, so information has to be as good as possible, not modelled. The timing of monitoring is crucial. Therefore, a field visit cannot be made at any time as erosion might be masked by vegetation growth, especially in the humid tropics where ground can be quickly covered.

Field-scale monitoring identifies rapidly which fields are visibly eroding, or not. Monitoring a landscape to assess the severity, frequency and extent of erosion is quick, easy and cheap.

Putting erosion rates into perspective suggests that in many localities rates predicted by models are too high, and hence soil thinning is much less important as a factor leading to poorer crop yields than is generally acknowledged. Such information suggests that causes other than erosion are important when considering soil degradation.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

John Boardman

References

Alstrom

Akerman

(1992) Contemporary soil erosion rates on arable land in southern Sweden. Geografiska Annaler 74A: 101–108.

Auzet

Boiffin

Papy

, et al. (1993) Rill erosion as a function of the characteristics of cultivated catchments in the north of France. Catena 20: 41–62.

Bakker

Govers

Rounsevell

MDA

(2004) The crop-productivity-erosion relationship: An analysis based on experimental work. Catena 57: 55–76.

Biddulph

Collins

Foster

IDL

, et al. (2017) The scale problem in tackling diffuse water pollution from agriculture: Insights from the Avon Demonstration Test Catchment Programme. River Research Applications 1: 1–12.

Boardman

(1983) Soil erosion at Albourne, West Sussex, England. Applied Geography 3: 317–329.

Boardman

(1988) Severe erosion on agricultural land in East Sussex, UK, October 1987. Soil Technology 1: 333–348.

Boardman

(1990) Soil erosion on the South Downs: A review. In: Boardman

Foster

IDL

Dearing

(eds) Soil Erosion on Agricultural Land. Chichester: Wiley, 87–105.

Boardman

(1991) Land use, rainfall and erosion risk on the South Downs. Soil Use and Management 7(1): 34–38.

Boardman

(1998) An average soil erosion rate for Europe: Myth or reality? Journal Soil and Water Conservation 53(1): 46–50.

10.

Boardman

(2003) Soil erosion and flooding on the South Downs, southern England 1976–2001. Transactions Institute British Geographers 28: 176–196.

11.

Boardman

(2006) Soil erosion science: Reflections on the limitations of current approaches. Catena 68(2/3): 73–86.

12.

Boardman

(2016) The value of Google Earth for erosion mapping. Catena 143: 123–127.

13.

Boardman

Favis-Mortlock

(1998) (eds) Modelling Soil Erosion by Water. NATO ASI Series. Berlin: Springer.

14.

Boardman

Favis-Mortlock

(1999) Frequency-magnitude distributions for soil erosion, runoff and rainfall – a comparative analysis. Zeitschrift fur Geomorphologie 115: 51–70.

15.

Boardman

Burt

Evans

, et al. (1996) Soil erosion and flooding as a result of a summer thunderstorm in Oxfordshire and Berkshire, May 1993. Applied Geography 16: 21–34.

16.

Boardman

Ligneau

De Roo

, et al. (1994) Flooding of property by runoff from agricultural land in northwest Europe. Geomorphology 10: 183–196.

17.

Boardman

Poesen

Evans

(2003) Socio-economic factors in soil erosion and conservation. Environmental Science and Policy 6: 1–6.

18.

Boardman

Shepheard

Walker

, et al. (2009) Soil erosion and risk assessment for on- and off-farm impacts: A test case in the Midhurst area, West Sussex, UK. Journal of Environmental Management 90: 2578–2588.

19.

Boardman

Vandaele

Evans

, et al. (2019) Off-site impacts of soil erosion and runoff: Why connectivity is more important than erosion rates. Soil Use and Management 35(2): 245–256.

20.

Borelli

Robinson

Fleischer

, et al. (2017) An assessment of the global impact of 21st century land use change on soil erosion. Nature Communications 8(8). DOI: 10.1038/s41467–017–02142–7.

21.

Castillo

Pérez

James

, et al. (2012) Comparing the accuracy of several field methods for measuring gully erosion. Soil Science Society of America Journal 76(4): 1319–1332.

22.

Cerdan

Le Bissonnais

Courturier

, et al. (2002) Rill erosion on cultivated hillslopes during two extreme rainfall events in Normandy, France. Soil & Tillage Research 67: 99–108.

23.

Cerdan

Govers

Le Bissonnais

, et al. (2010) Rates and spatial variations of soil erosion in Europe: A study based on erosion plot data. Geomorphology 122: 167–177.

24.

Colbourne

GJN

Staines

(1985) Soil erosion in south Somerset. Journal of Agricultural Science, Cambridge 104: 107–112.

25.

Collins

Zhang

Walling

, et al. (2010) Tracing sediment loss from eroding farm tracks using a geochemical fingerprinting procedure combining local and genetic algorithm optimisation. Science of the Total Environment 408: 5461–5471.

26.

Collins

Zhang

Duethmann

, et al. (2013) Using a novel tracing-tracking framework to source fine-grained sediment loss to watercourses at sub-catchment scale. Hydrological Processes 27: 959–974.

27.

Defra (2018) Health and Harmony: The Future for Food, Farming and the Environment in a Green Brexit. The Future Farming and Evidence Compendium. Cm 9577. London: HMSO.

28.

Den Biggelaar

Lal

Eswaran

, et al. (2003) Crop losses to soil erosion at regional and global scales: Evidence from plot-level and GIS data. In: Wiebe

(ed.) Land Quality, Agricultural Productivity and Food Security. Cheltenham: Edward Elgar, 262–279.

29.

Edwards

Owens

(1991) Large storm effects on total erosion. Journal of Soil and Water Conservation 46(1): 75–78.

30.

ELD Initiative (2015) The Value of Land. Available at: www.eld-initiative.org (accessed 24 October 2018).

31.

Evans

(1980a) Characteristics of water-eroded fields in lowland England. In: De Boodt

Gabriels

(eds) Assessment of Erosion. Chichester: John Wiley, 77–87.

32.

Evans

(1980b) Mechanics of water erosion and their spatial and temporal controls: An empirical viewpoint. In: Kirkby

Morgan

RPC

(eds) Soil Erosion. Chichester: John Wiley, 109–128.

33.

Evans

(1987) Membership News. Bristol: Soil Association.

34.

Evans

(1988) Water Erosion in England and Wales 1982–1984. Report for Soil Survey and Land Research Centre, Silsoe.

35.

Evans

(1990a) Soils at risk of accelerated erosion in England and Wales. Soil Use and Management 6: 125–131.

36.

Evans

(1990b) Water erosion in British farmers’ fields – some causes, impacts, predictions. Progress in Physical Geography 14: 199–219.

37.

Evans

(1992a) Erosion in England and Wales – The present the key to the past. In: Bell

Boardman

(eds.) Past and Present Soil Erosion. Oxford: Oxbow Books, 53–66.

38.

Evans

(1992b) Assessing erosion in England and Wales. In: Haskins

Murphy

(eds) People Protecting Their Land. Proceedings 7th ISCO Conference. International Soil Conservation Organisation, Sydney, 82–91.

39.

Evans

(1992c) Rill erosion in contrasting landscapes. Soil Use and Management 8: 170–175.

40.

Evans

(1993a) On assessing accelerated erosion of arable land by water. Soils and Fertilizers 56 (11): 1285–1293.

41.

Evans

(1993b) Extent, frequency and rates of rilling of arable land in localities in England and Wales. In: Wicherek

(ed.) Farm Land Erosion in Temperate Plains Environment and Hills. Amsterdam: Elsevier, 177–190.

42.

Evans

(1995) Some methods of directly assessing water erosion of cultivated land – a comparison of measurements made in plots and in fields. Progress in Physical Geography 19: 115–129.

43.

Evans

(1996a) Soil Erosion and Its Impacts in England and Wales. London: Friends of the Earth Trust, London.

44.

Evans

(1996b) Some soil factors influencing accelerated water erosion of arable land. Progress in Physical Geography 20: 205–215.

45.

Evans

(1998) Field data and erosion models. In: Boardman

Favis-Mortlock

(eds) Modelling Soil Erosion by Water. Berlin: Springer, 313–327.

46.

Evans

(2002a) Soil deterioration and loss of topsoil. In: Munn

(ed.) Encyclopedia of Global Environmental Change. Chichester: Wiley, 587–594.

47.

Evans

(2002b) An alternative way to assess water erosion of cultivated land – field-based measurements: An analysis of some results. Applied Geography 22: 187–208.

48.

Evans

(2005) Monitoring water erosion in lowland England and Wales – a personal view of its history and outcomes. Catena 64: 142–161.

49.

Evans

(2006) Land use, sediment delivery and sediment yield in England and Wales. In: Owens

Collins

(eds) Soil Erosion and Sediment Redistribution in River Catchments. Wallingford: CAB International, 70–84.

50.

Evans

(2009) Pesticide runoff into English rivers – a big problem for farmers. Pesticide News 85: 12–15.

51.

Evans

(2010) Land use and accelerated soil erosion by water in a small catchment on the South Downs, West Sussex, England – past and present. In: Cowley

Standring

Abicht

(eds) Landscapes Through the Lens. Oxford: Occasional Publication of the Aerial Archaeology Research Group No. 2, Oxbow Books, 129–142.

52.

Evans

(2012) Reconnaissance surveys to assess sources of diffuse pollution in rural catchments in East Anglia, eastern England – implications for policy. Water and Environment Journal 26 (2): 200–211.

53.

Evans

(2013) Assessment and monitoring of accelerated water erosion of cultivated land – when will reality be acknowledged? Soil Use and Management 29(1): 105–118.

54.

Evans

(2017) Factors controlling soil erosion and runoff and their impacts in the upper Wissey catchment, Norfolk, England: A ten year monitoring programme. Earth Surface Processes and Landforms 42(14): 2266–2279.

55.

Evans

Boardman

(1994) Assessment of water erosion in farmers’ fields. In: Rickson

(ed.) Conserving Soil Resources: European Perspectives. Wallingford: CAB International, 13–24.

56.

Evans

Boardman

(2003) The curtailment of muddy floods in the Sompting catchment, South Downs, West Sussex, southern England. Soil Use and Management 19: 223–231.

57.

Evans

Boardman

(2016a) The new assessment of soil loss by water erosion in Europe. Panagos P. et al. 2015 Environmental Science & Policy 54, 438–447 – a response. Environmental Science & Policy 58:11–15.

58.

Evans

Boardman

(2016b) A reply to Panagos et al., 2016 (Environmental Science & Policy 59, 53–57). Environmental Science & Policy 60: 63–68.

59.

Evans

Brazier

(2005) Evaluation of modelled spatially distributed predictions of soil erosion by water versus field-based assessments. Environmental Science & Policy 8: 493–501.

60.

Evans

Cook

(1986) Soil erosion in Britain. SEESOIL 3: 28–59.

61.

Evans

MacLaren

(1994) Monitoring Water Erosion of Arable Land. London: Friends of the Earth.

62.

Evans

Morgan

RPC

(1974) Water erosion of arable land. Area 6: 221–225.

63.

Evans

Nortcliff

(1978) Soil erosion in north Norfolk. Journal of Agricultural Science, Cambridge 90: 185–192.

64.

Evans

Collins

Foster

IDL

, et al. (2016) Extent, frequency and rate of water erosion of arable land in Britain – benefits and challenges for modelling. Soil Use and Management 32(Supplement 1): 149–161.

65.

Evans

Collins

Zhang

, et al. (2017) A comparison of conventional and ¹³⁷Cs–based estimates of soil erosion rates on arable and grassland across lowland England and Wales. Earth-Science Reviews 173: 49–64.

66.

FAO (2015) Status of the World’s Soil Resources: Main Report. Rome.

67.

Fischer

Kistler

Brandhuber

, et al. (2017) Validation of official erosion modelling based on high resolution rain data by aerial photo erosion classification. Earth Surface Processes and Landforms. DOI: 10.1002/esp.4216.

68.

Garcia-Ruiz

Begueria

Lana-Renault

, et al. (2015) A meta-analysis of soil erosion rates across the world. Geomorphology 239: 160–173.

69.

Govers

(1991) Rill erosion on arable land in central Belgium: Rates, controls and predictability. Catena 18: 133–155.

70.

Govers

Merckx

van Wesemael

, et al. (2017) Soil conservation in the 21st century: Why we need smart agricultural intensification. SOIL 3: 45–59.

71.

Govers

Poesen

(1988) Assessment of the interrill and rill contributions to a total soil loss from and upland field plot. Geomorphology 1: 343–354.

72.

Harrold

Edwards

(1972) A severe rainstorm test on no-till corn. Journal of Soil and Water Conservation 27: 30.

73.

Hjelmfelt

Kramer

Spomer

(1986) Role of large events in average soil loss. In: Proceedings Fourth Federal Interagency Sedimentation Conference, Vol. 1. Denver: US Geological Survey, 3.1–3.9.

74.

ISSS (1996) Terminology for Soil Erosion and Conservation. Wageningen, the Netherlands: International Society of Soil Science/International Soil Centre and Information Centre.

75.

Jarvis

Allen

Fordham

, et al. (1984) Soils and Their Use in South East England. Harpenden: Soil Survey of England and Wales.

76.

Jetten

de Roo

Favis-Mortlock

(1999) Evaluation of field-scale and catchment-scale soil erosion models. Catena 37: 521–541.

77.

Laflen

Moldenhauer

(2003) Pioneering Soil Erosion Prediction. The USLE Story. World Association of Soil and Water Conservation, Special Publication No. 1.

78.

Larson

Lindstrom

Schumacher

(1997) The role of severe storms in soil erosion: A problem needing consideration. Journal of Soil and Water Conservation 52(2): 90–95.

79.

Ledermann

Herweg

Liniger

, et al. (2010) Applying erosion damage mapping to assess and quantify off-site effects of soil erosion in Switzerland. Land Degradation & Development 21: 353–366.

80.

Ludwig

Boiffin

Chadoeuf

, et al. (1995) Hydrological structure and erosion damage caused by concentrated flow in cultivated catchments. Catena 2: 227–252.

81.

Montanerella

(2015) Govern our soils. Nature 528: 32–33.

82.

Montgomery

(2007) Soil erosion and agricultural sustainability. PNAS 104(33): 13268–13272.

83.

Murphy

Frewin

(1993) Rill erosion on a structurally degraded sandy loam surface soil. Australian Journal of Soil Research 31: 419–436.

84.

NRC (1986) Soil Conservation. Assessing the National Resources Inventory. Volume 1. Washington, DC: National Academy Press.

85.

Oldemann

Hakkeling

RTA

Sombroek

(1991) World Map of the Status of Human-induced Soil Degradation: An Explanatory Note. 2nd edn. Nairobi: GLASOD and International Soil Reference and Information Centre, United Nations Environment Programme.

86.

Panagos

Borelli

Poesen

, et al. (2015) The new assessment of soil loss by water erosion in Europe. Environmental Science & Policy 54: 438–447.

87.

Parsons

Foster

IDL

(2011) What can we learn about soil erosion from the use of ¹³⁷Cs? Earth-Science Reviews 108: 101–113.

88.

Prasuhn

(2011) Soil erosion in the Swiss midlands: Results of a 10-year field survey. Geomorphology 126: 32–41.

89.

Reed

(1979) Accelerated erosion of arable soils in the United Kingdom by rainfall and runoff. Outlook on Agriculture 10: 41–48.

90.

Robinson

Boardman

(1988) Cultivation practice, sowing season and soil erosion on the South Downs, England: A preliminary study. Journal Agricultural Science, Cambridge 110: 169–177.

91.

Sanchez

Recatala

Colomer

, et al. (2001) Assessment of soil erosion at national level: A comparative analysis for Spain using several existing maps. In: Villacampa

Brevia

Uso

(eds) Ecosystems and Sustainable Development III. Southampton: WIT Press, 249–258.

92.

Sonneveld

BGJS

Dent

(2009) How good is GLASOD? Journal of Environmental Management 90: 274–283.

93.

SSEW (1983) Soil Map of England and Wales – 6 Sheets. Legend for the 1:250,000 Soil Map of England Wales. Harpenden: Soil Survey of England and Wales.

94.

Steinhoff-Knopp

Burkhard

(2018) Soil erosion by water in northern Germany: Long-term monitoring results from Lower Saxony. Catena 165: 299–309.

95.

Syvitski

JPM

Kettner

(2011) Sediment flux and the Anthropocene. Philosophical Transactions of the Royal Society A 369: 957–975.

96.

Trimble

(1983) A sediment budget for Coon Creek basin in the Driftless Area, Wisconsin, 1853–1977. American Journal of Science 283: 454–474.

97.

Vandaele

Poesen

(1995) Spatial and temporal patterns of soil erosion rates in an agricultural catchment, central Belgium. Catena 25: 213–226.

98.

Van-Camp

Bujarrabal

Gentile

, et al. (2004) Reports of the Technical Working Groups Established under the Thematic Strategy for Soil Protection EUR 221319 EN/2, 2004, Office for Official Publications of the European Communities, Luxembourg.

99.

Vanmaercke

Maetens

Poesen

, et al. (2012) A comparison of measured catchment sediment yields with measured and predicted hillslope erosion rates in Europe. Journal of Soils and Sediments 12: 586–602.

100.

Van Oost

Quine

Govers

, et al. (2007). The impact of agricultural soil erosion on the global carbon cycle. Science 318 (5850): 626–629.

101.

Verstraeten

Poesen

(2002) Using sediment deposits in small ponds to quantify sediment yields from small catchments: Possibilities and limitations. Earth Surface Processes and Landforms 27: 1425–1439.

102.

Walling

Quine

(1990) Calibration of caesium-137 measurements to provide quantitative erosion rate data. Land Degradation and Rehabilitation 2: 161–175.

103.

Walling

Quine

(1991) Use of ¹³⁷Cs measurements to investigate soil erosion on arable fields in the UK: Potential applications and limitations. Journal of Soil Science 42: 147–165.

104.

Watson

Evans

(1991) A comparison of estimates of soil erosion made in the field and from photographs. Soil & Tillage Research 19: 17–27.

105.

Watson

Evans

(2007) Water erosion of arable fields in north-east Scotland, 1985–2007. Scottish Geographical Journal 123: 107–121.

106.

Wells

Momm

Bennett

, et al. (2016) A measurement method for rill and ephemeral gully erosion assessments. Soil Science Society of America Journal 80(1): 203–214.

107.

Zachar

(1982) Soil Erosion. Elsevier: Amsterdam.