A GIS‐based approach to site vegetated buffer strips for erosion control within an agricultural catchment in southern England

Agricultural soil erosion is largely attributed to arable intensification and increased mechanization. Runoff from arable land and intensively managed grassland transports sediment and contaminants across the landscape and into watercourses, causing crop loss, land degradation, and water quality issues. One low‐cost and low‐maintenance nature‐based mitigation approach is the implementation of vegetated buffer strips (VBS): grassland sited along field margins to trap sediment and contaminants, reducing transportation and diffuse pollution rates. GIS modelling using remotely sensed landscape indices and land parcel data can provide an efficient means of identifying priority areas for intervention at sub‐catchment or farm system scales. We develop and test a scalable runoff risk model in the lower Rother catchment, West Sussex. The model uses the Normalized Difference Vegetation Index (NDVI) applied to satellite images as an erodibility proxy and identifies locations along pathways that are conceivably at greatest risk of sediment accumulation and transfer, guided by field observations. We assess current and historical field boundaries near high‐risk locations, evaluating the potential capacity of their margins to contribute to runoff risk reduction using an innovative ranking system. Recommendations are made for VBS implementation and the value of historical field boundary and margin restoration is discussed. Our method offers a rapid approach with minimal data requirements to identify high‐risk sediment runoff locations and priority sites for intervention. The tool has the potential to guide decision‐makers responsible for targeting and implementing soil erosion and runoff control measures such as VBS, while also maximizing agri‐environmental and cultural benefits.


| INTRODUCTION
Agricultural diffuse pollution resulting from fine sediment and agrochemical runoff from farms presents significant challenges to land managers and incurs financial costs to farm businesses globally (Collins & McGonigle, 2008;Harrison et al., 2019;Vennix & Northcott, 2004).Soil erosion, land degradation, water quality, and alterations to biodiversity are present throughout many river catchments (Evans et al., 2017a), threatening the provision of natural capital goods and services (Papacharalampou et al., 2017).Farming intensification, arable expansion, and field boundary removal have increased soil loss, crop failure, muddy flooding, and in-stream sedimentation risk (Hess et al., 2010;Mullan et al., 2018;O'Connell et al., 2007).Considering future market and climate uncertainty, these issues must be addressed to support productivity whilst meeting objectives for achieving good ecological status under the Water Framework Directive (WFD) (Evans et al., 2017b;HM Government, 2003).
Land managers are increasingly adopting nature-based solutions for mitigating agricultural soil erosion and runoff (Miralles-Wilhelm, 2021).Vegetated buffer strips (VBS) are commonly implemented in agricultural catchments to slow flow and intercept sediment and contaminants (Lerch et al., 2017;Pistocchi, 2022).Reducing surface flow effectively filters runoff and increases infiltration and thus groundwater recharge, improving water quality (Lerch et al., 2017).These linear grassland features are positioned along arable field margins and offer additional benefits through increased habitat connectivity for hedgerow species (Broughton et al., 2021;Cole et al., 2020;Haddaway et al., 2016).VBS are generally more effective at trapping sediment when flow is shallow and slow (Barling & Moore, 1994).Their efficiency reduces with decreasing sediment size, with greater deposition of sandy loams than clayey and silty loams observed on gentler slopes (Cole et al., 2020;Marshall & Moonen, 2002).Flume experiments revealed a greater proportion of nutrient-rich finer silts and clays within emerging runoff that is either deposited immediately downslope of VBS or transported downstream to watercourses (Ghadiri et al., 2000(Ghadiri et al., , 2006)).It is understood that VBS can increase net sediment deposition and deliver benefits when implemented alongside on-site measures to reduce mobilization at source (Barling & Moore, 1994;Mekonnen et al., 2014).
Thus, the location of VBS on farmland and their proximity to high-risk erosion sites are key factors in their effectiveness in limiting runoff.At locations with known soil erosion problems, prioritization should be given to implementing VBS along arable field margins with limited or no vegetation.A catchment-based approach is ideal for evaluating options in the landscape, cognisant of local environmental, social, and economic conditions and broader projected climate change (Department for Environment, Food and Rural Affairs/ Defra, 2013).
Under the recent Sustainable Farming Incentive (SFI) scheme in England (Defra, 2020), payment options exist for the establishment of grassland buffers of at least 4 m width on arable and horticultural land and improved grassland to reduce runoff and trap pollutants (Defra and RPA, 2023).These simple intervention methods require limited investment and maintenance and do not require significant arable reductions, any losses of which can be offset by farming subsidies and improved in-field soil retention.
Evaluating soil erosion and runoff risk at landscape scale using GIS can help to identify locations that may contribute more material in a catchment (Hahn et al., 2014;Zhang et al., 2009).Modelling risk across agricultural landscapes has significant potential to identify priority areas where resources may be best allocated for reducing runoff and diffuse pollution (Heathwaite et al., 2005;Mullan et al., 2016;Reaney et al., 2019).Recent focus on sediment connectivity has aimed to improve potential pathway mapping from source to sink using remote sensing data and field observations of high-risk locations to aid decision-making (Boardman et al., 2019;Hudson & Soar, 2023;Najafi et al., 2021).In this way, GIS-based approaches can effectively identify risk at locations where ephemeral features may be difficult to observe (Liu et al., 2021).Topographic connectivity indices can be generated from flow accumulation modelling (Wilson et al., 2008) and derived from open-source high-resolution digital elevation models (DEMs) at 1 m resolution (Environment Agency, 2022).The Normalized Difference Vegetation Index (NDVI) is derived from high resolution satellite imagery and indicates vegetation density from which soil erodibility can be inferred.At suitable granularity, land cover data can aid delineation of field boundaries, enabling the assessment of their margins as potential options for implementing VBS.Furthermore, historical maps depicting former field boundaries can guide potential heritage margin restoration, offering additional options for siting VBS and providing a novel research component to our study.However, these data, if indeed available, should be interpreted with caution, cognisant of their original scope, purpose and method of collection.(Pearson et al., 2019b).
Several studies have employed GIS modelling to identify optimal sites for VBS implementation for sediment runoff mitigation.Bruin et al. (2009) determined field margin placement by optimizing cropped swath orientation based on field geometries, but offer no contextualization with high-risk erosion locations.Vennix and Northcott (2004) simulated sediment transport rates to evaluate conceptual riparian buffers using the Agricultural Non-Point Source (AGNPS) model (Young et al., 1987), generating broad scale outputs using reasonably coarse input data.Tomer et al. (2003Tomer et al. ( , 2009) ) employ terrain analyses to map wetness and erosion index values along riparian buffer zones.
However, these examples do not readily facilitate identification of the highest risk locations along pathways selected for their proximity to known sites of erosion, or utilize additional spatial data to inspect and evaluate plausible field margins for intervention at the farm scale.
Numerous alternative terrain models incorporate erodibility in modelling erosion and runoff, for example, the Revised Universal Soil Loss Equation (RUSLE) (Terranova et al., 2009), the Unit Stream Power based Erosion/Deposition model (Mitasova et al., 2001) and the Water Erosion Prediction Project (WEPP) model (Flanagan & Nearing, 1995).However, these models commonly require complex parameterisation of soil characteristics informed by empirical observations, have high data requirements and infrequently account for spatial variability in runoff (Leh et al., 2013).Several studies have applied the NDVI as a measure of soil erodibility, notably in the estimation of the cover management factor (C-factor) in GIS-RUSLE modelling (Durigon et al., 2014;Karaburun, 2010), and exploring the relationship between NDVI and soil erosion (Ayalew et al., 2020;Ouyang et al., 2010).These examples reveal mixed conclusions, with Ouyang et al. (2010) suggesting that the NDVI can effectively represent land cover and climatic scenarios, but Ayalew et al. (2020) highlighting its sensitivity to other soil, topographic and vegetation factors.On balance, the NDVI may offer a more useful indicator of relatively high-risk sediment accumulation sites, but these should be validated with more detailed desktop and field-based inspection where possible.
Other recent work has focused on the integration of heritage landscape features to determine the P-factor for GIS-RUSLE modelling.Panagos et al. (2015) utilized grass margins data to generate P-factor estimates across Europe, and Brandolini et al. (2023) calculated the P-factor by combining field boundaries and historical boundary loss information.This research contributes to an important topical development, promoting representation of cultural landscape features in the estimation of contemporary soil erosion rates using wellestablished numerical modelling.Collins and McGonigle (2008) and Deeks et al. (2012) highlighted significant scope for further GIS tools to guide optimal placement of erosion control features such as VBS.
Thus, there is an opportunity to integrate these concepts into an alternative, spatially explicit relative risk-based framework that addresses multidisciplinary management goals.
Our study aims to identify priority field boundaries for implementing VBS on margins at a small testbed site in the lower Rother catchment, southern England; a rural location well-known for its issues with agricultural soil erosion and runoff.We achieve this through (1) developing a model to explore the spatial pattern of relative sediment accumulation and runoff risk, (2) evaluating field boundaries based on their proximity to the highest risk locations along sediment pathways, and (3) recommending margins for implementing VBS for erosion control, with consideration to additional cultural heritage and environmental services.

| Study area
The lower catchment of the River Rother covers approximately 274 km 2 along a length of 51.5 km and is located within the South Downs National Park.The Rother originates north of Petersfield, Hampshire, and flows eastwards into West Sussex.It has two main tributaries, the Hammer stream and the Lod, and flows into the River Arun near Pulborough (Figure 1).Our study focuses on a small site of 149 hectares between Easebourne and Lodsworth, located approximately 800 m north of the main Rother channel (Figure 1a,b).The site sits within the Lower Greensand Group, broadly consisting of Hythe formation sandstones and mudstones in the upper sub-catchment reaches to the north, with Easebourne member sandstones in the mid sub-catchment zone (British Geological Survey, 2013).Lower Fittleworth member sandstone and mudstone dominate the lower region, bounded by Selham Ironshot Sands member sandstones to the east and west and at the sub-catchment outlet (Figure 1a).The catchment elevation ranges from c. 181 m asl to c. 41 m asl (Environment Agency, 2022).
The lower Rother catchment is predominantly rural, with 96.7% being arable fields, grassland or woodland.The Lower Greensand geology of sandstones and mudstones across the wider Rother catchment is highly conducive for arable cultivation (Boardman et al., 2009), with winter cereals, salad vegetables, oil seed rape, and potatoes being notable crops (Evans et al., 2017b).Steeper slopes to the southeast and southwest are dominated by improved grassland for livestock grazing and woodland.Within the entire Rother catchment and across the South Downs region, agricultural soil erosion is well-documented (Boardman, 2001;Boardman et al., 2020;Evans et al., 2017b;Guerra, 1994).Susceptibility of winter crop fields to erosion is enhanced by greater precipitation over autumn and winter (Boardman et al., 2009;Boardman et al., 2020).Field boundaries are commonly marked by hedgerows, a characteristic heritage landscape feature of the area (South Downs National Park Authority, 2019).
Regional losses of field boundaries and margins are associated with agricultural intensification and land consolidation (Boardman & Vandaele, 2016;Pearson & Soar, 2018).With fewer margins serving to intercept overland flow, fine sediment runoff from erodible fields promotes the formation of rills and gullies, presenting significant management issues (Boardman, 2016;Boardman et al., 2009).Muddy flooding risk across farm tracks and roads is enhanced (Boardman et al., 2019), contributing to fine sedimentation problems within main channels (Evans et al., 2017b;Foster et al., 2019) and disconnecting the floodplain (Environment Agency, 2021).Furthermore, high diffuse sediment loads and transportation of agricultural contaminants from fields to waterbodies contribute significantly to ecological problems, notably fish spawning habitat depletion (Boardman et al., 2009;Evans et al., 2017b).
These erosion issues are thought to be long-standing, with historical records indicating an extensive arable heritage landscape in the mid-19th century (Pearson et al., 2019a(Pearson et al., , 2019b)).Widespread conversion of grassland and meadows into arable land to improve post-war food security (Pearson & Soar, 2018) has been noted as a catalyst for contemporary soil erosion in the Rother catchment (Evans, 2010;Vanwalleghem et al., 2017).Attainment of 'good' ecological and chemical WFD status remains a significant challenge (Evans et al., 2017b).Indeed, based on monitoring between 2019 and 2022, the Rother's ecological status decreased from 'moderate' to 'poor', and its chemical status scored 'fail' in 2019 (not requiring assessment in 2022) (Environment Agency, 2023).VBS initiatives across the catchment are supported by local guidance for farmers and landowners (South Downs National Park Authority, 2017) in addition to broader environmental stewardship incentives (Defra and RPA, 2023).We suggest that the NDVI can effectively capture sub-field scale variation in vegetation cover, and utilizing the index values within a The study site and its geology.The locations of previous erosion observations from Boardman (2016)  LCM2021 is a national scale dataset derived from 10 m resolution satellite imagery and represents the configuration of land parcels that are greater than 0.5 hectares (Marston et al., 2022).We chose to manually align narrow parcels smaller than 0.5 hectares to available Ordnance Survey topographic reference data (Ordnance Survey, 2021) to ensure farm track and road boundaries intersecting derived flow pathways were represented.We also optionally considered historical field boundaries inferred from digitized agricultural tithe survey maps dated to c. 1840 (Pearson et al., 2019a).

| Stage one: Generating erodibility-weighted accumulation
The study site (Figure 1a) was selected based on the greatest density of observed erosion points (observation points of soil erosion, rills, gullies and fans on arable fields identified through field study and Google Earth imagery; Boardman, 2016), and the greatest arable cover as identified from the LCM2021 dataset (Marston et al., 2022).
The inverse range of positive NDVI values was used to weight soil erodibility to ensure bare soil (lower NDVI) and denser vegetation (higher NDVI) contributed more and less to sediment accumulation, respectively.Flow accumulation algorithms such as D8 are incapable of handling negative weights (i.e.'de-accumulation'), so negative values representing water were set to zero.Unlike erodibility risk classification by land cover type, which commonly relies on complex parameterisation informed by empirical knowledge of specific site conditions, NDVI provides a more objective weight index.While unweighted accumulation cumulatively sums cell values of 1 along pathways, weighting reassigns cell values to the value of the weight.This aimed to identify a risk distribution that better incorporates vegetation density and, thus, represents a more realistic depiction of soil detachment and transportation hotspots.

| Stage two: Determining the accumulation threshold
We developed a method for testing different flow accumulation thresholds in order to identify appropriate pathways for further analysis.The flow accumulation threshold is directly proportionate to the complexity of the derived channel network, whereby decreasing the threshold value generates a greater number of pathways and increasing the threshold value generates fewer pathways (Zhang et al., 2021).We focus on first order streams in order to promote early intervention in sub-catchment headwaters where erosion control benefits are thought to be greatest (Correll, 2005).Particular attention was paid to first order pathways coincident with prior erosion observation points (as mapped by Boardman, 2016), where VBS could have most impact.Intervention along pathways coincident with observed erosion features aims to limit the likelihood of eroded material reaching the main watercourse and contributing to sedimentation and water quality problems.Our method was designed to identify sites close to as many field observation points as possible, while generating a practical number of pathways for further analysis and potential VBS intervention.

| Stage three: Calculating accumulation risk along pathways
Derived pathways were assigned unique IDs and converted to points at 1 m intervals.For each pathway, ascending point IDs were assigned in order of ascending accumulation.Successive accumulation percentage increase was calculated at each pathway point, and accumulation increase values in the 99th percentile were determined as high-risk points (HRPs) of accumulation along each pathway.Elevation, NDVI and slope values were also extracted for each pathway point to provide additional context.Stages one to three were scripted as an ESRI ArcGIS Pro v.3 Python v.3 toolbox ('Sediment Accumulation Risk Model (SARM) Toolbox v1.0') using the Spatial Analyst extension (Hudson & Schaefer, 2023).We then selected a subset of pathways that were hydrologically connected (at our chosen accumulation threshold) to the majority of observation points within our study site to evaluate their suitability for VBS implementation.

| Stage four: Identifying suitable sites for VBS
The approach to identifying realistic options for the implementation of VBS was informed by government guidance (Defra and RPA, 2023;RPA, 2023), based on limiting surface water runoff and, thus, sediment transfer through the introduction of VBS along arable or improved grassland field margins.The initial accumulation output was inspected to identify current and historical boundaries located in-situ, upstream or downstream from each chosen pathway.For each boundary identified, we recorded the distance from that boundary to the nearest HRP along the pathway, and the accumulation increase value of the nearest HRP to that boundary.For each set of pathway boundaries, we applied min-max scaling to normalize (from 0 to 1): (1) the distance to HRP values, and (2) the accumulation increase values of the nearest HRPs.Combined scores were established for each boundary by calculating the average of ( 1) and (2) to form an overall ranking system, giving equal consideration to both the proximity to an HRP and the risk level of that HRP.Inverse normalized distance to HRP scores were calculated so that values closer to 1 scored higher in terms of sediment trap potential, and values closer to 0 scored lower.
We assigned a rank of 1 to the preferred boundary option, with ascending scores consecutively assigned to each boundary in order of descending preference.We then evaluated the potential utility of margins at ranked field boundary locations for the implementation or extension of existing VBS, considering a range of conceivable erosion control, cultural heritage and environmental drivers and benefits.
Aerial imagery (Getmapping, 2022) and DEM-derived slope data were used to identify existing field margins and guide our qualitative desktop evaluation of ranked boundaries.Figure S1 in the supporting information provides a broad summary of the method we have developed.

| Identifying a suitable threshold value for pathway mapping
An accumulation threshold of 3800 m 2 was found to be optimal for identifying first order pathways likely to influence most of the observation points in our study site (Figures 1 and 3).This threshold yielded 35 first order pathways for further analysis.Various thresholds were tested but these either omitted too many observation points, or they produced an impractical number of pathways for further analysis.Our chosen threshold produced flow pathways over a total length of 10.3 km.Visually, the accumulation points generally correlate with the inverse NDVI layer, with greater increases in accumulation common in locations with higher inverse NDVI scores (bare soil) and vice versa (Figure 2).

| Identifying specific pathways and risk hotspots
Visual inspection of the accumulation outputs revealed four pathways likely to influence most of the erosion observation points in the study site (Figures 2, 4, and 5).These four pathways are referred to hereafter as pathways 1-4 (P1-4).Accumulation values were exported for each pathway, successive percentage increases were then calculated between each pathway point, and points in the 99th percentile of percentage increase were identified to indicate HRP locations where flow accumulation increase rate was greatest.Using this approach, 11 HRPs were identified along our pathways.Figure 3

| Ranking field boundaries
We ranked boundaries close to our chosen pathways based on suitability for VBS intervention.Twenty-five boundaries were selected across P1-4 for evaluation as potential options.All pathways have boundaries intersecting the pathways that could serve to trap sediment directly within the pathway.Inspection of the initial accumulation output also revealed that additional boundaries that do not intersect the pathways, but are located upstream and downstream of them, should also be considered as potential options for trapping sediment upstream or downstream of the pathways.Seven boundaries were analysed for both P1 and P2, five for P3 and six for P4.Table 2 summarizes all boundaries, including whether the boundary is historical or current, its location relative to the pathway (i.e.upstream or downstream of, or intersecting), distance to the nearest HRP along each pathway (and inverse normalized distance values), the successive accumulation increase percentage value of that nearest HRP (and normalized values), combined rank scores and overall assigned rank.

| Recommendations for siting VBS
Our ranking analysis included all boundaries adjacent to pathways (in-situ, upstream and downstream), and considered both current and historical boundaries to compare locations and explore feasibility of historical margin restoration.However, we emphasize that reversion to former configurations would likely be more obtrusive on farming practice than implementing or widening existing VBS at existing field margins, and may be more appropriate if it is understood that doing so may significantly enhance sediment deposition, or that it carries substantial additional cultural heritage or ecological benefits.
The presence (or absence) of existing vegetation at existing margins was another qualitative factor in our recommendations as it is conceivable that widening existing VBS is preferential over implementation of new VBS.Margin connectivity was also assessed, whereby more connected vegetated corridors may be capable of delivering greater ecosystem benefits for farmland species than fragmented margins (Marshall, 2005).Figures 4 and 5 highlight the arrangement and ranking of boundaries along each pathway with reference to land cover and aerial imagery, and slope data, respectively.

| Pathway 1
Boundary 1 (the highest scoring) is a current boundary dividing two improved grassland fields and is situated 19 m downstream from the highest risk point along the pathway (Figure 4a).Aerial imagery reveals no adjacent margin vegetation, and slope data indicate a slight descent at the downstream margin (Figure 5a), which may accelerate Based on our ranking criteria, reinstatement of the historical boundary 7 (the lowest scoring) may deliver the least erosion control benefits.However, an erosion observation point is located north of P1 (inset on Figures 4a and 5a) and inspection of upslope pathways suggests this site of observed erosion is hydrologically connected to P1. Reinstatement of this historical margin with VBS could be plausible, particularly if ecological or cultural heritage benefits were key management priorities.

| Pathway 2
Boundary 1 is historical, and boundary 2 appears to represent its current configuration (Figure 4b); thus, rank score difference is negligible.
Boundaries 1 and 2 are located downstream of the pathway's four HRPs and divide two arable fields.Aerial imagery highlights an upslope track not depicted in current or historical land cover data, with a narrow hedgerow visible along the northern margin, contrasting with a more established VBS visible at the downslope margin.P2 follows the track westward for a short distance before crossing the boundary via a slight topographic depression (Figure 5b).Widening of vegetation along the upslope margin could thus be a suitable option for limiting muddy flooding across the track and limiting transfer into the downstream arable field.
The distribution of ranking scores for the remaining boundaries

| Pathway 3
The rank scores for boundaries 1-3, which divide arable fields, are similar, suggesting they could offer similar erosion control benefits.
Boundary 1 is historical and located upstream of all HRPs along the pathway, and 9 m upstream of the highest risk point (Figure 4c).Boundary 2 is current, likely indicating minor realignment of the former boundary.Aerial imagery highlights the presence of a wooded hedgerow and 5 m of VBS spanning between the two boundaries and slope data reveal a gentle downslope decline where P3 crosses (Figure 5c).Widening of the VBS to the north and south may dilute upslope sediment accumulation risk and limit runoff downslope.
Extending intervention eastwards along these margins could also limit F I G U R E 4 HRPs and suggestions for vegetated buffer strip implementation at P1 (a), P2 (b), P3 (c) and P4 (d) overlaid onto aerial imagery and contemporary land cover (Source: Getmapping, 2022;Marston et al., 2022).Boundaries are labelled according to their ranking.Magnitude of sediment accumulation percent increase along each pathway is shown as blue graduated circles, with P1-4 highlighted in darker blue and the remaining pathways in light blue.Historical field boundaries (from Pearson et al., 2019a) are overlaid as white dashed lines and contemporary field boundaries as black lines.Erosion observation points from Boardman (2016) 5d).Land managers could conceivably widen the existing VBS here and extend intervention eastwards to promote T A B L E 1 Elevation, slope, inverse NDVI and sediment accumulation increase percentages for all HRPs along P1-4, identified using the 99th percentile of sediment accumulation increase percentages.sediment deposition upstream of P4, reduce risk of sediment transfer into the downstream arable field and promote habitat connectivity.
Boundary 2 (current) is located 55 m from the nearest HRP and downstream of both HRPs along P4, and topography reveals a reasonable descent downslope of the boundary.Erosion has been observed in the arable field upslope of boundary 2, but aerial imagery and land cover data reveals its adjacency to well-established downslope woodland and VBS, which should already serve to limit sediment transfer downstream and promote habitat connectivity, suggesting intervention here is of lower priority.Boundary 4 appears to reflect the former configuration of boundary 2 and similarly, boundary 5 is presumed to represent the historical position of the current boundary 6.These boundaries largely overlap, producing negligible score differences and are located within woodland, 95 m downstream from the nearest HRP.This location is proposed to be least suitable for intervention, being a significant distance from an HRP and because aerial imagery indicates the presence of well-established woodland here (although underlying grassland cover is not clear).

| Evaluating the value of historical margin restoration
We view field margins as important locations for possible VBS intervention for erosion control.However, they have different uses and implications to different groups; for livestock farmers they facilitate barriers to contain livestock, to arable farmers they may be a source of pests requiring pesticide application, to ecologists they provide invaluable habitat for farmland species, and to the public they are a symbol of traditional ruralism.Their perceived value has fluctuated over time.Land consolidation, partly resulting from the enclosure movement, saw small parcels of land in England subsumed into large estates, which led to hedgerow removal during the latter half of the 19th century (Chapman & Sheail, 1994).This period saw increased mechanization with field margins and hedgerows increasingly removed to promote productivity, a trend that has continued during the post-war period (Robinson & Sutherland, 2002).However, during the late 20th century, a rising awareness of the ecological value of field margins and hedgerows shifted subsidies from productivityfocused to environmental stewardship (Cobb et al., 1999).The perceived value of field margins is impacted by the dynamic interplay between political, market, environmental and social drivers which must therefore be balanced in sustainable land management initiatives (Brandolini et al., 2023).
Our evaluation of field boundaries using our ranking criteria suggests that some historical boundary margins may have offered different erosion control benefits to current configurations.However, these findings must be considered in the context of wider natural and cultural values to deliver multidisciplinary management outcomes.In places our analysis highlighted trivial realignment of boundaries (e.g.boundaries 3-6 in Figures 4a and 5a); where reconfiguration has been inconsequential, it is logical to implement VBS along current field margins rather than restore historical margins for this purpose.
Boundary 7 at P1 and Boundary 5 at P3 are the only boundaries, which appear to have been removed entirely since the tithe survey period.Evaluation of both boundaries suggests that despite not scoring most favourably in our ranking, their restoration and placement of VBS along their margins could nevertheless contribute to sediment accumulation reductions whilst also delivering cultural heritage benefits and reconnecting former hedgerow habitats.
The development of 'shaws' along boundaries were woodland remnants deliberately left following the creation of agricultural fields, but were later deemed unnecessary by agricultural reformers.Shaw bases were kept clear for neat hedgerow shape (Chapman & Sheail, 1994), or often removed altogether to reduce tree shadow over crops, which may have indeed promoted sediment transfer between arable fields.By the time land enclosure restrictions were introduced in the mid-19th century, the landscape mosaic of shaws was so mature it would later "be mistaken, by later generations, for an integral part of 'our natural heritage'" (Chapman & Sheail, 1994, p. 10).Thus, it is conceivable that our observations of existing wooded margins from aerial imagery pertain to the predominance of traditional shaws in this locality, carrying credible cultural heritage value.These include boundaries 2 and 4 at P1, boundaries 2 and 4 at P2, boundaries 2 and 5 at P3 and boundaries 1, 2 and 6 at P4.To deliver multiple agri-environmental and cultural benefits at these sites, land managers could therefore consider pairing traditional shaws with VBS, commensurate with contemporary agri-environmental subsidy schemes.

| The value of an accessible tool for exploring sediment accumulation risk
Our method is designed for use by researchers and practitioners inter- supporting our emphasis on upper catchment intervention and selection of first order pathways for further analysis.Our relative modelling approach provides a spatial indication of risk along pathways rather than estimating sediment yields, so our findings are not directly com-  (Ayalew et al., 2020).However, we suggest that our use of the NDVI as a proxy for soil erodibility presents a new option for spatially distributed modelling, which can indicate relatively high-risk locations along pathways with minimal parameterisation and data requirements.Sentinel-2 satellite imagery offers a good spatial resolution for capturing sub-field scale variability of vegetation density and associated runoff attenuation capacity (Abdelsamie et al., 2023).Its resolution permits valid interpolation to align with higher resolution DTMs, assuming a suitable transformation method is used (i.e.bilinear).Its near universal coverage also facilitates NDVI calculation and similar assessment in other agricultural catchments.Frequent capture of Sentinel-2 data also permits shorter-term analysis of seasonality on sediment accumulation risk, which can help land managers identify locations that may be most impacted by more intense precipitation events (Vrieling et al., 2008).Future research could consider seasonal weighting of NDVI imagery to explore patterns of risk within sub-annual crop and climatic cycles.The risk outputs and ranked boundaries can be used to identify critical locations of sediment accumulation and feasible locations for intervention within our study site, respectively.
In this study area, material is commonly derived from arable fields north of the River Rother and transported downstream into the main channel where it can lead to smothering of the gravel bed and floodplain disconnection (Boardman et al., 2020;Evans et al., 2017b).VBS can mitigate these issues by trapping soil and reducing sediment load into watercourses (Cole et al., 2020).Ghadiri et al. (2000Ghadiri et al. ( , 2006) ) suggested that VBS effectively trap larger sandy material but output a greater proportion of finer nutrient-rich silts and clays downstream.In the Rother catchment, Evans, Foster, et al. (2017) discussed extensive smothering of gravel beds with eroded sands, and silts and clays transporting diffuse agricultural contaminants in the Rother catchment.
However, it was also noted that local erosion of sandy soils was more of a prevalent problem than clayey and silty soil erosion due to the former lacking the intrinsic cohesiveness of the latter.Thus, we suggest that VBS implementation in higher order streams in this locality can be employed as an effective measure for mitigating runoff of eroded material.However, our choice of pathways is not strictly pre-  et al., 2013;Mekonnen et al., 2014;Zak et al., 2018), or undertake broader arable reversion to reduce erosion from fields (Boardman et al., 2017;Hudson & Soar, 2023).It will also be important to consider how future climate change projections might impact the effectiveness of existing and planned VBS implementation (e.g.Mullan et al., 2016).Our approach provides a tool for agricultural land management policy development that is transferable to other catchments and considers both environmental and cultural service provision.

| CONCLUSIONS
We demonstrate the application of a novel, scalable model for indicating soil erosion and runoff risk along accumulation pathways, guided by empirical observations.The tool is applied to a 149 hectare study site within the agricultural lower Rother catchment in West Sussex, southern England.We applied the NDVI as a proxy for erodibility, assigning normalized inverse NDVI weights to the landscape to understand where vegetation density may limit runoff most.High-risk points (HRPs) were identified along pathways as locations with greatest runoff potential, determined by sediment accumulation surpassing the 99th percentile.Following the identification of HRPs, we explored field boundary options for siting VBS along margins for reducing soil erosion and accumulation of sediment along pathways which are hydrologically connected to observed erosion sites.The inclusion of historical field boundaries in our ranking analysis enabled us to evaluate the suitability of former configurations as potential blueprints for restoration, considering various management priorities.
We suggest a range of priority site options for VBS implementation along four first order pathways in the study site, classified by distance to, and magnitude of, the nearest HRP.Implementation of (or widening of existing) VBS adjacent to suitable boundaries is recommended as a simple and inexpensive nature-based solution for limiting agricultural soil loss and promoting habitat connectivity.For erosion control, VBS implementation should ideally be employed alongside other measures addressing risk generation at the source.
Our method is transferable, facilitating analysis of alternative study sites and can be applied at different spatial scales (field to sub-catchment).We have made our approach available as an accessible GIS tool that could be used by land managers as part of a suite of strategies for reducing soil erosion in agricultural catchments.
Furthermore, the South Downs Local Plan (South Downs National Park Authority, 2019) promotes the restoration of degraded characteristic landscape features including traditional hedgerow and other field boundaries.With the well documented and on-going soil erosion and runoff issues, and competing agricultural, cultural heritage and environmental interests, the study area provides a suitable testbed for developing and applying a new method for identifying suitable sites for VBS.

2. 2
| Data preparation Our method requires application of a Digital Terrain Model (DTM) and multi-spectral satellite imagery to calculate sediment accumulation risk along flow pathways.A comprehensive 1 m DTM (Environment Agency, 2022) was used with the D8 flow routing algorithm (O'Callaghan & Mark, 1984) to generate sediment accumulation pathways based on slope gradient and contributing area.In flow accumulation modelling, topographic sinks must initially be filled in order to generate a contiguous network (O'Callaghan & Mark, 1984), which may omit isolated natural depressions from a conditioned DTM.Derived pathways are therefore indicative only, and their validity is dependent on empirical observation data.However, higher resolution terrain data can effectively capture finer continuous topographic features (e.g.banks, slopes and drains); this is observable later in our findings where the effect of topographic features occasionally causes pathways to flow parallel to a boundary temporarily before crossing it.Four 10 m spatial resolution Sentinel 2 images spaced through the year (8th September 2021, 22nd November 2021, 26th April 2022 and 9th August 2022) were acquired from the ESA Copernicus Open Access Hub (https://scihub.copernicus.eu/)and used to create an NDVI layer, which served as an accumulation weight and proxy for sediment erosion and transportation (European Space Agency, 2023).
are shown as black triangles.(b) Location of the study site (black outline) within the lower Rother catchment (red outline).(c) The lower Rother catchment (red outline) within West Sussex and the South Downs National Park (grey shading).(d) Location within the UK.Source: British Geological Survey (2013); Natural England (2022); Ordnance Survey (2022, 2023).relative modelling framework adequately reveals magnitude of soil erodibility.Every effort was made to obtain imagery with sufficiently cloud-free coverage with even temporal spacing throughout the year, to represent annual average vegetation conditions.The NDVI is scaled from À1 to 1, with water areas producing very negative values, values close to 0 as bare soil, and high positive values representing denser vegetation.NDVI index values were resampled to match the DTM resolution using bilinear interpolation, suitable for capturing in-field variation.The UK Centre for Ecology and Hydrology Land Cover Map 2021 (LCM2021) was used to extract contemporary field boundaries.
shows these HRPs plotted in relation to sediment accumulation and inverse NDVI values.The highest risk points are generally located earlier on in the pathways, the exception being 1B along P1 (Figure 3a), with this high-risk site located further downslope.In most cases, high risk sites occur at locations preceded by relatively high or increasing NDVI values, excluding 1A, 2C, 3B, and 3C.P1 consists of 166 accumulation points along a length of 189.2 m, of which two points are classified as HRPs.Along the pathway, accumulation increased by 91.5%, or at a rate of 4.8% per 10 m.P2 consists of 347 accumulation points along a length of 371.9 m, with four points classified as HRPs.Accumulation increases by 600.9% (16.2% per 10 m).P3 consists of 283 accumulation points along a length of 339.6 m, with three points classified as HRPs.Accumulation increased by 33.5% (1% per 10 m).P4 consists of 114 accumulation points along a length of 131.4 m, with two points classified as HRPs.Accumulation increases by 158.1% (12% per 10 m).Overall, P2 has the greatest overall increase in accumulation rate, while P3 shows the lowest increase.Caution should be exercised when comparing pathways as they were subjectively generated at our chosen accumulation threshold and represent different locations within the hydrological network, thus exhibiting different characteristics.Percent accumulation increases are therefore relative to each pathway.Elevation, slope, inverse NDVI and sediment accumulation increase percentage values for all HRPs across P1-4 are summarized in Table1.
flow and in turn entrain and transfer greater sediment volume here.Despite limited habitat connectivity potential, this boundary could therefore conceivably offer significant potential for VBS implementation at its upslope margin by diluting risk close to where sediment transfer is greatest and limiting transfer from the first field to the next.Aerial imagery also highlights clear erosion in the southernmost corner of the upslope field, further promoting intervention at this existing margin.Boundary 2 is a former boundary 70 m downstream of P1 and 102 m downstream from the nearest HRP.It marks the western edge of a heritage farm track not depicted in contemporary land cover data, downslope of improved grassland.Close alignment of this historical boundary with aerial imagery suggests it represents the historical F I G U R E 2 Erodibility-weighted accumulation successive percentage increase across the site.Derived pathways are shown as dark blue lines, with larger light blue points indicating locations with greater percentage increases and vice versa, overlaying the inverse Normalized Difference Vegetation Index layer.The locations of previous erosion observations from Boardman (2016) are shown as black triangles.Light blue lines represent the adjoining downslope watercourse (Source: Ordnance Survey, 2021, 2022).configuration of boundary 1 and is contemporarily marked by a hedgerow.Aerial imagery also reveals the presence of an established VBS along the downslope margin and limited vegetation along the upslope margin.Where P1 crosses the boundary, slope data indicate a slight downslope descent onto the track, which then serves as a conduit further downstream.Three modelled pathways flow through this boundary; thus, implementation of VBS on the upslope margin may be a suitable option to promote sediment deposition, reduce risk of muddy flooding across the track and extend the existing vegetation corridor to the north of this upslope field.
(3-7) is compact.Boundary 3 appears to show the historical arrangement of boundary 4.These boundaries are sited approximately 365 m downstream of the nearest HRP, and downstream of two observed F I G U R E 3 Normalized sediment accumulation and inverse Normalized Difference Vegetation Index values for the four pathways identified for further analysis (P1-4), with locations of identified HRPs highlighted in red.erosion points.We suggest that widening of existing margin vegetation visible in aerial imagery may enhance erosion control to limit transfer of material sourced from upslope fields.Boundary 5 is an historical boundary 28 m upstream of the nearest HRP and appears to have been removed.Its reinstatement could offer benefits through early intervention erosion control and provide additional heritage and ecological value.Intervention adjacent to boundaries 6 (current) and 7 (historical) could aid runoff reductions close to an observed erosion location.However, boundary 6 could not be verified from aerial imagery inspection, therefore reinstatement of boundary 7 margin vegetation could collectively limit runoff risk and promote cultural heritage values.Overall there is little opportunity for improving habitat connectivity via field margins along P2.
are shown as black triangles.Grey arrows indicate flow direction.The inset on (a) reveals the northerly extent of boundary 7 upslope of P1, near to an erosion observation point.The red dashed line visible in (a) indicates the study area boundary.runoff from an adjacent pathway to the east and additionally extend the habitat corridor.Boundary 3 is current and situated 33 m upstream of the nearest HRP.It adjoins the boundary recommended for intervention at P2 (boundary 2), both following the upslope track edge, which is marked by a hedgerow inferred from aerial imagery.The topography reveals a reasonable descent from the downslope margin into the next arable field.Land managers might therefore consider broader continuous VBS implementation along this margin adjacent to the existing hedgerow to deliver combined erosion control benefits for both P2 and P3.This option could promote grass strip habitat provision in this location but may carry no heritage value.Boundaries 4 (historical) and 5 (current) appear to represent the same boundary.Aerial imagery highlights existing VBS approximately 5 m wide between boundaries 4 and 5. Erosion has been observed upstream, and the existing VBS may already be promoting sediment deposition downstream of boundary 5. Widening existing VBS in the northerly margin could therefore potentially enhance deposition of material sourced from the upslope arable field, limiting its transfer into the downslope field.4.1.4| Pathway 4 Boundary 1 is current and located 123 m upstream of the nearest HRP (Figure 4d).Notably, this is also the lowest ranking boundary for P3 (boundary 5), suggesting that widening of the vegetated margin north of this boundary may be an efficient measure for limiting runoff both downstream of P3 and upstream of P4.Boundary 3 appears to F I G U R E 5 Slope in degrees, showing steeper areas in darker orange and flatter areas in lighter orange, with all other map elements consistent with those shown in Figure 4. represent the former configuration of boundary 1, thus, rank scores are similar.Aerial imagery shows a wooded hedgerow bounded by a 6.5 m wide grassland strip between the boundary lines, and slope data indicate a reasonably steep downslope descent into the southerly arable field (Figure ested in evaluating the dynamics of soil erosion and runoff risk in agricultural environments.The model affords flexibility and scalability, enabling selection of specific catchment locations and pathways for VBS implementation, guided by empirical field observations and local management priorities.It generates a simple risk map of sediment accumulation showing highest risk locations along pathways.We observed more HRPs earlier on in our chosen pathways (Figures 3-5), scriptive; other users may qualify different pathways as more worthy of investigation depending on individual catchment conditions and management priorities.If practical, land managers should consider employing VBS alongside targeted on-site nature-based mitigation measures at high-risk locations within fields, for example, installing ponds or wetlands, which are optimally sited in wetter locations to further retain sediment and contaminants (Babbar-Sebens All field boundaries relevant to P1-4, showing source of boundary data, location in relation to the pathway, distance to the nearest HRP along that pathway in metres, sediment accumulation percentage increase value of the nearest HRP, inverse normalized distance to HRP scores, normalized HRP accumulation percentage increase scores, combined normalized scores and overall rank for that pathway.