Soil geohazard mapping for improved asset management of UK local roads

Unclassified roads comprise 60 % of the road network in the United Kingdom (UK). The resilience of this locally important network is declining. It is considered by the Institution of Civil Engineers to be “at risk” and is ranked 26th in the world. Many factors contribute to the degradation and ultimate failure of particular road sections. However, several UK local authorities have identified that in drought conditions, road sections founded upon shrink–swell susceptible clay soils undergo significant deterioration compared with sections on non-susceptible soils. This arises from the local road network having little, if any, structural foundations. Consequently, droughts in East Anglia have resulted in millions of pounds of damage, leading authorities to seek emergency governmental funding. This paper assesses the use of soil-related geohazard assessments in providing soil-informed maintenance strategies for the asset management of the locally important road network of the UK. A case study draws upon the UK administrative county of Lincolnshire, where road assessment data have been analysed against mapped clay-subsidence risk. This reveals a statistically significant relationship between road condition and susceptible clay soils. Furthermore, incorporation of UKCP09 future climate projections within the geohazard models has highlighted roads likely to be at future risk of clay-related subsidence.

in particular those clay soils prone to volumetric shrinking and swelling, have resulted in considerable structural damage to their highway networks (Table 1). Highways assets themselves exacerbate developing drought conditions by prohibiting the permeation of water into the underlying soils (Harrison et al., 2012).
A number of drivers of soil-related ground movement impact on UK highway 10 networks, including: clay shrinkage and swelling; sand-washout; compression of soft soils; and peat shrinkage (Pritchard et al., 2014). The county of Lincolnshire reported the largest impact of drought conditions on its highways in 2003 (Table 1), indicating that it is particularly prone to ground movement due to the county's abundance of shrink-swell susceptible clays soils. As clay-related shrink-swell is the dominant form 15 of ground movement in the UK, this paper examines the impact of clay-related shrinkswell on road infrastructure. Soil surveys, which classify the uppermost layer (0 to ca. 1.2 m) of the earth's surface, provide an understanding of soil properties and their spatial distribution. However, despite the intimate link between the soil and the infrastructure it supports, 20 the application of soil surveys in highways asset management has seldom been undertaken within the UK. Elsewhere, this is not the case, examples including: USA (Bauer, 1973;Santi and Martens, 2003;Allemeier, 1974;Lee and Griffiths, 1987;Beatty and Bouma, 1973), Netherlands (Westerveld and Van Den Hurk, 1973) and Australia (Murtha and Reid, 1976;Biggs and Mahony, 2004). Whereas it is recognised that soil 25 surveys have an established role (Hartnup and Jarvis, 1979), the UK still tends to look only to deeper geological maps for indications of surface movements.
Highways engineers and asset managers often have a civil engineering background, and many posess only general knowledge of earth sciences, geology and geotechnics. Introduction Clear spatial information which describes the risk of hazardous ground conditions, without the need for geological interpretation, is therefore potentially of great benefit to practicing engineers (Royse, 2011). Thematic soil-related geohazard maps, derived through reinterpretation of traditional soil maps and fusion with meterologial data, can provide decision makers with a clear view of the potential hazards affecting their assets.

5
This paper aims to demonstrate how soil-related geohazard assessments can serve within a decision support tool in the asset management of local highways. Many tangible associations have been posited with regard to the impact of geology/soils on road condition (Willway et al., 2008), especially in light of climatic change. However, quantitative analyses are currently lacking, providing this study with its novel approach. 10 To achieve this aim, empirical road condition data provided by Lincolnshire County Council (LCC) has been intersected spatially with an existing soil-related geohazard model to understand the statistical and spatial relationship between the two datasets. Furthermore, we investigate the use of UKCP09 climate projections to understand future clay-related subsidence risk in Lincolnshire. Subsequently, suggestions and 15 recommendations resulting from the use of this approach for planned operational maintenance in Lincolnshire is described. Finally, we consider how this can form a framework for other local authorities to follow, both in the UK and internationally.
2 Lincolnshire study area 2.1 Climatic, topographical and geological setting 20 The administrative county of Lincolnshire is situated in mid-eastern England, spanning from the Humber Estuary in the north, to the Wash in the south of the county (Fig. 1). As a result of its flat, fertile lands, a large area of the county is devoted to high intensity agriculture. The county's relief is predominantly low-lying (0-50 m a.s.l.) excepting the Lincolnshire Wolds in the central-northern area of the county; a chalk outcrop, where Introduction This area of Eastern England has a relatively mild, temperate climate. Annual rainfall is relatively low, for example in Lincoln it is 577 mm. Consequently, it is one of the driest areas in the country (Hough and Jones, 1997). This low rainfall, coupled with high evapotranspiration rates, contribute to some of the highest soil moisture deficits (driest soils) in the UK in an average year (Hodge et al., 1984;Robson, 1990).

5
The superficial geology and subsequently the soil parent material of Lincolnshire is predominantly derived from Pleistocene and more recent deposits (Hodge et al., 1984). The influence of previous glaciations and regressions and transgessions of the North Sea around the Wash have led to extensive deposits of silts, clays and the formation of peat material (Chatwin, 1961). Marine and riverine alluvium, and glacial till represent a large proportion of the deposits, with glaciofluvial deposits dominating the Lincolnshire Wolds. As a result, Lincolnshire contains a broad range of soil types, encompassing 73 soil associations (Cranfield University, 2015). Further details of soil types and their distribution in Lincolnshire can be found in Hodge et al. (1984).
Industrial drainage in the fenlands of Lincolnshire since the 1600's have caused such 15 extensive peat wastage that previously underlying clays and silts are now exposed at the ground surface. Substantial thicknesses of peat are therefore now confined to the edge of the fens or to areas remaining undrained (Hodge et al., 1984 The majority of unclassified highways in Lincolnshire, as with many other UK counties, are deemed to be evolved. Evolved roads have not been designed to modern engineering standards and have instead evolved from historic or even ancient roads, even dating in some specific instances to the Roman and Bronze Age eras (Astbury, 1958).

5
A Coarse Visual Inspection (CVI) is a nationally defined standard of assessing road defects and is principally used on the UK's minor road networks. Further technical information regarding this survey can be found in Wallis (2009). CVI surveys revealed that Lincolnshire's unclassified road network has been subjected to severe drought-related subsidence, particularly during 2003 (Table 1) (M. Coates, personal communication, 2015). Moreover, during the 2010-2011 period, approximately 154 road sections were highlighted by the CVI survey as having been damaged as a direct result of drought conditions realising the shrinkage potential of clay rich soils (Fig. 2). These two event years led Lincolnshire County Council (LCC) to place a bid to central government for additional emergency road funding. 15 LCC highlight several safety hazards arising from soil-related drought damaged roads, which are often unclassified, and have speed limits of up to 60 mph (miles per hour). Specific risks include: increased deterioration of the longitudinal profile of roads, requiring drivers to reduce speeds below those that would normally be appropriate for the width and 20 alignment of the road; severe localised transverse depressions, which require drivers to slow to speeds below 20 mph; longitudinal differential settlement and cracking, requiring supplementary road signs to warn drivers and allow roads to remain open;

25
defects which are not easily seen at night or in adverse weather conditions, increasing risk to motorists.

Natural Perils Directory (NPD)
The Natural Perils Directory ™ (NPD) geohazards thematic dataset, developed and maintained by staff at Cranfield University, provides a detailed and comprehensive assessment of the environmental vulnerabilities to building structures and 5 infrastructure posed by soil-related geohazards (Jones et al., 1995;Hallett et al., 1994). NPD represents a thematic reinterpretation of the national soil map (NATMAP) which shows the spatial distribution of all ∼ 700 soil series in England and Wales (Cranfield University, 2015). This soil data, alongside climatic data and expert knowledge are encompassed within a "Land Information System" (LandIS) (Keay et al., 2009). LandIS 10 is regarded as the principle source of soil information for England and Wales by Defra (Department for Environment, Food and Rural Affairs). A core component of the NPD is the clay-related subsidence model, or the underground foundation stability (UFS) model. Based upon a pedo-climatic approach, UFS assesses the likelihood of a soil to undergo shrink-swell and subsequently, 15 whether a potential soil moisture deficit (PSMD) is present for shrink-swell potential to be realised. Once classified, a 9 point vulnerability class, ranging from Extremely High to Extremely Low is assigned. The NPD model output is expressed in GIS (Geographical Information System) format on a vector polygon basis across England, Wales and Scotland (Scottish soils data provided by the James Hutton Institute). This 20 makes it suitable for easy integration with other geospatial data (e.g. infrastructure networks). It is predominantly used by the insurance, reinsurance and water sector. However, its use in highways asset management has not been explored until now.

Highway condition data
CVI assessment data for ∼ 4400 km (75 % , 2007-2011, 2011-2012, 2012-2013 and 2013-2014. CVI is expressed as a series of indices, including assessments for wearing course, edge effects and structural condition. Following discussion with LCC, it was identified that the "structural condition index" provided the most suitable index in understanding 5 the effects of soil on the network, also incorporating edge defects. Conversely, wearing course degradation is a factor of traffic use, direct climatic effects and road surfacing techniques and so therefore soils do not exert a direct effect. Each structural condition index GIS vector polygon represented a 50 m rolling average survey area, where increasingly higher values indicated a worsening structural 10 condition of the highway (P. Shevill, personal communication, 2015); values ranged between 0 and 93. Generally, a value of < 40 represents a road in a good state of repair, whereas values > 40 require further investigation and likely treatment options.

UKCP09 future climate projections
UKCP09 climate projections (Jenkins et al., 2009) indicate that the UK is likely to 15 experience hotter, drier summers and warmer wetter winters, especially in the southeast and east Anglian region of England, for the forthcoming century. Such weather patterns will exacerbate clay-related subsidence.
The long lifespan of road infrastructure in the UK leads to particular susceptibility to hazards under future climates (Willway et al., 2008). Research undertaken by 20 the authors as part of the Infrastructure Transitions Research Consortium (www. itrc.org.uk) project has involved the modelling of probabilistic potential soil moisture deficit (PSMD) for the future scenarios of 2030 (2020-2049) and 2050 (2040-2069). Model parameters, daily rainfall and potential evapotranspiration (Eq. 1) were derived from a spatially coherent, 5 km 2 resolution gridded output from the UKCP09 weather  PSMD represents the fundamental climatic control on clay-related subsidence and so has been incorporated within the NPD geohazard model (Pritchard et al., 2015). Future projections of clay subsidence risk for Lincolnshire are presented in Fig. 3. These enable assessment of future risk from ground movement for Lincolnshire's unclassified road network. 5 4 Risk assessment

Overview
We have determined statistically the relationship between CVI value and claysubsidence risk class from NPD for each survey period. This section describes the GIS and statistical framework used to assess the impact of clay-related subsidence 10 on Lincolnshire's unclassified road network, illustrated in Fig. 4. GIS software was employed to provide a platform for rapid analysis and handling of spatial data (Fedeski and Gwilliam, 2007), including CVI and clay subsidence risk class. CVI data provided by LCC required processing to make the data suitable for intersection with the NPD data (Fig. 3). Polygon data was converted to points in 15 ESRI's ArcGIS (v. 10.2) using the polygon centroid tools, and the open-source software, Geospatial Modelling Environment (www.spatialecology.com), was used to snap these points to the road section (vector line feature). Road sections (GIS format) for the entire Lincolnshire network, classified by road hierarchy, were provided by LCC. Processed CVI data was then intersected with the NPD geohazard dataset and the resultant 20 GIS shapefile attribute data imported directly into the statistical package, R (R Core Team, 2014) for further statistical analysis. The number of CVI points assessed per subsidence risk class are presented in Table 3. Similarly, future subsidence projections have been intersected with the Lincolnshire unclassified road network within the GIS in order to understand the change in potential exposure throughout these scenarios.

Current risk
Our results suggest that a spatial and statistical relationship exists between clay-related subsidence risk and CVI (structural index) value. The boxplots in Fig. 5 present the relationship between CVI over the survey periods of 2007-2011, 2011-2012, 2012-5 2013 and 2013-2014. Due to the extensive unclassified road network in Lincolnshire, a 4 year rolling CVI assessment program is used by the Council. Therefore, each survey year often represents different road sections to those assessed in the previous year. Ideally analysis would have considered many years/decades. However, the use of GIS in Lincolnshire's highways asset management remains a relatively emergent technology. Prior to GIS techniques, CVI data recording was predominantly paperbased and is therefore difficult to investigate alongside other environmental data.
Overall, the CVI structural index for Lincolnshire's unclassified network shows significant deterioration between the years 2007 and 2014. This is especially so on soils with an Extremely High to High risk of clay-related subsidence, where consistently 15 high CVI values (i.e. > 60) are observed; representing a significant deterioration of the structural road condition on these higher-risk soils: roads identified at current risk are provided in Fig. 7. Moreover, by 2012-2013 the CVI value on these soils has reached a critical point, whereby the roads are deemed to have failed structurally (i.e. CVI > 80), resulting in the enforcement of speed restrictions as a result. 20 CVI for roads on soils at a Medium Low risk for all survey periods showed a consistently high level of subsidence risk. However, this is only representative of a relatively low number of CVI points for each survey period (9, 10, 10 and 10, respectively) (Table 3). These values were therefore excluded from the analysis.

Future risk
The entire unclassified road network for Lincolnshire was intersected with future projections of clay subsidence risk for 2030 and 2050 (Fig. 3), using the 50th percentile or central estimate. From this, metrics were produced that identify lengths (kilometres) of road sections falling into each particular clay-related subsidence risk class. The 5 resultant graph (Fig. 7) shows clear shifts of road sections into higher vulnerability classes from the baseline (1961-1990) through to 2050 (2040-2069). This is especially so for the Extremely High class where between the baseline  and 2050, there is a 300 % increase in the amount of road length on these soils. There is also a ∼ 1200 % shift from roads being at a Medium risk to those becoming at Medium 10 High risk between baseline and 2030 and 2050 scenarios. However, Medium High risk soils do not appear to exert such structurally damaging effects on road condition as compared with the higher classes (Fig. 5).

Discussion and conclusions
The low traffic volumes on local road networks mean that the large capital investments 15 required to adequately engineer all unclassified roads on highly shrinkable soils are an unaffordable solution in mitigating the impact of soil-related subsidence. Road deterioration is affected by a number of factors, however, this paper has shown quantitatively that clay-subsidence prone soils exert a profound structural impact upon road condition in the Lincolnshire study area. This has resulted in subsequent 20 economic impacts, with calls made by the county for emergency funding from the DfT to clear the maintenance backlog. ADEPT (2009) argue that climatic change may lead to wide-scale failure of the UK minor highways network. As this work shows, UKCP09 scenarios indicating hotter, drier summers and warmer, wetter winters through to 2080 (Jenkins et al., 2009)  and less visible geohazards such as clay-related subsidence to highway infrastructure. An analysis of the impacts of current and future geohazards to road infrastructures nationally will lead to a greater awareness amongst local authorities and policy-makers. It will also lead to a better understanding of the viable adaptation and mitigation options which can be implemented to tackle the issue at hand (Williams et al., 2012) as well as informing the debate on infrastructure investment planning. The construction of entirely new roads in the UK is rare. More commonly, existing networks are upgraded (Brown, 2013). Rawlins et al. (2013) state that with new developments, an awareness of the potential hazards and the influence of climate change should be incorporated into design principles. As a result, the findings of this 15 research can aid planning of new highways.
The All Party Parliamentary Group on Highway Maintenace recently called for the incorporation of highway asset management plans (HAMP) within local authorities highways departments to become mandatory (APPGHM, 2013). Moreover, recommendations exist to make funding streams from the Department for Transport 20 accessible only if HAMP's are in place. Therefore, specific risk information, in this case relating to the spatial distribution of hazardous soils, are vital in supporting these asset management plans. Within LCC, integration of clay-related subsidence assessment within the planning of highway maintenance has provided the basis for a decisionsupport tool for establishing which specific treatments highways engineers can use 25 to improve drought-damaged road sections. Whereas previously a blanket approach to resurfacing would have been applied to all affected sections, it is now recognised that unclassified road sections on drought susceptible soils are predominantly prone to failure. Therefore, with large capital investment not being an option, other value-formoney options are being sought. LCC have made attempts to reinforce their road network, for example, with the use of steel reinforcing grids (e.g. at A1073, Crowland) which act to reinforce the road structure on clay-subsidence susceptible soils. Although this technique has proven succesful, it is both expensive and proves problematic when resurfacing works or a utility trench has to be emplaced below the road surface, so it is not a wholly viable option.
More recently, LCC have been trialling an in situ road recycling process known as retread, for specific drought affected sections of their network. This process involves the 10 re-incorporation of in situ road planings into the foundation of the road. Retread offers an in situ treatment, using cold-laying techniques, therefore heat and more energy is not required (Heaton, 2014). Being in situ this also reduces the high disposal costs of potentially hazardous (bitumen-containing) waste to landfill. Moreover, road planings acquired from other sites across the network have been imported in some areas 15 to further deepen road foundations to try and minimise the impacts of clay-related subsidence.
A sum of GBP 1 million, within the Lincolnshire road maintenance budget, has been sourced from a bid to central government, and assigned to the retread project over the period 2013-2015. This process, although having been implemented in other counties 20 for a number of years, is a new initiative by LCC. Due to this approach being novel in Lincolnshire, it is not currently known what the long-term reliability of this method will be on Lincolnshire's subsidence-prone soils, however, current results appear positive (Heaton, 2014).
Although not explored within this paper, the shrinkage of clay soils is a known 25 precursor to shallow and deep seated slope movements (Page, 1998). In this situation, shrinkage cracks allow water ingress to reach clay shear surfaces (Loveridge et al., 2010). This results in increased pore-water pressures, especially during intense storms. High pore water pressure with additional weight from large agricultural vehicles  (Hawkins, 2013;Loveridge et al., 2010). As a result of agricultural and drainage practices, much of the highway network in Lincolnshire is flanked by steep-sided drainage dikes, or has been left raised above the surrounding land due to peat shrinkage. Therefore, these steep sided embankments and the 5 presence of shrink-swell soils can lead to specific localised slope failure (e.g. Fig. 8). This paper has assessed the use of soil-related geohazard mapping in the asset management of minor highway networks within Lincolnshire. The risk-assessment framework (Fig. 4) presented has enabled LCC to prioritise its limited road maintenance capital on road sections at risk from drought-related clay subsidence. It has also acted 10 as the basis for a decision-support tool, making highways engineers more conscious as to where hazardous soils present themselves spatially, highlighting the damaging effect that these can have upon the unclassified road network during drought conditions, as well as guidance as to which treatments prove economically and structurally viable.
The study also highlights the structurally damaging effects that are likely to occur 15 on England's unclassified road network under the UKCP09 climate projections, and on soils prone to clay-related subsidence, especially in the south-east of England. The incorporation of probabilistic subsidence projections provides a novel approach to evaluate the future risk of soils to highway networks. The standardised CVI method of road condition survey within the UK means that rapid soil-geohazard assessment could 20 be readily applied to other administrative areas affected. Furthermore, with appropriate soils data (e.g. national soil survey maps) and empirically derived shrink/swell assessment, the methodology presented here could be used internationally. Further research is required to consider the potential impacts of other infrastructures (e.g. utility trenches and water mains failures) on the road network, and what impact these have 25 upon the soil structure on which these roads are founded.   Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Williams, K., Joynt, J. L. R., Payne, C., Hopkins, D., and Smith, I.: The conditions for, and challenges of, adapting England's suburbs for climate change, Build.