4 GIS-BASED ASSESSMENT 4.1 INTRODUCTION 4.2 BACKGROUND 4.3 DEBRIS FLOW POTENTIAL ASSESSMENT 4.4 GIS METHODOLOGY 4.5 RESULTS 4.6 CONCLUSIONS
4 GIS-BASED ASSESSMENT
by M Harrison, A Gibson, A Forster, D Entwisle and G Wildman
This section describes the development of the GIS-based assessment tool for debris flow hazard assessment in respect of the entire Scottish road network and illustrates the data and results from this process.
In order to ensure that a comprehensive knowledge of debris flows, the road network and the interaction between these two entities was fully captured within the work being undertaken a working group was formed. This group comprised the authors of this section, who were tasked with producing the actual assessment, the editors of this report and other specific individuals with relevant experience (see acknowledgements).
A series of four meetings were held to consider the following issues:
Meeting 1: The specification for the work.
Meeting 2: The available data sets and their relevance to the task in hand.
Meeting 3: The scorings and weightings to be assigned to each data set.
Meeting 4: Fine-tuning of the results from scorings and weightings in the light of the group’s knowledge and experience.
The process of knowledge capture and input used in this work was akin to the process that is used to capture information and develop rules for knowledge-based systems (e.g. Winter and Matheson, 1992). This approach forms a vital part of any knowledge and rule-based interpretation of data.
The methodology developed is based on assessing the propensity for debris flow formation. In order to establish the hazard to sites on the road network, further interpretation of the outputs is required in order to establish the likelihood of any given area that exhibits a propensity to debris flow formation producing a flow that might intersect the road (see Section 5).
GIS-based imagery relevant to the local road network was distributed to the individual Local Authorities for further action in the context of their particular needs.
4.2.1 Causal Factors for Debris Flow in Scotland
Winter et al. (2005a, p30, 31, 58), describing the findings of the initial study, identified 86 different factors that contribute to the debris flow hazard to the Scottish Roads network. Each of these factors is valid, and could be individually considered for inclusion in a system that seeks to model debris flow potential. However, for the purposes of this study it was important to use datasets that possessed reasonably consistent coverage across the whole country and for which some form of quality could be assured and for which availability could be guaranteed.
In effect this meant that many ‘point’ datasets, such as borehole records and site investigations could not be included within the analysis. However, many of the properties recorded as ‘point’ data are to some extent, described by spatially continuous datasets; for instance, data on permeability, grain size, and cohesion are intrinsically linked to polygons of different lithologies described by a geological map.
The working group identified three relevant data sources that were available for the entire study area:
1. BGS DiGMap: GIS layers of geology at 1:50 000 scale showing bedrock and superficial deposits (supplied by BGS). Each polygon of the geological map is attributed with a code that describes the litho-stratigraphic unit to which the rock type belongs. That is, each polygon is labelled with a code that describes the polygon in terms of the type and age of the rock.
2. NEXTMap Britain: a digital terrain model derived from the INTERMAP Digital Terrain Model product. NEXTMap is a high-resolution elevation model of Great Britain. It was generated from a 2005 airborne survey in Scotland where the time it takes for a signal to be sent down to the ground and bounce back was measured. This was calibrated with a GPS on board the plane to give the height of the ground surface, accurate to 0.5 m.
The initial NEXTMap product was a digital surface model that represented the height of the surface of the ground. This dataset contains all ‘cultural’ features such as buildings and wooded areas. The second product from INTERMAP is a digital terrain model, which is the same product but with the cultural features removed. The algorithms that have been employed to remove these features are in the most part very effective. However, some areas of woodland are still shown by areas of raised elevation. Although this can cause localised error in the data, the NEXTMap digital terrain model is a very accurate and high-resolution dataset, and it provides continuous coverage for all of Scotland.
3. CEH (Centre of Ecology and Hydrology) land use data: CEH Landcover 2000 is a digital map that gives a comprehensive picture of the UK Broad Habitats (LCM 2000). Sixteen Target classes (Level-1) and 27 subclasses (Level-2) allowed construction of the Broad Habitats. The subclasses are described in greater detail in Level-3. It was mapped by analysing satellite spectral reflectance data on a grid of approximately 25 m square pixels. Ground survey assessed the spectral characteristics of the Broad Habitats and an automated system selected the most likely class for each pixel in a remotely sensed image. Accuracy was checked against ground survey and other information. The minimum mappable unit is about half a hectare. Data is available as digital outlines of the level 2 subclasses, which are treated as ‘objects’ in ArcView.
At an early stage of the research, it was proposed that rainfall data was also included. However, the working group concluded that, as intense rainfall could occur anywhere within the geographical study area, it was not necessary to include this as a separate factor. Therefore, Meteorological Office rainfall data have not been utilised at this stage.
The research described by this report considered how best to integrate aspects from the three data sources described above to provide a reasonable model for debris flow hazards affecting the Scottish road network. This has mainly been carried out through an iterative process of attributing or manipulating each dataset to represent as many of the factors described by Winter et. Al. (2005a). Thus, expertise in the geology of Scotland has been applied to DiGMap to change the standard attribution of polygons (age and type of rock) to numerical codes that estimate bedrock permeability and the degree to which source material for debris flows can be formed.
The working party concluded in the initial study that five main components should be considered when determining the hazard potential of debris flows affecting the road network:
1. Availability of debris material.
2. Hydrogeological conditions.
3. Land Use.
4. Proximity of Stream Channels.
5. Slope Angle.
It was considered that information regarding each of these could be usefully retrieved from the datasets described in Section 4.2. The interpreted data could then be combined to produce a working model of debris flow hazard that could be validated by comparison with scenarios taken from accounts of investigated debris flows.
The possibility of seismic acceleration as a causal factor was raised at a working group meeting. After discussion it was thought unlikely to be a significant factor in the generation of debris flows that could impact significantly of the trunk road network as the ground accelerations developed by anticipated earthquakes were an order of magnitude less than those typically generated by heavy construction plant. However, further research into this subject may be useful at a later stage to properly consider the implications of a major seismic event.
For a debris flow to occur, there must be an available source of material, usually granular, often with a very wide particle size range in such a state that it would easily be mobilised by the action of a fluid (usually water). Thus the material that has the highest potential for debris flow activity is likely to be non-cohesive, with significant particle granularity. Material that is cohesive due to high clay content or inter-granular cement would be difficult to mobilise.
126.96.36.199 Analytical Method
On this basis, the lithologies represented by polygons in DiGMap were interpreted against a scale that indicated the degree to which the bedrock or superficial unit at surface would provide non-cohesive granular material as a source of debris. The ROCK_D (BGS Rock Description code) attribute of each polygon was re-interpreted by asking the following two questions:
‘Is this material capable of being mobilised by water into a debris flow in its fresh, unweathered state?’
‘Is this material likely to have a weathered regolith or covering of head that could be mobilised by water to form a debris flow?’
Each ROCK_D description has been assigned a number on a scale of 1 to 10 to give an indication of its potential to supply the material, from within its outcrop, that would be capable of generating a debris flow. The judgement is based on the indicated grain size distribution or its assumed probable grain size distribution (for superficial material) or the likely ‘block size’ distribution of near-surface material/regolith (bedrock materials) as inferred by expert judgment. Table A.1 (Appendix A) shows the criteria used to assign each of the 4200 BGS codes identified. These are based upon the general principles outlined by Terzaghi (1955a, 1955b) and updated in BS8002:1994, BGS geologists’ knowledge of each lithology and guidance from the working group.
In consultation with the working group, these were adjusted to account for the significance given to chemical weathering of certain rock types and the lower likelihood of the generation of clays by chemical weathering in some lithologies in Scotland. The interpreted codes were stored within the BGS Scottish Debris Flow Attribution Database (BGS SDABD).
188.8.131.52 Analytical Method – ‘Accumulation Zone’ Supplementary Dataset
It was recognised by the working group that in many locations, BGS data did not record the presence of peat or other deposits that may form sources of debris flow material. This was a function of the age of the BGS data used and the mapping methods historically employed by BGS mapping teams. In previous decades, priority was given to recording the presence of bedrock materials – superficial materials were considered ubiquitous and not mapped. To counter this, the working group recommended that a method be sought that could identify areas where deposits of non-cohesive material could collect and form source areas for debris flows.
The NEXTMap digital elevation model was analysed, using GIS to identify those areas where material was likely to accumulate. The analytical method used, highlighted those areas where changes in the shape (morphology) of the ground meant, that a flow of water would be slowed, and any material held in the water flow might be deposited. The two types of slope are identified by the method are:
1. Convergent slopes – where horizontal bends in the ground mean that flows come together to form ‘sinks’, (Figure 4.1). This is termed a change in horizontal or ‘plan curvature’.
2. Slope bases – where the relief of the ground changes quickly from a steep gradient to a shallow gradient, (Figure 4.2). This is termed a change in vertical or ‘profile curvature’
Figure 4.1 – Diagram showing how material flows away from (a) higher, convex (divergent) ground (b) and towards lower, concave (convergent) ground (after Shary, 2002).
Figure 4.2 – Diagram to show where accumulation might be expected in an area of relief change, where slope gradient changes from steep to shallow (after Shary, 2002).
Characterisation of plan curvature and profile curvature were analysed together, to identify zones where the ground surface is ‘convergent’, where a flow of water would decelerate and deposition can occur (Figure 4.3). Likewise, the analysis can identify ‘divergent’ ground, where a flow of water would accelerate and erosion would occur.
Figure 4.3 – Landform classification based upon a combination of horizontal curvature and vertical curvature (Troeh, 1964, Shary, 2002).
184.108.40.206 Method to Calculate the ‘Accumulation Zone’ Supplementary Dataset
Although a number of methods are available to perform this analysis, experience in other BGS projects has shown the K-Accumulation model by Shary (2002) works well and is flexible to different types of terrain. The method can be given by the algorithm:
Kaccum is a value indicating the shape of the ground, positive values indicate areas of accumulation.
Meancurve, Plancurve and Profilecurve are all numbers that mathematically describe the shape of the ground, calculated from the 25 m pixel NEXTMap DEM using ArcGIS.
- Meancurve is the average normal section. Positive values highlight a broadly convex slope, negative values describe broadly concave landforms.
- Plancurve is the rate of change of horizontal curvature. Positive values highlight a divergent slope, negative values a convergent slope (Figure 4.4).
- Profilecurve is the rate of change of vertical curvature or slope. Positive values indicate convex slope, negative values indicate concave profiles (Figure 4.4).
Figure 4.4 – Visual representation of plan curvature and profile curvature.
The output from the algorithm is a number rating (Kaccum) that indicates the likelihood that material will be deposited, as a result deceleration of a flow. Although previous use of the method by BGS was to estimate areas where head may accumulate in the south of England, it has been possible to adapt the method to indicate areas where material may accumulate in the Scottish Highlands. This was carried out by an iterative process whereby the original formula was applied to known areas and a visual assessment made, comparing the estimated area of material deposition with local knowledge of areas of deposition. Where the estimated area of deposition was incorrect (for instance on cliffs or in stream channels), the score given by the formula was discounted. After a number of iterations, it was decided that those areas likely to be depositional zones had a values in the range 0 – 1. Table A.2 (Appendix A) shows the result of this analysis, with suggested values to be used in the overall debris flow algorithm.
Figure 4.5 shows the availability of debris material scoring for two areas – Glen Ogle and Inverness – as examples. These same examples will be followed for each of the stages in the report. The diagrams show that the generally more granular, alluvial or glacial materials in the base of the valley in Glen Ogle and more widely distributed in Inverness, have higher scores and therefore greater influence on the result in this part of the methodology. In this analysis these materials are regarded a potential sources areas for landslide debris.
Figure 4.5 – Availability of debris material. Excerpts from the GIS showing the variation in availability of debris material scoring for two areas; (left) Glen Ogle and (right) Inverness. The legend in the diagram refers to this as lithology index, reflecting the scoring of the source material. (OS Data © Crown Copyright. All rights reserved Scottish Government 100020540, 2008.)
4.3.2 Water Conditions
Two aspects of water conditions are relevant to the generation of debris flows.
1. The ability of water, as rainfall or overland flow to infiltrate a potentially mobile deposit (permeability of the deposit).
- This has been taken into account in the scoring for Availability of Debris Material (Section 4.3.1), which combines judgements on grain-size and permeability.
2. The ability of water to remain within the deposit to an extent where pore water pressures can build to a level where the shear strength is sufficiently reduced to initiate failure (permeability of the underlying material).
- A factor was required that would take account of the permeability of the underlying bedrock, this is considered in this section.
The substrate beneath potentially mobile deposits may exert either a positive (destabilising) or a negative (stabilising) input to debris flow generation. A positive input will be generated where the substrate is impermeable. In such a case, infiltration through the surface material is impeded, leading to a build up of pore-water pressures, a lowering of effective shear strength and increasing the likelihood of a ground failure. Most bedrock materials may be locally expected to be relatively impermeable with regard to the timescale of a high intensity rainfall event.
A negative (stabilising) input to debris flow potential is generated where the substrate is permeable. If a debris flow moves over permeable ground it may be slowed by under-drainage – (water draining from the moving mass into the substrate) with a consequent increase in shear strength. It is unlikely that this mechanism will have a significant effect except where the debris flow has flowed onto shallow, very permeable slopes and has spread out to allow under drainage over a large area (as seen in the lowest part of some debris flows).
The permeability of a rock type will be a function of grain size distribution for superficial materials and discontinuity spacing and dilation for bedrock materials. For superficial materials, coarse, clean gravels will be the most permeable and clay the least permeable. Consideration of the permeability of bedrock in Scotland needs to consider the possibility that in most places, relatively impermeable bedrock lies beneath a potentially permeable and mobile regolith. However, depending upon specific rock type, discontinuities in the bedrock may have been developed and dilated by thermal, physical and chemical weathering. At depth, most bedrock lithologies in the study area are likely to be interlocked and unlikely to be incorporated in a debris flow.
It should be borne in mind that, in many locations, there will often be a pre-existing drainage system that will have a significant impact upon the nature and distribution of pore-water pressures. Although such systems are likely to be a significant control on debris flow potential, there is no proven method available at this time that can be used to digitally analyse this using existing data. It was considered by the working group that this may be an avenue for further investigation at another stage of the research.
220.127.116.11 Analytical Method
Lithologies represented in DiGMap were interpreted on a scale that indicated the relative permeability of substrate materials. The ROCK_D (BGS Rock Description code) attribute of each polygon was interpreted by asking the following question:
‘What is the permeability of this rock type?’
Each ROCK_D description has been assigned a number on a scale of 1 to 10 to give an indication of its permeability within its outcrop. The judgement is based on the indicated grain size distribution or its assumed grain size distribution (superficial material), consolidation/cementation and discontinuities. Table A.3 (Appendix A) shows the criteria used to assign each code.
Figure 4.6 shows the result of these assessments in the two example areas. High values represent areas of low permeability hence a higher likelihood of contributing to debris flow formation.
4.3.3 Vegetation and Land Cover
Vegetation may have three beneficial effects in maintaining slope stability:
1. Intercepting rainfall to reduce infiltration into the
2. Removing soil moisture
3. Reinforcement of the ground by a root network.
The amount by which particular vegetation improves the stability of a slope will vary with the type of vegetation. Trees are likely to be more beneficial than shrubs, which would be better than grass.
Other land uses are likely to have adverse influence on slope instability, for instance, bare soil or cultivated (bare) ground would be prone to debris flow, as it is often unbound and in a loose condition. Urban or rural development may also be detrimental to stability due to the possibility of the inappropriate disposal of surface water, or leaking services that may feed water into a susceptible slope leading to high antecedent water level prior to a high magnitude event or a focusing of a high magnitude event such as to initiate debris flow activity.
Figure 4.6 – Excerpts from the GIS showing the variation in water conditions (permeability) for two areas; (left) Glen Ogle and (right) Inverness. (OS Data © Crown Copyright. All rights reserved Scottish Government 100020540, 2008.)
18.104.22.168 Analytical Method
Expert judgement has been used to assign appropriate scores for the land use categories in the CEH land use dataset (Table A.4 of Appendix A) by asking the following question:
‘What is the likely effect of this landcover upon debris flow potential at a site?’
The judgement was based upon the assumptions described above and on guidance from the working group. Each cover type was given a rating between 0.7 and 1.2 to indicate by how much the vegetation may improve stability. The lowest value is for woodland and the highest value for annual crops where the ground is regularly disturbed producing an open structure with little root strengthening. Other land use and vegetation cover have intermediate values.
Figure 4.7 shows the result of these assessments in the two example areas. As can be seen in the Inverness area, the built environment gives a high stabilising factor, whereas in the Glen Ogle example, the vegetation would have a more limited affect on stabilisation.
Figure 4.7 – Excerpts from the GIS showing the variation in Vegetation conditions from satellite data for two areas; (left) Glen Ogle and (right) Inverness. (OS Data © Crown Copyright. All rights reserved Scottish Government 100020540, 2008.)
Peat was initially considered separately, as it may fail in slides and bog bursts in certain environments. When it fails it may form a source of water and material to debris flows and may be part of the initial slide or, if part way down slope, may add impetus to the flow. A method of combining peat and slope that identified areas of peat above roads with slope of greater than 5° was discussed. However, the stability of peat involves many complex factors, some listed below, and a proper understanding of peat behaviour would require field assessment. As peat could not be suitably assessed using available national datasets, it was decided that this was a specialist issue and would not be pursued in detail during this project.
Some factors affecting the stability of peat:
- Peat layer overlies a relatively impermeable material.
- A convex slope or break of slope at its head.
- Proximity to local drainage including seeps, flushes and subsurface flow.
- Connection between surface drainage and the base of the peat.
As a primer to the peat assessment, BGS have used the NEXTMap data to calculate flat areas (that could contain peat) that lie above the trunk road network, with a connecting slope.
The first step was to locate areas of steep and flat ground (Figure 4.8). For the purpose of this exercise, steep ground over which peat could move was assessed as anywhere that had a slope greater than 5 degrees. Flat ground, where peat materials may form was identified as anywhere with a slope less than 2 degrees.
This was performed on a slope model for Scotland, derived from NEXTMap Digital Terrain Models (DTM), resampled to a 50m cell size. ESRI’s Spatial Analysis extension was used to create two grids: one for steep ground and one for flat ground.
The steep and flat grids were converted to shapefiles, resulting in polygons representing instances of steep and flat ground. In some cases these polygons were quite large in size, so they were broken up into smaller polygons for analysis. This was achieved by intersecting the polygons by the trunk road and a catchment dataset calculated from the underlying DTM.
It was decided that only flat areas within a 3km buffer from the roads should be included for analysis, to reduce the data volumes. The intersected layers for steep and flat areas were both clipped to a 3km buffer of the trunk road network.
The polygons representing steep ground for the area around Glen Ogle are shown below.
Figure 4.8 – Ground sloping greater than 5 degrees in the Glen Ogle area. (OS Data © Crown Copyright. All rights reserved Scottish Government 100020540, 2008.)
Zonal statistics, in ArcMap were applied to the intersected shapefiles to find the average height of each polygon. The height was abstracted from the NEXTMap DTM.
Every flat polygon was interrogated. If the flat polygon was within 150m of a steep polygon, and the average height of the flat polygon was larger than the average height of the steep polygon, then the steep polygon was interrogated. If the steep polygon was within 150m of a trunk road, and the average height of the steep polygon was higher than the average height of the road, then the flat polygon that originally intersected the steep polygon was exported to another layer. (The statistics for each road segment have already been supplied as part of the project deliverables. These statistics include average height of road segments, which have been used here.) This exercise was repeated for each flat polygon.
The algorithm used is detailed in Table A.5 (Appendix A).
22.214.171.124 Peat Limitations
This is a fully automated methodology, and is only as accurate as the data used.
The methodology identifies areas of flat ground above both steep ground and roads. It does not necessarily identify the ideal profile for peat flow.
4.3.4 Stream Channels
Stream channels are often associated with debris flows. This is primarily because they may focus the flow of water during extreme events and supply large volumes of water that can mobilise available material. They may also act as collectors for loose material during moderate flows forming debris dams and at times of extreme flow there is the possibility of their actively promoting landsliding of additional material from the walls of the channel and from these debris dams. Thus the working group concluded that identifiable streams should be buffered for an appropriate distance from their centre line to take into account the erosion catchment area and nature of the adjacent material. Discussions within the working group suggested that a buffer, at least, as wide as an assumed 15° side slope should be employed. For a 3m deep channel this would give a buffer width of ±15m and for a 10m deep channel a ±50m buffer.
Using this method, it was found that the buffer covered very large areas of ground. Therefore, for the first iteration of the dataset, it is proposed to use a 50 m buffer centred on stream channels and to score this 10.
126.96.36.199 Analytical Method
The location of streams were automatically generated from NEXTMap digital terrain models using hydrological modelling techniques. The NEXTMap dataset is detailed in Section 4.2.2.
ESRI’s hydrologic modelling toolset from the Spatial Analyst extension in ArcGIS 9.1 was used to generate the stream network. A filter of 1500m was used to ensure that the correct density of streams was identified. Full details of the method are given in by Tarboten et al (1991). As described above, the automatically generated stream network was buffered to a width of 50 m. Any ground within this buffer zone has been given a score of 10.
Figure 4.9 shows the stream locations calculated from the DTM in the example areas. The low-lying Inverness area has many more streams present than the steeper and more deeply-incised Glen Ogle area.
4.3.5 Slope Angle
Slope angle influences the balance of stabilising and destabilising forces on all slopes. When the destabilising forces exceed the shear strength of the materials forming the slope the failure occurs. Therefore, the steeper the slope the greater is the susceptibility of the material to initiate a debris flow.
Figure 4.9 – Excerpts from the GIS showing the processed stream data as extracted from the Digital Terrain Model for two areas; (left) Glen Ogle and (right) Inverness. (OS Data © Crown Copyright. All rights reserved Scottish Government 100020540, 2008.)
188.8.131.52 Analytical Method
The slope categories significant in the generation of debris flows that were indicated by Winter et al. (2005a) were modified following further discussions within the working group, based on the experience of those present. These were used as the criteria to allocate scores to be included in the overall debris flow hazard assessment Table A.6 (Appendix A)
Figure 4.10 shows the effect of these classifications in the two example areas. Slope is one of the most significant factors in the initiation of debris flows and as can be seen in the diagrams, in the low-lying Inverness area the slope index is very low. In Glen Ogle the pale colours indicate high slope values above the A85.
4.3.6 Weighting of Causal Factors
It was recognised by the working group that it would be important to include some form of weighting factor into the algorithm to allow the relative importance of each factor to be expressed. Although it is impossible to understand, in detail, the precise interaction between each of the factors described, the working group generated a series of weighting factors. These are based upon the knowledge and experience of members of working group involved in the investigation and management of debris flows in Scotland. This allowed different scenarios to be modelled in working group meetings to use real-world examples to validate the model results. The factors are given in Table A.7 (Appendix A).
Figure 4.10 – Excerpts from the GIS showing the processed slope index values as extracted from the Digital Terrain Model for two areas; (left) Glen Ogle and (right) Inverness. (OS Data © Crown Copyright. All rights reserved Scottish Government 100020540, 2008.)
The five variables listed in Section 4.2 (availability of debris material, water conditions, vegetation and land use, stream channels, and slope angle) have been combined in a geographical information system (GIS) in order to analyse their distribution and be able to spatially combine their contributing hazard scores.
The system used to prepare data was ESRI’s ArcGIS 9.1 and ArcWorkstation 9.1. For example, the geology was selected and clipped (cut-out) to the area of interest, and the stream network was ‘cleaned’ to remove anomalous areas such as lochs from the dataset. Once the datasets were clean and ready for analysis they were converted into grids and processed using ESRI’s ArcWorkstation 9.1. ArcWorkstation is a command line driven GIS. Although it doesn’t display the data graphically, so certain processor-intensive functions can be performed more efficiently.
All data were converted into grids with a cell size of 25 m. This was necessary in order to process the data efficiently, though the conversion process was not always straightforward. Grids are a very efficient way of processing large volumes of data and they are ideal when applying weighting factors. Using a simple arithmetic grid calculator, it is possible to multiple every cell in a grid by a certain amount. This enables any final weighting factor to be easily incorporated into the methodology. The weighting factor for each variable is then applied to the grid and resultant grids are added together to produce a final model representing the landslide potential.
Figure A.1 (Appendix A) shows a series of flow charts that summarise each of the methodologies.
For those who do not have access to full GIS, ArcReader is available as a free download from http://www.esri.com/software/arcgis/arcreader/indexl using this tool the GIS format outputs can be viewed, simply use the <Scottish landslide.pmf> (ArcReader published map document) to view the data. The landslide data are provided with the Scottish trunk road network and local road network that BGS were provided with, that has been built into a network that can be viewed in the ArcReader software. For convenience, also located on the data DVD is a coastline as downloaded from the National Oceanic and Atmospheric Administration (NOAA) World Vector Shoreline website and Shuttle Radar Topography Mission (SRTM) 90 m Digital Elevation Model. These two products can be freely distributed. When analysing the Scottish Landslide Data against the SRTM data care should be taken as the Landslide data were produced using a more accurate and validated Digital Terrain Model.
Geotiff (Georegistered tiff images) have been created to load into any GIS or CAD software, or Adobe Acrobat format PDF files have been created for data inspection.
In order to present the data in this form, a legend was developed that summarised the data into classes Table A.7 (Appendix A) show the class values used. The classes are identified on an A to E scale, where A has the least potential to initiate debris flow landslides and E has the highest potential. Figure 4.10 shows the landslide hazard layers for the two example areas. As one would expect from the topography and general lack of contributory factors, the Inverness area in the diagram has very low potential for debris flow initiation. In contrast, the Glen Ogle area shows several areas of high potential mainly focused along stream channels. Certain of these channels were the initiators and focus of the landslide events of August 2004 that began the present study.
That the models identify these areas, indicates that the working group have been able identify the principal factors that led to the 2004 events (Figure 4.11). However the power of the GIS technique is that these same groups of factors have been identified for the whole of the Scottish road network as can be seen in the diagrams Figure 4.12 and 4.13. This is a 1:2,000,000 printout from the final landslide hazard layer. At this scale it is impossible to see the detail included in the 1:50000 modelling, however an overall indication of the level of hazard from debris flows across the Scottish Road network is possible.
The data is also presented in a tabulated form found on the data DVD as ScottishRoadLandslideStatistics.mdb. In this form the data were summarised using the database primary key of the trunk road network that BGS originally received. This allows users with no access to GIS to open the data in a spreadsheet or Microsoft Access database. Because of restrictions in field name lengths, these statistics are produced against shortened names. For a full explanation of these see Table A.9 (Appendix A).
Figure 4.11 – Excerpts from the GIS showing the landslide hazard assessment for two areas: (left) Glen Ogle and (right) Inverness. (OS Data © Crown Copyright. All rights reserved Scottish Government 100020540, 2008.)
Figure 4.12 – Indicative 1:2,000,000 debris flow hazard across Scotland. For more detail please refer to the digital data available separately. (Note that as reproduced herein the map is not to scale.)
Figure 4.13 – Indicative 1:2,000,000 debris flow hazard across Scotland showing the Trunk Road Network. For more detail please refer to the digital data available separately. (Note that as reproduced herein the map is not to scale.)
This has been a desk-based study, undertaken by BGS and the working group. Resources were not available for field survey. The results presented here from the first part of a landslides study. Because of the techniques employed in the processing of these data, there are some notable areas where misleading results could be inferred. These errors relate to factors that are not simple to encode into a system generated in the way described in the foregoing chapters. Figure 4.14 indicates an example of the errors that should be expected. In the Montrose basin, Raised Marine deposits of sand silt and clay, classified in our assessment as having moderate potential as source material for landslides have steeply sided gullies incised into them, which are highlighted by the slope model. This leads to a high score for landslide potential, especially when considering the land classification of bare ground and the presence of streams. In this instance, with extra knowledge that these deposits do not sit above a road, we can assume that they will not be involved in debris flow activity, however, the computer system does not know this fact, nor would it be straightforward on a national-scale to calculate this. It is at this stage that human intervention is required.
Figure 4.14 – Data anomalies in the Montrose Basin. (OS Data © Crown Copyright. All rights reserved Scottish Government 100020540, 2008.)
A methodology has been agreed by the Scottish Road Network Debris Flow Hazard Project Group to provide an outline assessment of debris hazard to selected sections of the Scottish Road Network.
The selection of factors to be included in the study was based upon relevance, usability and availability over the whole of Scotland.
Factors represented in the GIS methodology are the availability of debris material, water conditions, land cover, proximity of stream channels and slope angle. These have been combined using a GIS to estimate the hazard to the road network from debris flows.
The results have been tested against a number of areas where the degree and spatial extent of debris flow hazard are reasonably well known by members of the working party. The results of these tests have been used to ‘tune’ the methodology to better represent real world conditions.
The datasets provide a reasonable estimation of the hazard to the Scottish Road Network, as carried out at a national scale and are fit for purpose for use in helping to determine priority areas for the next phase of work in the Scottish Road Network Debris Flow Assessment.