Difference between revisions of "OR/14/051 Erosion and deposition due to road construction"

From Earthwise
Jump to navigation Jump to search
[unchecked revision][checked revision]
m (1 revision imported)
(No difference)

Latest revision as of 17:08, 15 March 2016

Whitbread, K. 2014. The geomorphic impact of road construction: a case study of the A9 in Scotland. British Geological Survey Internal Report, OR/14/051.

Estimates of the net transfer of material resulting from road construction can be made using the mean surface change values derive by direct comparison of the borehole ground levels and the DTM described above. Alternatively mathematical algorithms can be used to reconstruct the pre- road ground surface from the DTM and volumes of material lost or gained can then be derived by comparison of the ‘before’ and ‘after’ DTMs. These two methods are compared in the following sections

Method 1: Direct borehole — DTM surface analysis[edit]

Estimates of the volume of material excavated from cuttings and piled in embankments were calculated using the average depths of MGR and WGR from Table 4, combined with the mapped areas of each unit. A triangular cross-section was assumed for all the artificial ground areas.

The results demonstrate that there has been a net loss of material from along the road route in both the test areas. The net volume change in the southern area is two thirds that of the northern area despite the greater section length. This finding reflects a greater proportion of worked ground and a lower average thickness of made ground in the northern area.

Table 5    Details of the surface area and volumes associated with areas or artificial ground along the A9. Volumes were derived for a triangular cross-section using the mean thickness of made and worked ground from Table 4
Test area Proportion of road length Artificial Ground Type Area (m2) A/Road Length (m2/m) Volume (m3) V/Road Length (m3/m) Net volume change (m)
North Daviot-Moy (12.5 km) 0.36 MGR 89715 8.3 134573 12.4 -604895
0.47 WGR 172370 15.9 739468 75.4
South Dunkeld (30 km) 0.25 MGR 239763 8.0 808001 27.0 -370585
0.31 WGR 279286 9.3 1178586 39.4

Method 2: DTM reconstruction[edit]

Reconstructed DTMs were derived for the test areas to model the ground surface prior to road construction. Areas of mapped artificial ground along the road route were removed from the DTM, and the remaining DTM was re-interpolated using three different methods for comparison: inverse distance weighting, regularised spline and triangulation (see DEM re-interpolation). The inverse distance weighting interpolation was repeated using power factors of 1.5, 2 and 3, reflecting an increasing degree of influence given to points near to the cell.

Comparison of reconstructed DTMs with pre-road borehole ground levels[edit]

The surface elevation values in each of the reconstructed DTMs was compared with ground surface elevations from the pre-road (c.1970’s) borehole records to assess how well they estimate the original ground surface level. The elevation differences between the borehole ground level and the reconstructed DTM were compared separately for areas of MGR and WGR.

For the northern area (Daviot-Moy), the IDW re-interpolation give the closest fit to the actual borehole ground levels, with the mean difference in elevation between the borehole start height and the re-interpolated DTM not significantly different from zero for either the MGR or WGR (p>0.05; Table 6). Both the spline and triangulation methods yield DTM elevations that are not significantly different from the borehole ground levels in areas of MGR, but are significantly different in areas of WGR; mean differences in WGR areas are 2.17 and 1.15 respectively, indicating that these DTMs tend to underestimate the actual ground level.

In the Southern test area (Dunkeld), all of the re-interpolation methods yield DTMs that are significantly different to the actual ground level. Mean differences between the borehole start height and the DTM are on average 1.31–1.86 m higher than the borehole ground level in areas of MGR, and 1.47–2.83 m below the borehole ground level in areas of WGR (Table 7). These offsets may arise if the delineation of areas of artificial ground does not quite cover the full extent of the embankment or cutting. Buffering the areas of made and worked ground prior to extracting them from the original DTM may help to minimise these differences in future.

Table 6    Statistics for the difference between the pre-road borehole ground levels
(boreholes dated 1970–1977) and the re-interpolated DTMs for the northern test area (Daviot-Moy).
The different types of DTM interpolation are discussed in the text
Northern test area

IDW 1.5





Mean 0.28 1.13 0.27 1.13 0.24 1.13 -0.84 2.17 -0.69 1.15
Standard Error 0.59 0.72 0.59 0.72 0.59 0.72 0.55 0.41 0.34 0.33
Median -0.08 2.31 -0.11 2.28 -0.16 2.22 -1.42 2.25 -0.62 1.01
Standard Deviation 2.03 4.09 2.04 4.08 2.04 4.07 1.90 2.35 1.18 1.87
Kurtosis 5.97 0.83 6.02 0.84 6.13 0.85 4.74 -0.57 0.71 0.69
Skewness 2.20 -0.99 2.21 -0.99 2.24 -0.98 1.98 0.11 -0.50 0.15
Max 6.0 7.8 6.0 7.8 6.0 7.8 4.3 6.6 1.2 5.3
Min -1.6 -9.2 -1.6 -9.1 -1.7 -9.1 -2.5 -1.9 -3.2 -3.5
Count 12 32 12 32 12 32 12 32 12 32
2-tailed t-test P value 0.64 0.13 0.66 0.13 0.70 0.13 0.15 <0.01 0.07 <0.01
Table 7    Statistics for the difference between the pre-road borehole ground levels
(boreholes dated 1970–1977) and the re-interpolated DTMs for the southern test area (Dunkeld)
Southern test area

IDW 1.5





Mean -1.31 1.47 -1.31 1.47 -1.33 1.46 -1.86 2.83 -1.48 1.91
Standard Error 0.26 0.54 0.26 0.54 0.26 0.54 0.29 0.49 0.23 0.39
Median -0.69 0.99 -0.69 0.99 -0.74 0.98 -0.98 2.09 -1.00 1.12
Standard Deviation 2.65 4.21 2.65 4.21 2.66 4.21 2.92 3.79 2.94 3.06
Kurtosis 1.69 2.01 1.68 2.01 1.65 2.00 2.06 2.15 1.22 2.06
Skewness -0.80 0.20 -0.80 0.20 -0.80 0.21 -0.95 1.10 -0.78 0.97
Max 5.5 13.4 5.5 13.5 5.5 13.5 5.3 16.8 3.9 12.5
Min -10.7 -11.9 -10.7 -11.9 -10.7 -11.8 -13.4 -5.5 -10.6 -4.8
Count 105 60 105 60 105 60 105 60 105 60
2-tailed t-test P value <0.01 0.01 <0.01 0.01 <0.01 0.01 <0.01 <0.01 <0.01 <0.01

In both test areas, the IDW approach gives the lowest mean difference, which appears to indicate that it more closely reflects the actual ground surface than the other re-interpolation methods. However, the IDW method also has higher standard deviations and a larger range in the data values, with differences between the actual and predicted ground levels of ±8–9 m in the northern test area and ±12–14 m in the southern area. By contrast, differences in actual and predicted ground levels for the spline interpolation are up to ±7 m and ±17 m in the north and south respectively and for the triangulation are ±5 m in the north and up to ±13 m in the south but with a lower range in values (Tables 6 and 7).

The volumes of cuttings and embankments[edit]

Volumes of MGR and WGR were derived for each test area by summing the volumes of cuttings and embankments as defined by the Cut-Fill tool in ARCGIS. This tool compares the pre-road construction ground surface (re-interpolated DTM) and the post-road construction surface (original DTM) to define regions of net gain or net loss of material. Values for the triangulation and IDW 2 DTMs only are shown; the IDW 1.5 and 3 interpolations are similar to the IDW 2 interpolation, and the spline interpolation results in the poorest correspondence with the pre-road borehole ground levels so the volumes derived are likely to be the least accurate (Tables 6 and 7).

The results are given in Table 8 and are compared with values derived from the average vertical ground level change estimated directly from comparison of the pre-road borehole records with the original DTM (Table 5).

In both test areas, the volume of WGR exceeds that of MGR indicating that there has been a net loss of material during road construction. This reflects the fact that areas of worked ground are more extensive than made ground in both areas, and that cuttings tend to be deeper than embankments (Table 5).

The re-interpolated DTMs tend to underestimate the depth of cuttings by 1–2 m and the heights of embankments by up to 1.5 m (Tables 6 and 7). Moreover, the underestimates for WGR are greater than those for MGR. This observation indicates that any future improvements in DTM re-interpolation methods for reconstructing the pre-road ground surface may result in even greater differences in the volumes of MGR and WGR.

The volumes of material estimated from direct comparison of the borehole ground levels with the DTM are lower than those derived using the re-interpolated DTMs (Table 8), and yield lower net differences. This may be because the direct method is based on the derivation of an average surface lowering value that depends on the number and distribution of borehole records available. It is possible that the average depth of cuttings and height of embankments is underestimated through this method, particularly in the northern area, due to the limited number of borehole records (Table 4) and lack of sampling in the deeper parts of many cuttings.

Table 8    Calculated volumes of MGR and WGR for the test areas derived from comparison of re-interpolated DTMs (IDW2 and triangulated DTMs only) with the original DTM and from direct assessment of the difference between the pre-road boreholes and the DTM (Boreholes-DTM). Note that the aggregate base and tarmac comprising the road surface constitutes MGR that will be present in areas of cuttings but is not accounted for in these figures. The volume of MGR along the road route is thus likely to be slightly greater than shown

Northern area (length 12.5 km)

Southern area (length 30 km)

Length and area Prop. of road length





Area (m2)







                                         Derivation method
Volume (m3) Triangulation





















Average vertical ground surface change due to construction (m) Triangulation





















Erosion flux (m3/m/yr)* Triangulation









* assuming road construction took 2 years

If the net loss of material from the road sections were distributed evenly along the road length, it would amount to surface lowering of 2 to 2.5 m in the northern area, and 0.4–0.7 m in the southern area. The higher value of vertical lowering in the northern area reflects extensive cuttings, and the relatively thin nature of the MGR along this road section.