Susceptibility models for pipe collapse in loess-derived soils in a temperate humid climate*

A. Logistic regression

Els Verachtert (1), Miet Van Den Eeckhaut (2,1), Jean Poesen (1), Gerard Govers (1) and Jozef Deckers (3)

(1) Division of Geography, Department of Earth and Environmental Sciences, K.U.Leuven, Belgium; (2) Land Management and Natural Hazards Unit Institute for Environment and Sustainability, Joint Research Centre (JRC) - European Commission, Ispra, Italy; (3) Division of Soil and Water Management, Department of Earth and Environmental Sciences, K.U.Leuven. E-mail address: els.verachtert@ees.kuleuven.be

B. Topographic thresholds (SA-relation)Soil piping (tunnel erosion) has been recognized as an important erosion process in collapsible loess-derived soils in a temperate humid climate, which can cause collapse of the topsoil and formation of discontinuous gullies. Still, information on the spatial patterns of collapsed pipes and regional models describing these patterns is limited.Therefore, this study aims at predicting pipe collapses and a better understanding the factors controlling their spatial distribution.

Problem statement

A dataset with parcels suffering from piping (560 collapsed pipes) and parcels without collapsed pipes was compiled through a detailed survey in a 236 km² study area in the Flemish Ardennes (Belgium; Fig. 2).

ConclusionsFor logistic regression, locations in NCP-parcels that were similar to locations of collapsed pipes were selected to allow for correct comparison. Pipe collapse is more likely to occur when a topographical threshold depending on both slope gradient and upslope area is exceeded in zones where subsurface flow occurs, due to topographical convergence and/or the presence of a clay-rich lithology.

The logistic regression model allowed identifying the most important factors controlling pipe collapse: i.e. S, distance-to-thalweg, A, wet and silty clay loam soil, west-facing slopes, (silty) clay lithology and concave plan curvature.

Methods

Logistic regression was carried out to find the best model describing the relationship between the presence/absence of a collapsed pipe and a set of independent explanatory variables (slope gradient, drainage area, distance-to-thalweg, curvature, aspect, soil type and lithology). Special attention was paid to the selection procedure of the grid cells without piping erosion. Part A

A second piping susceptibility map was made based on topographical thresholds of slope gradient (S) and upslope drainage area (A). Part B

Figure 1. Pictures of (a) collapsed pipes (n=3; diameter ca. 0.6 m; depth ca. 0.7 m) in a pasture (Kluisbergen, April 2009), (b) a collapsed pipe (Maarkedal, February 2008) and (c) a typical pipe cross-section (Kluisbergen, March 2010)

(a) (b) (c)

ReferencesHolden J. 2005. Controls of soil pipe frequency in upland blanket peat. J. Geophys. Res. Earth Surf. 110, F01002.Jones JAA. 1986. Some limitations to the a/s index for predicting basin-wide patterns of soil water drainage.Z.

Geomorph. N.F., Suppl.-Bd 60, 7-20.Jones JAA. 1997. Pipeflow contributing areas and runoff response. Hydrol. Process.11, 35-41. Verachtert E, Van Den Eeckhaut M, Poesen, J, Deckers J. 2010. Factors controlling the spatial distribution of soil piping

erosion on loess-derived soils: a case study from central Belgium. Geomorphology 118 (3-4), 339-348.* Verachtert E, Van Den Eeckhaut M, Poesen J, Govers G, Deckers J. 2011. Prediction of spatial patterns of collapsed

pipes in loess-derived soils in a temperate humid climate using logistic regression. Geomorphology, 10.1016/j.geomorph.2011.03.015

Fig. 3. Methodology used for modelling piping susceptibility using logistic regression.

Fig. 4. Classified map indicating susceptibility to pipe collapse produced with logistic regression.

Fig. 6. Classified map indicating susceptibility to pipe collapse produced with logistic regression.

Fig. 5. Topographical thresholds for pipe collapse used for classification of the SA derived piping susceptibility map (1 very low, 2 low, 3 moderate, 4 high, 5 very high).

Fig. 2. Location of the study area in Belgium and inventory of the collapsed pipes (n=560) projected on a map with indication of the clay-rich Aalbeke Member (Tertiary lithology).

This is in contrast to findings from the UK (Jones, 1986, 1997; Holden, 2005) where other factors (e.g. land management) are more important than surface topography.

In the Flemish Ardennes, the SA relation seems to represent the spatial distribution of other factors such as lithology as well.

The use of the SA-thresholds allowed producing a map with reasonable predictions of susceptibility to pipe collapse with a minimum of information…for this environment!