Modelling the location of shallow landslides and their effects on landscape dynamics in large watersheds: An application for Northern New Zealand

Abstract

In this study we propose a model to assess the location of shallow landslides and their impact on landscape development within a timeframe of years to decades. Processes that need to be incorporated in the model are reviewed then followed by the proposed modelling framework. The capabilities of the model are explored through an application for a forested 17 km² study catchment in Northern New Zealand for which digital elevation data are available with a grid resolution of 25 × 25 m. The model predicts the spatial pattern of landslide susceptibility within the simulated catchment and subsequently applies a spatial algorithm for the redistribution of failed material by effectively changing the corresponding digital elevation data after each timestep on the basis of a scenario of triggering rainfall events, relative landslide hazard and trajectories with runout criteria for failed slope material. The resulting model will form a landslide module within the dynamic landscape evolution model LAPSUS. The model forms a spatially explicit method to address the effects of shallow landslide erosion and sedimentation because digital elevation data are adapted between timesteps and on- and off-site effects over the years can be simulated in this way. By visualization of the modelling results in a GIS environment, the shifting pattern of upslope and downslope (in) stability, triggering of new landslides and the resulting slope retreat by soil material redistribution due to former mass movements can be simulated and assessed.

Introduction

Shallow landslides are one of the most common types of landslides, occurring frequently in steep, soil mantled landscapes in different climatic zones (e.g. Kirkby, 1987, Benda and Cundy, 1990, Selby, 1993). In many cases, failure and transport of slope material by landsliding is one of the principal processes of soil redistribution and hillslope development in landscapes.

Landslides have traditionally been regarded as key indicators of forest disturbance, particularly in association with logging activities (Douglas, 1999, Douglas et al., 1999, Montgomery et al., 2000, Fannin and Wise, 2001), land use and climatic change (Van Beek, 2002, Vanacker et al., 2003) or as response to human imposed changes such as road building (Brand and Hudson, 1982, Larsen and Torres-Sanchez, 1998). Over the past decades most model studies dealing with landslides have mainly focussed on landslide hazards and terrain stability mapping for regional and urban planning (Van Westen, 1993, Guzzetti et al., 1999) and the impact of landslides on basin sediment yield (Burton and Bathurst, 1998). Also a number of recent landslide studies have been concerned with identifying rainfall thresholds required to trigger landslide events (Crozier, 1999) or magnitude and frequency of landsliding in a specific catchment (Reid and Page, 2003). However, as a mass movement process, rainfall-triggered landslides have been underestimated as contributors to slope development and denudation in the past (Dykes, 2002). Geomorphological investigations in steeplands show evidence that shallow landslides play a key role in the maintenance of characteristic slope forms and probably also make significant contributions to ecological dynamics (Dykes and Thornes, 1996, Claessens et al., 2006).

At present relatively few attempts have been made to model the effects of shallow landslides on landscape development. Modelling the relevant processes in a dynamic way requires a methodology aggregating several mechanisms acting at different spatial and temporal scale levels but remaining at least spatially explicit over a timeframe concerning redistribution of soil material. Kirkby (1987) identifies two major difficulties in the context of models of longer term landscape development: work on rapid mass movements has concentrated on stability analysis, so that forecasting of destinations for slide debris is very inexact, even for an individual slide. Modelling of slope profile evolution is severely restricted by a lack of studies on the factors controlling travel distances of mobilized material, which are crucial to the development of an overall mass budget in a forecasting context. The second major problem lies in aggregating from the individual slide to the assemblage of slides over a long period wherefore detailed meteorological records are generally not available. The change of temporal scale requires models that are built from somewhat different premises than those used for short-term stability analysis, although obviously they must be compatible.

The practical significance of shallow landsliding has motivated many approaches of mapping and predicting potential landslide initiation (for a review see Montgomery and Dietrich, 1994). A recent approach, which is proven to be very practical, is the use of digital elevation models and simple coupled hydrological and slope stability models (Dietrich et al., 1992, Dietrich et al., 1993, Montgomery and Dietrich, 1994, Wu and Sidle, 1995, Borga et al., 1998, Pack et al., 2001, Borga et al., 2002). Availability of GIS technology permits then to resolve and display spatial patterns of landslide susceptibility at the same scale and resolution as the digital terrain model.

In general landslides triggered on forested slopes release such energy and mass that a debris flow nearly always develops. This flow erodes the unstable material in its path and continues to move downslope until the gradient falls below that needed to maintain flow (Burton and Bathurst, 1998). To study the role of rainfall-induced landslides within the hierarchy of hillslope erosion processes on the longer term, it is therefore important not only to know the spatial distribution of possible landslide initiation sites but also to characterize erosion and deposition patterns caused by slope failure. Removal of failed landslide material can potentially increase the local slope by taking away initial support and may trigger subsequent upslope failure. Additional downslope erosion and failure may occur along the debris flow erosional pathway and by loading and steepening downslope material by debris flow deposition. Once a debris flow emerges, the problem of determining its path becomes complicated by the ability of the flow to erode, to spread, to plug, and to alter its direction. The rate of volume transport of a debris flow and its change with time, viscosity and hillslope morphology are some factors important for debris flow erosion and deposition.

Using data from individual events, two-dimensional mathematical models for flows on fans have been calibrated to determine flow depths, velocities, impact forces and areas of deposition (Mizuyama and Ishikawa, 1990, Mizuyama et al., 1987, O'Brien and Fullerton, 1990, Takahashi, 1991). These methods require a large amount of data along with numerous assumptions about the characteristics of the debris flow. Formulations from laws of mass and momentum conservation are complex and more appropriate for examination of the detailed behaviour of an individual debris flow for which the flow material composition and hillslope characteristics are well specified (Bathurst et al., 1997). For our goal a simpler, rule-based and spatially explicit approach is used that can easily be applied to multiple landslides occurring throughout a catchment over periods of time ranging from single rainstorms to several years and for which the data specification is more general.