Go to Paper
Return to GeoComputation 99 Index

The Validity of Using a Simplified Distributed Hydrological Model for Estimation of Landslide Probability Under a Climate Change Scenario

GRIFFITHS, James A.(Jim.Griffiths@kcl.ac.uk), COLLISON, A.J.C. and WADE, S.W., King's College London, Geography Department, Strand, WC2R2LS, U.K.

Key Words: hydrological modelling, landslides, climate change, validation

In order to assess the implications of envisaged future climate change on the stability of shallow-translational landslides, a simple distributed hydrological model was designed for use within humid-temperate and sub-humid environments. In order to simulate the hydraulic response of a landslide over relatively long timescales, however, it was necessary to simplify both its structure and related hydrological processes. The vertical soil profile of the feature was represented by just three layers (root zone, colluvium, and underlying impermeable layer), while the landslide surface was divided into 10 by 10m cells using a digital elevation model (DEM).

Though not strictly physically-based, processes of infiltration, unsaturated and saturated flow, and throughflow, all were represented in some way within the model. Vertical water movement between soil layers was simulated using a tank-model approach; i.e., downward moisture movement is gravity driven at a rate determined by soil conductivity and the capacity of the underlying layer to accept moisture. Lateral movement occurs in the direction of adjacent cells with the lowest moisture content at a rate determined by a derivative of Darcy's Law. For each daily timestep then, moisture movement was modelled vertically between layers but not between cells; and horizontally between cells, but not between layers.

In the first instance the suitability of the model under present day conditions was addressed. Using the dynamic modelling-based GIS package "PCRaster," timeseries output of predicted local water table heights could be obtained for any cell within the DEM, and then compared with observations made within the field. Model output was optimised by using different interpolation techniques applied to sampled soil and vegetation properties.

The suitability of the model for longer term prediction was then assessed with respect to stochastically generated daily rainfall and temperature data, derived from downscaled mean monthly GCM predictions. Initial results assessed through water-table ascribed thresholds of instability indicate that though climate change increases the probability of slope instability through greater precipitation, this effect may be offset by greater levels of evapotranspiration. Modelled using a temperature dependent empirical relationship, initial sensitivity analysis of evapotranspiration concurs with previous research that variation in parameters of vegetation height and canopy conductivity are of greatest significance to ET prediction.

To conclude, it is recognised that though relevant hydrological processes are numerically modelled to an acceptable standard within the model, the results obtained describe changes to slope stability in relation only to rainfall and temperature. In order to account for the full effects of climate change, a better knowledge of the potential variation in relative humidity; net radiation; cloud cover; and regional patterns of rainfall intensity, duration and seasonality, are needed. As much of this information has yet to be quantitatively determined, the described model may be regarded as a best first estimate given the data currently available.