SPATIALLY DISTRIBUTED PROBABILISTIC LANDSLIDE HAZARD MODELING AS A FIRST STEP TOWARDS QUANTITATIVE RISK ASSESSMENT

A first-order, second-moment (FOSM) approximation can be used to quantify the influence of parameter uncertainty and variability on the calculated stability of infinite slopes. The FOSM method described in this presentation allows the analyst to select input probability distributions, takes into account the influence of all variables in the factor of safety equation (including the effects of vegetation if desired), and is not restricted to a particular output probability distribution. Uncertainties in slope angles calculated from digital elevation model (DEM) data are calculated using a separate FOSM approximation based on the RMS elevation error of the DEM. Perhaps most importantly, the FOSM method is particularly well suited to spatially distributed or raster GIS applications because it is non-iterative and computationally straightforward. This simplicity makes the FOSM method a potentially valuable tool for watershed analysis, land use planning, transportation corridor routing, and regional hazard analysis. Results can be presented as 1) the factor of safety mean and variance, 2) the probability of sliding or stability, 3) a distribution-independent slope reliability index, and 4) binary hazard zonation maps with boundaries identified using digital image analysis of the model results. Calculation of the probability of sliding does require a priori knowledge of the underlying probability distribution. Extensive Monte Carlo simulations have shown that the factor of safety distribution is generally most faithfully replicated using a log-normal distribution. Conditional probabilities for pore water pressure or seismic acceleration thresholds can also be calculated and used as the basis for time-dependent estimates of the probability of sliding. Application of the method is illustrated using a test area near Wheeling, West Virginia for which an existing landslide inventory map was available for comparison.