Probabilistic model calibration for soil slope under rainfall: effects of measurement duration and frequency in field monitoring

Abstract

Field measured data reflect real response of soil slopes under rainfall infiltration. Based on field measured pore-water pressure histories, prediction models for slope design can be calibrated. In this study, a probabilistic approach based on the Bayesian theory is adopted to calibrate a slope-stability model using time-varying response data and to justify appropriate monitoring duration and data recording interval for field monitoring programmes. As an illustrative case study, field measured pore-water pressure responses in a natural terrain slope in Hong Kong are used to calibrate a prediction model using the Markov chain Monte Carlo simulation method. It is found that with the increase of measurement duration for model calibration, the uncertainties of soil parameters can be reduced. The duration of the calibrated data should be sufficiently long so as not to underestimate the model error. With regard to the effect of measurement interval, both the uncertainties of input parameters and the model error are increased with the increase of data recording interval. The correlations among the soil parameters are subject to greater uncertainties and the prediction accuracy of the calibrated model is reduced with a larger recording interval.