Probabilistic analysis of suction in homogeneously vegetated soils

Abstract

It is unlikely that natural vegetation growth induced soil suction can be predicted deterministically. Probabilistic approach may be an ideal tool to analyse effects of evapotranspiration (ET) induced suction by vegetation, which are commonly found on slopes. Most of the previous deterministic studies quantifying suction are carried out on natural slopes, which are heterogeneous in nature. The main objective of this study is to conduct probabilistic analysis of measured soil suction in homogeneously vegetated slopes. In the present work, a probabilistic model is developed for the measured values of suction, since the measurement process itself is very uncertain and is subjected to a wide range of climatic fluctuations owing to a wide monitoring interval. Expected values of soil suction are found to be significantly higher in treed slope as compared to grassed and bare slope. In contrast to this, the coefficient of variation of soil suction is found to be similar for all slopes irrespective of vegetation types. The probability distribution is found to be dynamically changing with time and also significantly different between bare slope (Weibull distribution) and the vegetated slopes (mostly Normal distribution). Using the estimated probability distributions of suction, the probability of failure is estimated using a simple limit state function using 1000 trials of Monte-Carlo simulation. It is observed that the distribution of suction alters significantly for bare soils during wetting events as compared to the grassed and treed soils. The probability of failure of the bare slopes increases significantly under wetting events.