As the exposure to rockfall increases to both infrastructure and the population of New Zealand. The need for reliable and accurate modelling tools becomes necessary in order to predict, protect and prevent against rockfall.

A rockfall events boulder distributions are largely governed by the terrain properties (slope topography and substrate) and the boulders characteristics (shape, size and density). It is the interactions between these two physical properties that control the boulder’s behaviour as it travels down the slope. Numerical rockfall models such as RAMMS Rockfall account for these physical interactions with a complex algorithm. The purpose of this thesis is to assess the individual input parameters sensitivities to change within this program. Once these were understood they could then be ranked in order of influence on the model.

The greater understanding of these sensitivities would allow the practitioner to confidently use the model in areas that the model has yet to be calibrated for, or in areas where no previous rockfall has been recorded. The location used for the purpose of this analysis was the Heathcote Valley situated within the Port Hills of Christchurch, New Zealand.

After approximately 1000 hours of modelling in this area, it was found that the RAMMS::Rockfall software was most sensitive to changes in surface material, followed by that of boulder characteristics (shape, size and density) and vegetation variations respectively.

These findings were then used to create suggested modelling workflow that could be used to help streamline the modelling process. The workflow makes suggestions for both Greenfield and active rockfall regions and what the primary focus should be for each scenario.