Wildfire disturbance and shallow landsliding in coastal British Columbia over millennial time scales: A numerical modelling study

An algorithm for wildfire occurrence is introduced for incorporation into a numerical model of drainage basin evolution. Within the model, fire return intervals are determined using a stochastic rule set and fire sizes are assigned according to a pareto distribution. A Weibull distribution was fit to millennial-scale lake charcoal data for calibration of fire return intervals in this study, while data for the Olympic Peninsula were used to determine the fire size distribution. Loss of tree root strength resulting from stand-replacing wildfires has been found to be comparable to loss after timber harvesting. Hence, data for increased shallow landslide activity associated with logging were used to calibrate transport equations for years following wildfire incidence. A weathering relation is incorporated in the model, which drives the supply of sediment. When sediment supply is depleted in the model, the shallow landsliding module shuts off until new material is available for transport. Time since last fire analysis indicates that the Weibull parameters defined for the Clayoquot charcoal data produce results that are weighted too heavily towards recent fires. Sensitivity analysis was undertaken, suggesting alternative values may be more reasonable for further model runs. The introduction of a weighted probability factor for aspect, whereby south-facing slopes have a higher likelihood of fire initiation, was not found to significantly affect patterns of fire age in the landscape. Patterns of net topographic change after 6000 years of model run time indicate that greatest net erosion is found on steep, upper slopes, with the highest deposition occurring in hillslope hollows and low-order channels. Depositional grid cells in interfluves show a variable response to the inclusion of wildfire in the model. Net erosion occurs over a relatively low proportion of hillslope area, with wildfire model runs showing a strong erosional signal in these locations. For stream grid cells, the additional deposition due to the inclusion of wildfire in model runs is about 9–10% for 1st to 3rd order streams. Overall, deposition rates increase with decreasing stream order, reflecting greater hillslope/channel coupling in low-order streams.