Post-fire changes in sediment rating curves in a wet Eucalyptus forest in SE Australia

Summary: Empirical power law sediment rating curves of the form C = aQb (where C is the sediment concentration, Q is the discharge rate, and the coefficient a and exponent b are fitted parameters), or the alternative form given by log(Qs )=log(a)+(b +1)log(Q) (where Qs = CQ is the sediment delivery rate, and the natural logarithm is used) are widely used for characterising sediment transport across a broad range of spatial and temporal scales. Fire frequently has a large effect on erosion rates and sediment delivery. We investigate the temporal changes in the coefficient a and exponent b for 3years following a wildfire between February 2003 and April 2006, for two South Eastern Australian Eucalyptus forested catchments (136 and 244ha) with mean annual rainfall of 1900mm. Storm-event integrated sediment loads and discharges were calculated for each of 596 storm events using stage-discharge control structures and in situ turbidity measurements at 15min intervals. Measurements were converted to sediment concentrations using regression relationships developed from storm activated water auto-samplers. The analysis identified: (i) strong negative linear relationships between the rating coefficient log(a), and the rating exponent b, reflecting sediment rating curves that “pivot” around a common fulcrum point in log–log space, (ii) a systematic shift in this linear relationship between log(a) and b as a function of time since fire, (iii) maximum values of b of ca. 2.5 (i.e. maximum non-linearity in the relationship between discharge and sediment delivery, and therefore maximum sensitivity to high peak-discharge events) immediately following the fire, which decline rapidly and monotonically by a factor of 10 to ca. 0.25 in the first 8months, attributed to a dominance of hillslope erosion processes and declining rill and interill erodibility, (iv) irregular patterns in the value of b during the vegetation recovery period, probably reflecting a shift from hillslope dominated erosion processes to a greater contribution from channel processes (8–24months after the fire) and (v) annual cycles in the value of b in the recovered state (24–38months after the fire) ranging from a minimum of ca. 0.5 in the dry season to a maximum of ca. 1.5 in the wet season. The physical cause of these cycles could not be isolated in this study. The results provide a robust quantitative perspective on the magnitude and temporal variability of the sensitivity of catchments recovering from wildfire. [Copyright &y& Elsevier]