Discussion

Vegetation change and climate variability were considered as main factors affecting annual streamflow in this study, and the evaluation of such impacts were implemented at mean annual scale. At seasonal or monthly time scale, the climate variability impact on streamflow may be different since the streamflow would be impacted by the seasonality of rainfall which adds an additional layer of complexity. Another time scale issue which should be focused on is the period selected for bushfire impact assessment. This study mainly focuses on an immediate post bushfire period. When the bushfire period considered is extended the increased impact can be subdued due to regrowth and runoff will decrease when there is more vegetation.
In this study, the estimation of streamflow change due to climate variability was calculated by a sensitivity-based method based on the Budyko framework (Budyko, 1958). This study not only tested the method of Zhang et al., (2001), but also other Budyko curves such as Fu curves (Fu, 1981) and both methods yield similar results. These tests further indicate that Budyko curves are robust for estimating mean annual streamflow change induced by climate variability. Generally speaking, the approach can yield more reliable results if multiple linear regression considered PET as well as P as descriptors to predict the mean annual streamflow during the testing period. For this reason, the multilinear regression considering P and PET was tested as well. Overall, the fitting coefficient of determination (R2 ) of single linear regression (0.74) and multiple linear regression (0.76) are very similar. In addition, the proportion of bushfire and climate variability impact on streamflow didn’t show a dramatic change when the single regression was replaced by multiple regression. As a result, we conclude that the linear regression relationships between annual rainfall and annual streamflow are satisfactory for estimating vegetation change induced impacts on streamflow.
As shown in Figure 5, the relationship can be described by a linear regression model. The fitting coefficient of determination (R2 ) is 0.69 which is satisfactory. it is possible to further improve quantitation of the relationship between burnt percentage and percentage of mean annual streamflow increase due to bushfire, which is constructed to specify how bushfire extent influences the bushfire impact on streamflow.
This study uses two independent methods to estimate the effects of vegetation change caused by bushfire and climate variability on annual streamflow. The independent evaluation of the two impacts show that the time-trend analysis and sensitivity-based methods are accurate for our research purpose. Further, to meet the assumption that climate variability and bushfire are the main dominant factors to affect the streamflow, efforts has been made when choosing these target catchments (i.e. adequate data record and impacted by bushfire to certain extent). Our study clearly indicates the burnt percentage shows a good positive relationship with percentage of mean annual streamflow change due to bushfire, which suggests higher and more serious burnt percentage contributes a greater increase in streamflow.