The effect of fire on hydrological response and the subsequent effect on streamflow

[摘要] ENGLISH ABSTRACT: Fire is a destructive force that destroys thousands of hectares of both urban and rural vegetation in South Africa every year. Fire not only destroys vegetation, but also affects the infiltration and percolation capacity of soil. With fire destroying vegetation and changing soil characteristics, it is therefore within reason to assume that the hydrological response of a catchment could be affected by fire. The main aim of this research was to investigate the hydrological changes caused by fire on a catchment scale. On the 9th of March 2015, a wildfire started in Jonkershoek nature reserve, which destroyed indigenous fynbos vegetation and afforested areas. Within the nature reserve, there are multiple rainfall and runoff stations, which provided a means of measuring possible hydrological changes caused by the fire event on different catchments. There were four catchments used for the research, one main catchment (fynbos area) and three sub-catchments (afforested areas). Fifty-six percent of the main catchment burned, while two sub-catchments were completely burned and the other was primarily unaffected by the fire. Themain catchment's hydrological response due to rainfall events were analysed by comparing the hydrographs of comparable pre- and post-fire rainfall events. Eighteen comparable events were used for the analysis. The mean runoff volume increased by 6.8% and mean peak flow by 50%, after the fire. The Wilcoxon signed ranked test confirmed that the increase in volume was significant (p < 0.05), however the increase in peak flow was not significant (p = 0.053).Since all of the sub-catchments were similar in size and were close to the same rainfall station, there were two affected catchments and one control catchment identified for further comparison. Before the fire, the average daily streamflow over the control and burned sub-catchments was similar, however after the fire the average daily streamflow of the burned sub-catchments in comparison to the control catchment, had increased by 45% and 50% respectively. The two-way mixed Analyses of Variance (ANOVA) confirmed that the differences were statistically significant (p < 0.01). The mean volume runoff, after the fire for individual events increased with 72.4% and 54.7% for the two burned -sub-catchments respectively, in comparison to the control sub-catchment. The mean peak flows increased with 116.7% and 183.3% in the burned-sub-catchments respectively, in comparison to the control sub-catchment. The paired catchment method was used to test whether the results were significant by using multiple linear regression. The runoff volume before the fire was statistically significant (p < 0.05) when comparing the control and affected sub-catchments. With the addition of an interaction term (F) for indicating the effect of fire, the predictability of the model did not increase, which indicates that fire was not a significant term. By using the same approach on the peak flows, it was found that with the addition of an interaction term (F), fire did increase the predictability of the model.

[发布机构] Stellenbosch University