Investigation on the Factors Controlling the Response of Mountain Rivers to Extreme Flood Event (Case Study: Upstream Ilam Dam)

Introduction
Large severe floods can have enormous influence on the fluvial system in comparison with the floods with lower magnitude and more frequencies. This work addresses the geomorphic response of mountainous rivers to extreme floods to explore the relationships between morphological changes and controlling factors. In October 2015, following the occurrence of a sudden extreme rainfall, a large and devastating flood occurred in Ilam province. The flood caused major changes in the morphology of Ilam's rivers. The rate of channel expansion is various in different sections of the studied rivers. Thus, we can examine the influential and controlling factors that led to diversity of river behavior. The hypothesis of this research is that explanation of geomorphic effects requires models that include other variables, e.g., lateral confinement, degree of sediment, besides hydraulic related variables (cross-sectional or unit stream power). The main purpose of this research is to explore the relationship between channel widening and arrange of controlling factors. We have addressed channel width (i.e: pre- or post-flood width) to calculate unit stream power in order to have a better explanation of channel response? Since few studies have been done in this field, this research was conducted with the aim of investigating the factors controlling the response of Mountain Rivers to extreme flood events in upstream of Ilam dam.
Materials and methods
The research has examined three tributaries of the Konjancham River (upstream of Ilam dam) whose catchments were affected by an extreme flood on 7th October, 2015. An integrated approach was taken to study this flood, including (i) Analysis of channel width changes by comparing aerial photographs before and after the flood, (ii) Estimation of peak discharges in studied reaches, and (iii) Determining the degree of sedimentation in studied reaches. Delineation of spatial units was carried out according to the approach proposed by Rinaldi et al. (2013), which is a modification of the approach by Brierley and Fryirs (2005). According to the approach, stream sectors were defined as macro reaches having similar characteristics in terms of lateral confinement, while the reaches are homogeneous in terms of channel morphology (channel pattern, width, and slope) and hydrology. We have used the reach scale (reach length was commonly from 200 to 1300 m) for an overall assessment of magnitude of channel changes and for a preliminary investigation of controlling factors. The dominant process observed in the study reaches was channel widening, which was analyzed by comparing aerial photographs taken before and after the flood. To assess the changes in channel width, channel banks, and islands, these features were digitized on pre- and post-flood orthophotos. The channel width was calculated by dividing channel area by the length of the reach, and changes in channel width were expressed as a width ratio (ratio of channel width after the flood to channel width before that flood). The estimation of peak discharges has been used to calculate cross-sectional stream power and unit stream power. The last part of the methodological section deals with statistical analysis carried out to explain channel response to the flood event by exploring the relationships between the changes in channel width and controlling factors.
Results and discussion
The relationships between the degree of channel widening and possible controlling factors were explored using multiple regression analysis. The analysis was carried out for the widening (width ratio) at reach scale. The entire data set includes 38 reaches. We analyzed seven controlling variables including confinement index, percentage of reach length with artificial structures, degree of sedimentation, channel slope, cross-sectional stream power, and unit stream power using pre-flood and post-flood channel width. Each regression model included only three to four variables. Each model included only one of the variables expressing potential or flood flow energy, e.g., channel slope, cross-sectional stream power, unit stream power. All four multiple regression models turned out to be significant (p< 0.001) and gave high coefficients of multiple determinations. The values of R²and adjusted R² are ranged between 0.73 and 0.8 and between 0.69 and 0.77, respectively. The best model embraced unit stream power calculated based on pre flood channel width and confinement index as explanatory variables.
Conclusion
The results confirmed the main hypothesis of this work that hydraulic variables alone are not sufficient to explain channel response to an extreme flood event. The inclusion of other factors, specifically lateral confinement, degree of sedimentation, and percentage of reach length with artificial structures can lead to satisfactory models explaining the observed variability in the degree of channel widening. These results suggest that the widening process is essentially controlled by two factors: flood power and valley confinement. Flood duration exceeding a critical threshold was not included in our analysis, but it is a variable that very likely would increase the robustness of regression models in these reaches. The analysis carried out in the three subcatchments of the Konjancham River basin showed that unit stream power calculated based on pre-flood channel width has stronger relations with channel widening in comparison with that based on post-flood channel width and cross-sectional stream power. Because peak discharge was used for stream power calculation, we are aware that neither pre-flood nor post-flood channel width is actually appropriate for the estimation of unit stream power, as the most appropriate would be the (unknown) width at the flood-peak time. The pre-flood width has stronger relations with the degree of channel widening (width ratio). This could suggest the width changes occurred after the flood peak.