The main motivation of this study is to develop a method and an elaborate index that are
capable of identifying effective changes in flood runoff processes over time. The effective
change proxy selected is the stream discharge time series as it is already an integration of
spatio-temporal variations of processes within a catchment. As a result of the changes in
the boundary conditions of a hydrological system, such as the characteristics of climate,
river drainage, and catchments land surface, change in the runoff dynamics is expected.
Urbanization, climate change, and different water management practices could have a
substantial impact on the characteristics of floods, such that the causative mechanism
related process typology of a flood could become unusual from how it is used to be. In
addition, different processes can also cascade to cause unprecedented flood dynamics. In
this study, the aforementioned effective change discusses on visible characteristics change
of the catchment outlet runoff that can be triggered by any modified boundary conditions.
The common practice of analyzing the occurrence of an extreme or rare flood is calculating
its return period over a long flood time series. However, this return period is often
based upon the index of the peak discharge or maximum water level and therefore is
not comprehensive enough to describe the rarity of the process typology. In order to
extend the analysis to include more comprehensive runoff characteristics, we introduce the
characterization of runoff dynamics that considers the shape of the hydrograph, portrayed
as a phase space trajectory. To distinguish the event hydrograph further, the approach
is taken to the next level by considering the non-linear and non-monotonic relationship
between magnitudes of different time points by using Taken’s time delay embedding
theorem. This takes the implicit temporal succession of discharge values into consideration,
such that the impact of the initial conditions on the flood events is considered as a
characteristics vector in the multi-dimensional time delay phase space. This mentioned
temporal succession of discharge values is herein called a temporal cascade. This study
argues that the proposed characterization of runoff dynamics which includes the continuous
shape of hydrograph shape and the temporal cascade is more elaborate and is a better
proxy to detect rare flood processes.
As the first application attempt in Hydrology, Recurrence Plot (RP) and Recurrence
Quantification Analysis (RQA) are used to visualize and determine the overall similarity
between flood runoff dynamics i.e. through a quantitative index as to whether or not a
certain dynamic is rare among historical observations. Rooted from the field of theoretical
physics, these tools have gained considerable popularity over the past decades in several scientific disciplines, from economics, physiology, neuroscience, paleoclimatology, astrophysics
to engineering, especially for non-linear time series analysis and studying the behavior of
a complex system. This study includes application examples dedicated to hydrologists
to better understand the concept of characterizing runoff dynamics and the usefulness of
the additional hydrograph similarity index. This study also extends the current state of
RP and RQA with improved robustness towards artifacts and the influence of noise and
is adapted to the observation runoff series. This includes the practical method to safely
parameterize the time delay embedding and RP, and an extended version of the RP and
RQA to reduce the influence of noise, in order to prevent further artifacts. The examples
utilize the runoff time series from the Dresden gauging station of the Elbe river catchment
located in East Germany from the period of 1901 to 2010. In this study, we showcase
examples of rare and unseasonal floods detected by their unusual runoff dynamics that are
found to be related to their documented causative mechanism. The advantage of using
such a rarity index over the approach of comparing conventional hydrograph indices is
assessed and discussed.
Thesis (PDF Available) · January 2019
https://www.researchgate.net/publication/337033870_Recurrence_Plots_and_Quantification_Analysis_of_Flood_Runoff_Dynamics