Introduction
Precipitation and river discharge are among the most important variables
of the global water cycle as they reflect a holistic image about the
hydrological processes happing within and over a river basin.
Precipitation has enormous spatiotemporal variations which pose a great
challenge to maintain sufficient estimates (Ji and Kang, 2015).
Streamflow monitoring, on the other hand, is extremely useful to address
various water-related applications.
The climate records of the past decades had documented evolving
recurrent extreme weather events and natural disasters around the world
(Arnell, 1999; Nohara et al. , 2006), subjected with negative
socioeconomic implications (Wang and Zhang, 2018). Climate change
results in a raise in atmospheric temperature coupled with obvious
alteration to precipitation patterns (Labat et al. , 2004; Guptaet al. , 2015). As a consequence of climate change which is
evolving apparently in the long run, the frequency of natural disasters
such as typhoons, severe tropical cyclones, floods and droughts have
been intensified (Hein et al. , 2019). Moreover, it was reported
by the United Nations International Strategy for Disaster Reduction
(UNISDR), between 1998 and 2017, that 91% of all documented disasters
in the whole world were induced by extreme weather events including
floods, droughts, heatwaves, etc. (Wallemacq and Below, 2017).
In fact, the Japanese Archipelago has a distinctive position where
numerous natural disasters happen frequently including seismic, volcanic
activities, tsunamis, typhoons, and floods mainly due to being located
in the Ring of Fire (Shimokawa et al. , 2016). These natural
disasters degrade the national sustainable development aspirations and
pose additional serious barrier for Japan as it faces multiple
challenges fundamentally in terms of population shrinking and declining
skilled workforce. Therefore, natural disaster prevention and mitigation
framework must be carried out and implemented by a national committee,
and hence the role of a multi-disciplinary team is to understand
vulnerability and how to minimize and overcome the potential adverse
implications (Alcántara-Ayala, 2002). As a result, additional
contributions are still required to investigate and address how
ecosystems are directly and/or indirectly modified by various
hydrological processes during normal and extreme climates.
For hydrological and water resources studies, it is vital to understand
streamflow properties induced by extreme climate response during both
short-term (few hours) and long-term (several days to several years) and
impact of human activities as well. Indeed, diverse parameterization
methods have been developed to characterize river discharge patterns and
identify changes and complexity in them (Pan et al. , 2012; Stosicet al. , 2018).
Mountain rivers over the world
have a vital role in in maintaining water ecology and conserving
biodiversity, as well as, their key functions in flood control (Chenet al. , 2019). However, mountain rivers are significantly
vulnerable to problems associated with heavy rains during short time.
This could be attributed to the fact that stream velocity in mountain
regions can vary within a system and subjected to chaotic turbulence
(Mihailović et al. , 2014). Therefore, investigating the
fluctuation and complexity of flow properties for mountain streams will
deliver profound understanding about streamflow patterns and their
corresponding responses influenced by hydrologic climates and/or human
activities.
Considering the future scenarios of streamflow in East Asia, it was
projected in the literature that there will be an increase in river
discharge by the coming decades (e.g. Arnell, 1999; Nohara et al.,
2006). Sato et al., (2012) inferred that at the end of this century
river flow will rise due to increases in precipitations. Likewise,
Higashino and Stefan (2019) concluded that the annual maximum discharge
in Japanese streams are expected to be increased.
Certainly, heavy seasonal rainfall that occur frequently in the western
part of Japan is among the worst destructive disasters, since it
accompanies by landslides and mudflows. Furthermore, out of all Japanese
prefectures, Hiroshima was ranked as the first prefecture in Japan that
has the highest number of mountainous slopes (\(\sim\)32,000) to be
susceptible to landslide and mudflow disasters (Tsuchida et al., 2014),
followed by Shimane and Yamaguchi prefectures 22,300 and 22,250,
respectively. Due to the frequent huge precipitation and associated
sudden landslide and mudflow disasters, for the time being, the Ministry
of Land Infrastructure and Tourism (MLIT) of Hiroshima prefecture
installed a network of real-time water level measurement that collects
measurements at multiple sites over Hiroshima’s streams, aiming to build
up a profound knowledge about the different characteristics related to
streamflow response during various rainy events and hence to mitigate
the potential risk accompanied with heavy precipitation.
In the recent years, information-based theories have received increased
interest in the hydrological studies to detect and address the
variability in numerous hydrological variables including precipitation,
temperature, and streamflow (e.g. Brunsell, 2010; Elsner and Tsonis,
1993; Koutsoyiannis, 2005; Mishra et al., 2009). Pan et al., (2012)
documented the benefits of information-based metrics in their capability
to interpret how a model presents patterns of information content and
complexity exist in hydrological dataset. Indeed, considerable efforts
had shown the applicability of the information and complexity measures
to characterize the various patterns in time series analysis. In
particular, (Pachepsky et al. , 2006, 2016; Pan et al. ,
2011, 2012), extensively utilized the information and complexity metrics
to characterize various soil moisture, streamflow, and rainfall time
series using a straightforward symbolic strings approach of 2 characters
length per word for system description which is very useful but
uncomplicated classification. Nonetheless, there is no work had
discussed the importance of considering complex patterns of words to
characterize different system states. In other words, how to recommend
using short or long length of words to describe different patterns
embedded in a hydrological system. In addition, there is almost no work
that clearly highlighted the transformation of a system from a state to
another especially during short and long terms.
Accordingly, one of the
fundamental research questions that we aim to answer is what are the
information and hidden hydrological phenomena that can be detected by
characterizing streamflow patterns using more complex patterns according
to information and complexity theory and how to define the appropriate
pattern length that describe the different potential states of a system
(dataset, time series, etc.). Therefore, one of the main contributions
of the present research is to shed light on streamflow variations in a
mountainous river and the nested relationships within its tributaries
located at Hiroshima prefecture that has been extremely and repeatedly
deteriorated from severe floods. The particular novelty is to examine
temporal streamflow patterns at high-frequency scales using real
discharge data obtained from both classic and novel hydroacoustic
system, also at low-frequency that happen over a basin and sub-basin
scales. We also proposed a new extension for the information-based
metrics to assess streamflow patterns during flood periods. After
describing the monitoring sites in section 2, the methods are given in
section 3. Results and discussion are provided in sections 4 and 5,
respectively. Eventually, section 6 shows the research conclusions.