RESEARCH
From Rivers to Watersheds, and to Society.
INTRODUCTION
In recent years, the intensification and increasing frequency of extreme rainfall events associated with climate change have made flood risk management an increasingly important societal challenge.
Our laboratory conducts research on water flows and flood disasters in rivers, paddy fields, and urban areas, with high-resolution two-dimensional numerical modeling using unstructured triangular meshes as a common foundation for our research.
We seek to understand hydraulic phenomena through detailed numerical simulations while advancing the speed and capability of computational analysis by integrating high-performance computing and artificial intelligence.
We further extend these technologies to integrated watershed management, flood prediction, and evacuation simulation, aiming to develop an integrated research framework that connects the understanding and prediction of water-related disasters with practical applications for disaster prevention and mitigation.
01
High-Resolution Flood Inundation Modeling
Flood inundation modeling is one of the fundamental technologies supporting our research.
We develop two-dimensional shallow water flow models using unstructured triangular meshes
to represent complex terrain and features such as rivers, levees, roads, and buildings in detail.
The flexibility of unstructured meshes allows us to represent river channels and urban areas
within a single computational domain,
enabling the integrated simulation of both riverine flooding and pluvial flooding caused by rainfall.
By incorporating detailed topographic and land-surface information into the numerical model,
we aim to capture not only the spatial extent of flooding but also how inundation evolves over time.
We also utilize GPU-based high-performance computing
to accelerate large-scale, high-resolution simulations.
This high-resolution flood inundation model provides a common foundation for our other research themes,
including paddy field dams, evacuation simulation, and AI-based rapid flood prediction.
02
River Hydraulics & Quasi-Three-Dimensional Analysis
Understanding the structure and characteristics of river flows is essential
for river planning and flood risk management.
Our laboratory investigates quasi-three-dimensional river flow
based on two-dimensional numerical analysis to better understand complex flow phenomena in rivers.
We numerically reproduce flow characteristics in rivers with complex channel geometries
and analyze water-surface profiles, velocity distributions,
and other hydraulic properties under flood conditions.
By connecting numerical simulations with actual river phenomena,
we aim to develop hydraulic analysis technologies
that can contribute to river planning, management, and flood control.
03
Sediment Transport & Riverbed Dynamics
During floods, riverbed morphology can change significantly as sediment is transported by flowing water.
Our laboratory investigates sediment transport and riverbed dynamics
during floods through numerical modeling.
By analyzing the interactions between flow and sediment transport,
we seek to clarify how changes in riverbed morphology affect flood flows and river safety.
The knowledge gained from these studies contributes
to river maintenance, channel management, and effective flood control strategies.
04
Integrated Watershed Management & Paddy Field Dams
In response to increasingly severe rainfall disasters associated with climate change, it is becoming increasingly important to utilize the flood mitigation functions of not only rivers but also various lands and facilities throughout the watershed.
As one practical approach to integrated watershed management, our laboratory investigates the flood mitigation effects of paddy field dams.
Rather than treating paddy fields simply as storage facilities, we model the topography and structures of individual paddy fields in detail and analyze the processes of water inflow, storage, and outflow during flood events.
Using high-resolution numerical simulations, we quantitatively evaluate how flood mitigation effects vary with rainfall conditions and the location and configuration of paddy field dams, and investigate more effective strategies for their implementation.
05
Flood-Informed Evacuation Simulation
Protecting lives from floods requires understanding not only where flooding occurs,
but also how flood conditions change over time.
Our laboratory combines detailed flood inundation modeling with evacuation simulation
to investigate evacuation behavior under dynamically changing flood conditions,
while explicitly representing buildings, roads, and other urban features.
Using time-series inundation information generated by the flood model,
we incorporate changes in walking speed according to flood depth
and dynamically identify safe evacuation routes as flood conditions evolve over time.
We also evaluate how the addition or relocation of evacuation shelters affects evacuation time
and evacuation completion rates,
contributing to the development of more effective and practical evacuation plans.
06
AI & High-Performance Computing
High-resolution flood inundation modeling is essential for accurately understanding flood dynamics.
However, large-scale and high-resolution simulations can require substantial computational time.
Our laboratory therefore investigates
both high-performance computing for accelerating physics-based models
and AI-based surrogate modeling for rapid flood prediction.
AI models are trained using simulation results generated
by the high-resolution flood inundation models that we have developed and validated.
Our goal is not to use AI alone to directly estimate real-world floods.
Instead, we aim to develop surrogate models that can reproduce the results of validated,
high-fidelity physics-based models in a fraction of the computational time.