Research Areas

Hydrologic Modeling & Extremes

Hydrological modeling · Design rainfall · Floods & droughts

We build and calibrate hydrological models for flood design and risk assessment, with special attention to basins where observations are scarce, and we analyze how climate change is reshaping hydrologic extremes.

• Hydrological modeling and parameter estimation in ungauged basins
• Design rainfall, peak discharge, and probable maximum flood (Clark unit hydrograph, storage effect)
• Climate change and extreme events — floods, droughts, compound dry–wet transitions
• Stochastic and statistical analysis; time–frequency analysis with wavelet transform
• Spatiotemporal variability of hydrometeorological and environmental variables

Machine Learning for Water

AI-driven hydrometeorological forecasting

We develop hybrid machine learning frameworks that pair signal decomposition and feature selection with deep learning to forecast key water variables more accurately.

• Reference evapotranspiration and rainfall forecasting for sustainable agriculture
• Drought forecasting with wavelet-based preprocessing (TQWT, MODWT)
• Wastewater treatment plant inflow forecasting
• Streamflow and climate projections under CMIP6 scenarios

Seeing and Hearing the Rain

Image- and sound-based rainfall observation

We turn everyday sensors into rain gauges. Surveillance cameras, infrared imaging, and raindrop sounds recorded by low-cost IoT devices can provide dense, real-time rainfall information where conventional gauges are sparse.

• Rainfall intensity and raindrop size distribution from CCTV and infrared cameras, including night-time conditions
• Rain-streak extraction and rainfall image databases built in a climatic environment chamber
• Acoustic rainfall estimation with CNNs on spectrograms; vibration-sensor rainfall information
• Citizen-science rainfall observation with low-cost IoT sensors

Satellites for Floods & Disasters

Satellite-based disaster detection

We use satellite data — especially Synthetic Aperture Radar, which sees through clouds and at night — to map floods and monitor disaster risk, and satellite reanalysis products to characterize rainfall across the Korean Peninsula.

• Flood detection and mapping using PolSAR decompositions and deep learning (C-band, ALOS-2)
• Satellite image analysis for disaster detection and infrastructure risk monitoring
• Grid-based precipitation quantiles from satellite/reanalysis data (ERA5-Land)
• Satellite land surface temperature and snow-disaster susceptibility mapping