Recently, developing countries have steadily been pushing for the construction of stream-oriented smart cities, breaking away from the existing old-town-centered development in the past. Due to the accelerating effects of climate change along with such urbanization, it is imperative for urban rivers to establish a flood warning system that can predict the amount of high flow rates of accuracy in engineering, compared to using the existing Computational Fluid Dynamics (CFD) models for disaster prevention. In this study, in the case of streams where missing data existed or only small observations were obtained, the variation in flow rates could be predicted with only the appropriate deep learning models, using only limited time series flow data. In addition, the selected deep learning model allowed the minimum number of input learning data to be determined. In this study, the time series flow rates were predicted by applying the deep learning models to the Han River, which is a highly urbanized stream that flows through the capital of Korea, Seoul and has a large seasonal variation in the flow rate. The deep learning models used are Convolution Neural Network (CNN), Simple Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM) and Gated Recurrent Unit (GRU). Sequence lengths for time series runoff data were determined first to assess the accuracy and applicability of the deep learning models. By analyzing the forecast results of the outflow data of the Han River, sequence length for 14 days was appropriate in terms of the predicted accuracy of the model. In addition, the GRU model is effective for deep learning models that use time series data of the region with large fluctuations in flow rates, such as the Han River. Furthermore, through this study, it was possible to propose the minimum number of training data that could provide flood warning system with an effective flood forecasting system although the number of input data such as flow rates secured in new towns developed around rivers was insufficient.

The water-level time series of a tidal river is influenced by various factors and has a complex structure, which limits its use as hydrological forecast data. This study proposes a methodology for decomposing the water-level time series of a tidal river into various components that influence the water level. To this end, the tide, wave, rainfall-induced runoff and noise components were selected as the main components that affect the water-level time series. The tide component and the wave component were first separated through wavelet analysis and curve fitting and then they were removed from the water-level data. A high-pass filter was then applied to the resulting time series to separate the rainfall-induced runoff component and the noise component. These methods made it possible to determine the rate of influence that each component has on the water level of a tidal river. The results could be used as a basis for calibrating a rainfall-runoff model and issuing flood forecasts and warnings for a tidal river.

Stream gauge stations are facilities for measuring stream water levels and flow rates, and their main purpose is to produce the data required to analyze hydrological phenomena. However, there are no specific criteria for selecting the locations and installation densities of stream gauge stations, which results in numerous problems, including regional imbalances and overlapping. To address these issues, a stream gauge network was constructed in this study considering both the transinformation of entropy (objective function 1) and the importance of each stream gauge station (objective function 2). To account for both factors, the optimal combinations that satisfied the two objective functions were determined using the Euclidean distance. Based on the rainfall runoff analysis results, unit hydrographs reflecting stream connectivity were derived and applied to entropy theory. The importance of each stream gauge station was calculated considering its purposes, such as flood control, water use, and environment. When this method was applied to the Namgang Dam Basin, it was found out that eight out of 12 stream gauge stations were required. The combination of the selected stations reflected both the transinformation of entropy and the importance of each station.

Monitoring tidal dynamics is imperative to disaster management because it requires a high level of precision to avert possible dangers. Good knowledge of the physical drivers of tides is vital to achieving such a precision. The Taehwa River in Ulsan City, Korea experiences tidal currents in the estuary that drains into the East Sea. The contribution of wind to tide prediction is evaluated by comparing tidal predictions using harmonic analysis and three deep learning models. Harmonic analysis is conducted on hourly water level data from 2010–2021 using the commercial pytides toolbox to generate constituents and predict tidal elevations. Three deep learning models of long short-term memory (LSTM), gated recurrent unit (GRU), and bi-directional lstm (BiLSTM) are fitted to the water level and wind speed to evaluate wind and no-wind scenarios. Results show that Taehwa tides are categorized as semidiurnal tides based on a computed form ratio of 0.2714 in a 24-h tidal cycle. The highest tidal range of 0.60 m is recorded on full moon spring tide indicating the significant lunar pull. Wind effect improved tidal prediction NSE of optimal LSTM model from 0.67 to 0.90. Knowledge of contributing effect of wind will inform flood protection measures to enhance disaster preparedness.

The choice of the computational time step (dt) value and the method for setting dt can have a bearing on the accuracy and performance of a simulation, and this effect has not been comprehensively researched across different simulation conditions. In this study, the effects of the fixed time step (FTS) method and the automatic time step (ATS) method on the simulated runoff of a distributed rainfall–runoff model were compared. The results revealed that the ATS method had less peak flow variability than the FTS method for the virtual catchment. In the FTS method, the difference in time step had more impact on the runoff simulation results than the other factors such as differences in the amount of rainfall, the density of the stream network, or the spatial resolution of the input data. Different optimal parameter values according to the computational time step were found when FTS and ATS were used in a real catchment, and the changes in the optimal parameter values were smaller in ATS than in FTS. The results of our analyses can help to yield reliable runoff simulation results.

This paper presents an algorithm that can efficiently simulate a grid-based one-dimensional distributed rainfall-runoff model by performing parallel computations using flow accumulation values for individual grid cells, which are calculated through an eight flow direction method. This parallel computation algorithm uses information about flow accumulation to automatically find parallel computation target grid cells within the overall area and perform parallel computation on the grid by unit. The Microsoft.NET Parallel class was used to apply and evaluate the parallel computation algorithm independently on two machines. The results showed that the time reduction effect of parallel computation differed for each target domain, because flow accumulation values varied depending on the domain. Parallel computation reduced computation time by around 40% to 78% in virtual domains and around 63% in the real domain compared to sequential computation. The results of this study can be utilized to reduce the computation time of distributed models.