• Magazine of the Korean Society of Agricultural Engineers
• /
• v.45 no.2
• /
• pp.45-57
• /
• 2003

An artificial neural network model was developed to analyze and forecast Short-term river runoff from the Naju watershed, in Korea. Error back propagation neural networks (EBPN) of hourly rainfall and runoff data were found to have a high performance In forecasting runoff. The number of hidden nodes were optimized using total error and Bayesian information criterion. Model forecasts are very accurate (i.e., relative error is less than 3% and $R^2$is greater than 0.99) for calibration and verification data sets. Increasing the time horizon for application data sets, thus mating the model suitable for flood forecasting. decreases the accuracy of the model. The resulting optimal EBPN models for forecasting hourly runoff consists of ten rainfall and four runoff data(ANN0410 model) and ten rainfall and ten runoff data(ANN1010 model). Performances of the ANN0410 and ANN1010 models remain satisfactory up to 6 hours (i.e., $R^2$is greater than 0.92).

• Journal of Korea Water Resources Association
• /
• v.53 no.3
• /
• pp.225-236
• /
• 2020

It is not easy to provide sufficient lead time for flood forecast in urban and small mountain basins using on-ground rain gauges, because the time concentration in those basins is too short. In urban and small mountain basins with a short lag-time between precipitation and following flood events, it is more important to secure forecast lead times by predicting rainfall amounts. The Han River Flood Control Office (HRFCO) in South Korea produces short-term rainfall forecasts using the Mcgill Algorithm for Precipitation-nowcast by Lagrangian Extrapolation (MAPLE) algorithm that converts radar reflectance of rainfall events. The Flash Flood Research Center (FFRC) in the Korea Institute of Civil Engineering and Building Technology (KICT) installed a flash flood forecasting system using the short-term rainfall forecast data produced by the HRFCO and has provided flash flood information in a local lvel with 1-hour lead time since 2019. In this study, we addressed the flash flood forecasting system based on the radar rainfall and the assessed the accuracy of the forecasting system for the recorded flood events occurred in 2019. A total of 31 flood disaster cases were used to evaluate the accuracy and the forecast accuracy was 90.3% based on the probability of detection.

• Journal of Korea Water Resources Association
• /
• v.48 no.5
• /
• pp.409-423
• /
• 2015

In this study the very short-term rainfall forecasting and storm water forecasting using the weather radar data were implemented in an urban stream basin. As forecasting time increasing, the very short-term rainfall forecasting results show that the correlation coefficient was decreased and the root mean square error was increased and then the forecasting model accuracy was decreased. However, as a result of the correlation coefficient up to 60-minute forecasting time is maintained 0.5 or higher was obtained. As a result of storm water forecasting in an urban area, the reduction in peak flow and outflow volume with increasing forecasting time occurs, the peak time was analyzed that relatively matched. In the application of storm water forecasting by radar rainfall forecast, the errors has occurred that we determined some of the external factors. In the future, we believed to be necessary to perform that the continuous algorithm improvement such as simulation of rapid generation and disappearance phenomenon by precipitation echo, the improvement of extreme rainfall forecasting in urban areas, and the rainfall-runoff model parameter optimizations. The results of this study, not only urban stream basin, but also we obtained the observed data, and expand the real-time flood alarm system over the ungaged basins. In addition, it is possible to take advantage of development of as multi-sensor based very short-term rainfall forecasting technology.

• Magazine of the Korean Society of Agricultural Engineers
• /
• v.37 no.5
• /
• pp.81-89
• /
• 1995

One of the most difficult problem to estimate the flood inflow is how to understand the effective rainfall. The effective rainfall is absolutely influenced by the condition of soil moisture in the watershed just before the storm event. DAWAST model developed to simulate the daily streamflow considering the meteologic and geographic characteristics in the Korean watersheds was applied to understand the soil moisture and estimate the effective rainfall rather accurately through the daily water balance in the watershed. From this soil moisture and effective rainfall, concentration time, dimensionless hydrograph, and addition of baseflow, the rainfall-runoff model for flood flow was developed by converting the concept of long-term runoff into short-term runoff. And, real-time flood forecasting model was also developed to forecast the flood-inflow hydrograph to the river and reservoir, and called RETFLO model. According to the model verification, RETFLO model can be practically applied to the medium and small river and reservoir to forecast the flood hydrograph with peak discharge, peak time, and volume. Consequently, flood forecasting and warning system in the river and the reservoir can be greatly improved by using personal computer.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2004.05b
• /
• pp.1285-1289
• /
• 2004

Elman Discrete Recurrent Neural Networks Model(EDRNNM) was used to be a suitable short-term hydrological forecasting tool yielding a very high degree of flood stage forecasting accuracy at Musung station of Wi-stream one of IHP representative basins in South Korea. A relative new approach method has recurrent feedback nodes and virtual small memory in the structure. EDRNNM was trained by using two algorithms, namely, LMBP and RBP The model parameters, optimal connection weights and biases, were estimated during training procedure. They were applied to evaluate model validation. Sensitivity analysis test was also performed to account for the uncertainty of input nodes information. The sensitivity analysis approach could suggest a reduction of one from five initially chosen input nodes. Because the uncertainty of input nodes information always result in uncertainty in model results, it can help to reduce the uncertainty of EDRNNM application and management in small catchment.

• Journal of Korea Water Resources Association
• /
• v.48 no.12
• /
• pp.995-1009
• /
• 2015

Weather radar rainfall data has been recognized for making valuable contributions to short-term flood forecasting and management over the past decades. There are several advantages to better monitoring rainfall in ungauged area compared to ground-based rain gauges with which spatial patterns of the rainfall are not effectively identified. Hence, this study aims to develop a new scheme to forecast spatio-temporal rainfall field. The proposed model was based on an advection scheme to track wind patterns and velocity. The results showd a promising forecasting skill with quantitative and qualitative measures. It was confirmed that the forecasted rainfall may be effectively used an input data for a distributed hydrological model.

• Journal of Korea Water Resources Association
• /
• v.37 no.1
• /
• pp.67-75
• /
• 2004

In this study, real-time forecasting model(Neural Dam Inflow Forecasting Model; NDIFM) based on neural network to predict the dam inflow which is occurred by flood runoff is developed and applied to check its availability for the operation of multi-purpose reservoir Developed model Is applied to predict the flood Inflow on dam Nam-Gang in Nak-dong river basin where the rate of flood control dependent on reservoir operation is high. The input data for this model are average rainfall data composed of mean areal rainfall of upstream basin from dam location, observed inflow data, and predicted inflow data. As a result of the simulation for flood inflow forecasting, it is found that NDIFM-I is the best predictive model for real-time operation. In addition, the results of forecasting used on NDIFM-II and NDIFM-III are not bad and these models showed wide range of applicability for real-time forecasting. Consequently, if the quality of observed hydrological data is improved, it is expected that the neural network model which is black-box model can be utilized for real-time flood forecasting rather than conceptual models of which physical parameter is complex.

• Korean Journal of Agricultural Science
• /
• v.46 no.4
• /
• pp.843-856
• /
• 2019

Real-time flood prediction has an important role in significantly reducing potential damage caused by floods for urban residential areas located downstream of river basins. This paper presents an effective approach for flood forecasting based on the construction of a deep neural network (DNN) model. In addition, this research depends closely on the open-source software library, TensorFlow, which was developed by Google for machine and deep learning applications and research. The proposed model was applied to forecast the flowrate one, two, and three days in advance at the Son Tay hydrological station on the Red River, Vietnam. The input data of the model was a series of discharge data observed at five gauge stations on the Red River system, without requiring rainfall data, water levels and topographic characteristics. The research results indicate that the DNN model achieved a high performance for flood forecasting even though only a modest amount of data is required. When forecasting one and two days in advance, the Nash-Sutcliffe Efficiency (NSE) reached 0.993 and 0.938, respectively. The findings of this study suggest that the DNN model can be used to construct a real-time flood warning system on the Red River and for other river basins in Vietnam.

• Journal of Korea Water Resources Association
• /
• v.54 no.12
• /
• pp.1329-1338
• /
• 2021

Rain radar provides high spatio-temporal radar rainfall that can be used as input data to short-term precipitation forecasting models. Korea Institute of Civil Engineering and Building Technology (KICT) has developed a flash flood forecasting system that is providing flash flood forecasting based on short-term rainfall forecasts estimated by the radar rainfall. Accuracy of the radar rainfall as well as the short-term rainfall forecasts, however, can deteriorate when radar polarimetric variables have error. In this study, we develope real-time program that can correct the error inherent in the radar polarimetric variables. First, effect according to the correction of the error was verified using 363 rainfall events on non real-time. The accuracy (1-NE) of the radar rainfall was approximately 70% and correlation coefficient was higher than 0.8 after correcting the error on non real-time. The accuracy (1-NE) using the real-time program was also approximately 70% after correcting the error.

• Korean Journal of Agricultural Science
• /
• v.47 no.3
• /
• pp.471-483
• /
• 2020

Flood prediction is an important issue to prevent damages by flood inundation caused by increasing high-intensity rainfall with climate change. In recent years, machine learning algorithms have been receiving attention in many scientific fields including hydrology, water resources, natural hazards, etc. The performance of a machine learning algorithm was investigated to predict the water elevation of a river in this study. The aim of this study was to develop a new method for securing a large enough lead time for flood defenses by predicting river water elevation using the a long- short-term memory (LSTM) technique. The water elevation data at the Oisong gauging station were selected to evaluate its applicability. The test data were the water elevation data measured by K-water from 15 February 2013 to 26 August 2018, approximately 5 years 6 months, at 1 hour intervals. To investigate the predictability of the data in terms of the data characteristics and the lead time of the prediction data, the data were divided into the same interval data (group-A) and time average data (group-B) set. Next, the predictability was evaluated by constructing a total of 36 cases. Based on the results, group-A had a more stable water elevation prediction skill compared to group-B with a lead time from 1 to 6 h. Thus, the LSTM technique using only measured water elevation data can be used for securing the appropriate lead time for flood defense in a river.