• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2021년도 학술발표회
• /
• pp.131-131
• /
• 2021

Deep learning models, especially those based on long short-term memory (LSTM), have presented their superiority in addressing time series data issues recently. This study aims to comprehensively evaluate the performance of deep learning models that belong to the supervised learning category in streamflow prediction. Therefore, six deep learning models-standard LSTM, standard gated recurrent unit (GRU), stacked LSTM, bidirectional LSTM (BiLSTM), feed-forward neural network (FFNN), and convolutional neural network (CNN) models-were of interest in this study. The Red River system, one of the largest river basins in Vietnam, was adopted as a case study. In addition, deep learning models were designed to forecast flowrate for one- and two-day ahead at Son Tay hydrological station on the Red River using a series of observed flowrate data at seven hydrological stations on three major river branches of the Red River system-Thao River, Da River, and Lo River-as the input data for training, validation, and testing. The comparison results have indicated that the four LSTM-based models exhibit significantly better performance and maintain stability than the FFNN and CNN models. Moreover, LSTM-based models may reach impressive predictions even in the presence of upstream reservoirs and dams. In the case of the stacked LSTM and BiLSTM models, the complexity of these models is not accompanied by performance improvement because their respective performance is not higher than the two standard models (LSTM and GRU). As a result, we realized that in the context of hydrological forecasting problems, simple architectural models such as LSTM and GRU (with one hidden layer) are sufficient to produce highly reliable forecasts while minimizing computation time because of the sequential data nature.

• 한국물환경학회지
• /
• 제28권3호
• /
• pp.359-366
• /
• 2012

Accurate predictions about the water quality of a river have great importance in identifying in-stream flow and water supply requirements and solving relevant environmental problems. In this study, the effect of water release from upstream dam on the downstream water quality has been investigated by applying a hydological model combined with QUAL2E to Geum River basin. The ESP (Ensemble Stream Prediction) method, which has been validated and verified by lots of researchers, was used to predict reservoir and tributary inflow. The input parameters for a combined model to predict both hydrological characteristics and water quality were identified and optimized. In order to verify the model performance, the simulated result at Gongju station, located at the downstream from Daecheong Dam, has been compared with measured data in 2008. As a result, it was found that the proposed model simulates well the values of BOD, T-N, and T-P with an acceptable reliability.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2010년도 학술발표회
• /
• pp.1279-1283
• /
• 2010

미국의 경우 1994년 발생한 대홍수(Great Flood)에 대해 사건조사를 수행하면서 예측에 포함되는 불확실성 정도를 제공하지 못하는 확정적 예측의 위험성 및 확률유량예측에 대한 필요성이 부각되었으며, 앙상블 유량예측(Ensemble Streamflow Prediction, ESP) 기법을 활용한 확률유량예측 방안에 대해 지속적으로 연구가 수행되고 있다. 국내에서도 확률예측에 대한 필요성이 인식되면서 기존 국외 연구사례를 토대로 국내 환경에 적용 가능한 방안에 대한 연구가 진행되었으며, 중장기 앙상블 유량예측의 경우 현업에서 다양한 형태로 활용되고 있다. 앙상블 유량예측의 기본이론은 예측시점의 초기조건 하에서 예측기간에 발생 가능한 기상 앙상블 시나리오를 수문모형의 입력자료로 사용하여 불확실성 범위를 설명 가능한 유량 앙상블을 모의하는 기법이다. 이러한 이론적 단순함 때문에 쉽게 현업의 유량예측 시스템 내에서 사용할 수 있다는 장점이 있으나, 동시에 기법적 특성으로 인하여 유량예측의 신뢰도가 현업에서 활용되기 어려울 정도로 낮아지는 관계로, 이러한 한계점을 극복하기 위해 그동안 기상자료 및 수문모형으로 인한 불확실성 저감에 대한 연구가 수행되었다. 하지만 예측 및 시나리오 기간의 잘못된 설정으로 기존의 불확실성 저감을 위한 연구의 적용에도 불구하고 앙상블 유량예측의 신뢰도가 오히려 낮아질 수 있으므로, 본 연구는 시나리오 기간에 따른 오차의 양상과 예측기간의 증가에 따른 초기조건의 영향을 분석하여 앙상블 유량예측의 기법적 특성 하에서 신뢰도 높은 예측을 기대할 수 있는 예측 및 시나리오 기간을 제안하였다.

• 한국물환경학회지
• /
• 제29권1호
• /
• pp.88-96
• /
• 2013

Much attention has been needed in water resource management at the watershed due to drought and flooding issues caused by climate change in recent years. Increase in air temperature and changes in precipitation patterns due to climate change are affecting hydrologic cycles, such as evaporation and soil moisture. Thus, these phenomena result in increased runoff at the watershed. The Soil and Water Assessment Tool (SWAT) model has been used to evaluate rainfall-runoff at the watershed reflecting effects on hydrology of various weather data such as rainfall, temperature, humidity, solar radiation, wind speed. For bias-correction of RCP data, at least 30 year data are needed. However, for most gaging stations, only precipitation data have been recorded and very little stations have recorded other weather data. In addition, the RCP scenario does not provide all weather data for the SWAT model. In this study, two scenarios were made to evaluate whether it would be possible to estimate streamflow using measured precipitation and long-term average values of other weather data required for running the SWAT. With measured long-term weather data (scenario 1) and with long-term average values of weather data except precipitation (scenario 2), the estimate streamflow values were almost the same with NSE value of 0.99. Increase/decrease by ${\pm}2%$, ${\pm}4%$ in temperature and humidity data did not affect streamflow. Thus, the RCP precipitation data for Hongcheon watershed were bias-corrected with measured long-term precipitation data to evaluate effects of climate change on streamflow. The results revealed that estimated streamflow for 2055s was the greatest among data for 2025s, 2055s, and 2085s. However, estimated streamflow for 2085s decreased by 9%. In addition, streamflow for Spring would be expected to increase compared with current data and streamflow for Summer will be decreased with RCP data. The results obtained in this study indicate that the streamflow could be estimated with long-term precipitation data only and effects of climate change could be evaluated using precipitation data as shown in this study.

• 한국수자원학회논문집
• /
• 제39권3호
• /
• pp.187-198
• /
• 2006

한국수자원공사에서는 매월 1일 생성한 월 앙상블 유량예측(Ensemble Streamflow Prediction, ESP)을 근거로 월 최적운영 모형인 SSDP모형을 통해 월말목표저수량을 산정할 수 있는 실시간 물 관리 시스템을 구축하였다. 그러나 월 중간에 발생하는 수문기상학적 변화를 반영할 수 없다는 단점을 가지고 있어 이를 보완하자는 필요성이 제기되었다. 이를 위해 본 연구에서는 1일부터 예측시점까지는 그 동안 발생한 강우 관측자료를 이용하고 이후 기간에는 발생 가능한 모든 과거 강수시나리오를 이용하는 기법을 통해 매월 10일 간격으로 3차례 앙상블예측을 갱신하여 예측정확도를 향상시켰으며, 예측정확도에 따른 저수지 운영효과의 개선여부를 정량적으로 분석하였다. 그 결과 월중 수문상황을 반영한 SSDP/ESP21모형이 그렇지 못한 SSDP/ESP01모형에 비해 연평균 $1\;X\;10^6\;m^3$의 용수부족 감소효과를 가졌으며 전반적으로 이수기에 총 유량이 적으면서 예측정확도가 향상된 경우 상당한 용수부족 감소효과를 나타냄을 또한 알 수 있었다.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2005년도 학술발표회(1)
• /
• pp.724-724
• /
• 2005

• 한국환경과학회지
• /
• 제13권7호
• /
• pp.611-618
• /
• 2004

The Neural Network Models which mathematically interpret human thought processes were applied to resolve the uncertainty of model parameters and to increase the model's output for the streamflow forecast model. In order to test and verify the flood discharge forecast model eight flood events observed at Kumho station located on the midstream of Kumho river were chosen. Six events of them were used as test data and two events for verification. In order to make an analysis the Levengerg-Marquart method was used to estimate the best parameter for the Neural Network model. The structure of the model was composed of five types of models by varying the number of hidden layers and the number of nodes of hidden layers. Moreover, a logarithmic-sigmoid varying function was used in first and second hidden layers, and a linear function was used for the output. As a result of applying Neural Networks models for the five models, the N10-6model was considered suitable when there is one hidden layer, and the Nl0-9-5model when there are two hidden layers. In addition, when all the Neural Network models were reviewed, the Nl0-9-5model, which has two hidden layers, gave the most preferable results in an actual hydro-event.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2015년도 학술발표회
• /
• pp.439-439
• /
• 2015

The temporal and spatial relationship of the weather elements such as rainfall and temperature is closely linked to the streamflow simulation, especially, to the flood forecasting problems. For the study area, Imjin river basin, which has the specific characteristics in geography with river cross operation between North and South Korea, the meteorological information in the northern area is totally deficiency, lead to the inaccuracy of streamflow estimation. In the paper, this problem is solved by using the combination of global (such as soil moisture content, land use) and local hydrologic components data such as weather data (precipitation, evapotranspiration, humidity, etc.) for the model-driven runoff (surface flow, lateral flow and groundwater flow) data in each subbasin. To compute the streamflow in Imjin river basin, this study is applied the hydrologic model SURR (Sejong Univ. Rainfall-Runoff) which is the continuous rainfall-runoff model used physical foundations, originally based on Storage Function Model (SFM) to simulate the intercourse of the soil properties, weather factors and flow value. The result indicates the spatial variation in the runoff response of the different subbasins influenced by the input data. The dependancy of runoff simulation accuracy depending on the qualities of input data and model parameters is suggested in this study. The southern region with the dense of gauges and the adequate data shows the good results of the simulated discharge. Eventually, the application of SURR model in Imjin riverbasin gives the accurate consequence in simulation, and become the subsequent runoff for prediction in the future process.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2005년도 학술발표회 논문집
• /
• pp.1450-1454
• /
• 2005

수자원관리에 있어서 미래시구간에 대한 유량예측은 수자원시스템운영자에게 있어서 의사결정에 결정적인 영향을 미치는 가장 중요한 요소 중의 하나이다. 효율적 물배분이나 발전 등의 이수활동을 위해서 최소 월단위 이상의 장기유량예측이 필요하며, 이를 위해서는 강우예측이 선행되어야 하는데, 본 연구에서는 통합 실시간 물관리 운영시스템을 위한 중장기 유량예측을 목표로 방법론을 제시하고자 한다. 중장기 유량예측을 수행하는 대표적인 방법 중의 하나는 앙상블 유량예측(ESP; Ensemble Streamflow Prediction) 기법이다. ESP란 현재의 유역상태를 초기조건으로 사용하고 과거의 온도나 강수 등의 시계열앙상블을 모형입력으로 이용해서 강우-유출모형을 통하여 유출량을 예측하는 기법이다. ESP는 결국 현재의 유역상태와 유역에서의 과거강우관측기록, 미래강우예측에 대한 정보를 조합하여 그에 따른 유출앙상블을 생산해 내게 된다. 유출앙상블은 각 앙상블 트레이스가 갖게 되는 가중치에 따라 확률분포를 달리 갖게 되고 경우에 따라서는 유량으로부터 2차적으로 유도되는 변수들의 확률분포로 전이되기도 한다. 기존의 ESP 이론은 미국 NWS의 범주형 확률예보를 근간으로 하고 있어, 이를 국내 환경에 그대로 적용시키기에 어려움이 있어 왔다. 따라서 본 연구에서는 국내 기상청의 월간 강수전망을 이용하고, 이러한 정보의 특성에 맞는 ESP기법을 제시하였다. 더 나아가 중장기 수자원운영을 위한 일단위 월강수시나리오 구성을 위해서 수치예보와 월강수전망을 조합하여 ESP를 사용하는 기법을 제시하였다.

• Water Engineering Research
• /
• 제5권2호
• /
• pp.101-109
• /
• 2004

Aggregating information by combining forecasts from two or more forecasting methods is an alternative to using forecasts from just a single method to improve forecast accuracy. This paper describes the development and use of a monthly inflow forecast model based on an optimal linear combination (OLC) of forecasts derived from naive, persistence, and Ensemble Streamflow Prediction (ESP) forecasts. Using the cross-validation technique, the OLC model made 1-month ahead probabilistic forecasts for the Chungju multi-purpose dam inflows for 15 years. For most of the verification months, the skill associated with the OLC forecast was superior to those drawn from the individual forecast techniques. Therefore this study demonstrates that OLC can improve the accuracy of the ESP forecast, especially during the dry season. This study also examined the value of the OLC forecasts in reservoir operations. Stochastic Dynamic Programming (SDP) derived the optimal operating policy for the Chungju multi-purpose dam operation and the derived policy was simulated using the 15-year observed inflows. The simulation results showed the SDP model that updated its probability from the new OLC forecast provided more efficient operation decisions than the conventional SDP model.