• Journal of The Korean Society of Agricultural Engineers
• /
• v.64 no.1
• /
• pp.39-50
• /
• 2022

In this study, we suggested the optimal training period for predicting the streamflow using the LSTM (Long Short-Term Memory) model based on the deep learning and CMIP5 (The fifth phase of the Couple Model Intercomparison Project) future climate scenarios. To validate the model performance of LSTM, the Jinan-gun (Seongsan-ri) site was selected in this study. We comfirmed that the LSTM-based streamflow was highly comparable to the measurements during the calibration (2000 to 2002/2014 to 2015) and validation (2003 to 2005/2016 to 2017) periods. Additionally, we compared the LSTM-based streamflow to the SWAT-based output during the calibration (2000~2015) and validation (2016~2019) periods. The results supported that the LSTM model also performed well in simulating streamflow during the long-term period, although small uncertainties exist. Then the SWAT-based daily streamflow was forecasted using the CMIP5 climate scenario forcing data in 2011~2100. We tested and determined the optimal training period for the LSTM model by comparing the LSTM-/SWAT-based streamflow with various scenarios. Note that the SWAT-based streamflow values were assumed as the observation because of no measurements in future (2011~2100). Our results showed that the LSTM-based streamflow was similar to the SWAT-based streamflow when the training data over the 30 years were used. These findings indicated that training periods more than 30 years were required to obtain LSTM-based reliable streamflow forecasts using climate change scenarios.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.58 no.1
• /
• pp.27-38
• /
• 2016

More severe and frequent flood and drought have increased the attentions on the river management. In particular, baseflow is an important element among many streamflow characteristics because streamflow is mainly consisted of direct runoff and baseflow. In this regard, this study attempted to analyze the relationship between streamflow variability and baseflow contributions on Nakdong river basin. For this, two Streamflow Variability Indices (SVI) were used: Coefficient of Variation (CV) and Coefficient of Flow Regime (CFR). Furthermore, baselow separation was individually conducted by three methods (PART, WHAT and BFLOW), and based on this, Baseflow Index (BFI) was calculated. Also, we used the daily streamflow data retrieved from 27 gauge stations in Nakdong river basin for baseflow separation. The results showed that BFI calculated by three models ranges from 0.14 to 0.90 for 27 gauge stations. For SVI, BFI has much higher correlation with CV than with CFR. Also, the inversely proportional relationship between BFI and CV showed that higher baseflow contribution, less streamflow variability.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.165-165
• /
• 2023

Streamflow prediction is a critical task in water resources management and essential for planning and decision-making purposes. However, the streamflow prediction is challenging due to the complexity and non-linear nature of hydrological processes. The transfer learning is a powerful technique that enables a model to transfer knowledge from a source domain to a target domain, improving model performance with limited data in the target domain. In this study, we apply the transfer learning using the Informer model, which is a state-of-the-art deep learning model for streamflow prediction. The model was trained on a large-scale hydrological dataset in the source basin and then fine-tuned using a smaller dataset available in the target basin to predict the streamflow in the target basin. The results demonstrate that transfer learning using the Informer model significantly outperforms the traditional machine learning models and even other deep learning models for streamflow prediction, especially when the target domain has limited data. Moreover, the results indicate the effectiveness of streamflow prediction when knowledge transfer is used to improve the generalizability of hydrologic models in data-sparse regions.

• Magazine of the Korean Society of Agricultural Engineers
• /
• v.32 no.E
• /
• pp.47-58
• /
• 1990

Abstract A regionalized daily streamflow model using a modified retention parameter in the SCS method was developed to predict the daily streamflow of a natural series for Korean watersheds. Model verification showed that it is possible to use the model for extending short period records in a gaged watershed or for predicting daily streamflow in any ungaged watershed, with reasonable accuracy by simply inputing the name of the watershed boundary, the watershed size, the latitude and longitude of the watershed, and the daily areal rainfall.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.59 no.6
• /
• pp.101-108
• /
• 2017

This study attempted to improve the accuracy of streamflow and baseflow prediction of Soil and Water Assessment Tool (SWAT) by applying baselfow recession constants for each sub-watershed. This study set two different scenarios (S1 and S2) to evaluate the impact of application of baseflow recession constants for each sub-watershed on streamflow prediction. In S1, Only the baseflow recession constant obtained from the streamflow station located in the final outlet of study area was applied for whole sub-watersheds. In S2, baseflow recession constants obtained from six different streamflow stations were applied for each sub-watershed. Then, baseflow was separated form the measured streamflow data and the predicted streamflow of S1 and S2 using Web-based Hydrograph Analysis Tool (WHAT). The results showed Nash-Sutcliff efficiency (NSE) and $R^2$ of S2 were a little higher than these of S1 in both streamflow and baseflow prediction results. However, it is important that S2 reflected physical meaning of baseflow recess. Also, recession part of hydrograph in S2 was calibrated better than that of S1 compared to the measured hydrograph.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.140-140
• /
• 2022

Recently an accurate quantification of streamflow under various climatological and anthropogenic factors and separation of their relative contribution remains challenging, because variation in streamflow may result in hydrological disasters. In this study, we evaluated the factors responsible for variation in streamflow in Korean watersheds, quantified separately their contribution using different Budyko-based functions, and identified hydrological breakpoint points. After detecting that the hydrological break point in 1995 and time series were divided into natural period (1966-1995), and disturbed period (1996-2014). During the natural period variation in climate tended to increase change in streamflow. However, in the disturbed period both climate variation and anthropogenic activities tended to increase streamflow variation in the watershed. Subsequently, the findings acquired from different Budyko-based functions were observed sensitive to selection of function. The variation in streamflow was observed in the response of change in climatic parameters ranging 46 to 75% (average 60%). The effects of anthropogenic activities were observed less compared to climate variation accounts 25 to 54% (average 40%). Furthermore, the relative contribution was observed to be sensitive corresponding to Budyko-based functions utilized. Moreover, relative impacts of both factors have capability to enhance uncertainty in the management of water resources. Thus, this knowledge would be essential for the implementation of water management spatial and temporal scale to reduce the risk of hydrological disasters in the watershed.

• Journal of Korea Water Resources Association
• /
• v.34 no.3
• /
• pp.275-285
• /
• 2001

Since the majority of streamflow during dry periods is provided by groundwater storage, the streamflow depends on a basin moisture state recharged from rainfall during wet periods. This hydrologic characteristics dives good condition to predict long-term streamflow if the basin state like groundwater recharge is known in advance. The objective of this study is to examine groundwater recharge effect to monthly streamflow, and to attempt monthly streamflow prediction using estimated groundwater recharge. The ground water recharge is used as an independent variable with streamflow and precipitation to construct multiple regression models for the prediction. Correlation analysis was performed to assess the effect of groundwater carry-over to streamflow and to establish the associations among independent variables. The predicted streamflow shows that the multiple regression model involved groundwater recharge gives improved results comparing to the model only using streamflow and precipitation as independent variables. In addition, this paper shows that the prediction model with the effect of groundwater carry-over taken into account can be developed using only precipitation.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.48 no.5
• /
• pp.29-38
• /
• 2006

Geographic Information System has extended to higher assessment of water resources. GIS linking with hydrological model becomes a trend in water resource assessment modeling. One of the most popular models is SWAT2000 which have effectiveness in multi-purpose processes for predicting the impact of land management practices on water, sediments and chemicals yields in large complex watershed with varying soils, land uses, and management conditions over long period of time. In this study, SWAT2000 model was applied to Gap stream watershed in Daejeon city where TMDL (Total Maximum Daily Load) Regulation would be implanted. The Gap Stream watershed was partitioned into 8 subbasins, however, only 3 out of 8 subbaisns were observed for having practical gauged data on the basis of streamflow from the year of 2002 to 2005. Gauged streamflow data of Indong, Boksu and Hoeduck stations were used for calibration and validation of the SWAT Streamflow simulation. Estimation Efficiency Analysis (COE), Regression Analysis ($R^{2}$), Relative Error (R.E.) were used for comparing observed streamflow data of the 3 subbasins on the daily and monthly basis with estimated streamflow data in order to fix optimized parameters for the best fitted results. COE value for the daily and monthly streamflow was ranged from 0.45 to 0.96. $R^{2}$ values for daily and monthly streamflow ranged from 0.51 to 0.97. R.E. values for total streamflow volume ranged from 3 % to 22.5 %. The accuracy of the model results shows that the SWAT2000 model can be applicable to Korean watersheds like the Gap Stream watershed that needs to be partitioned into a number of subbasins for TMDL regulation.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2008.05a
• /
• pp.1388-1392
• /
• 2008

Water resources management depends on dealing inherent uncertainties stemming from climatic and hydrological inputs and models. Dealing with these uncertainties remains a challenge. Streamflow forecasts basically contain uncertainties arising from model structure and initial conditions. Recent enhancements in climate forecasting skill and hydrological modeling provide an breakthrough for delivering improved streamflow forecasts. However, little consideration has been given to methodologies that include coupling both multiple climate and multiple hydrological models, increasing the pool of streamflow forecast ensemble members and accounting for cumulative sources of uncertainty. The approach here proposes integration and coupling of global climate models (GCM), multiple regional climate models, and numerous hydrological models to improve streamflow forecasting and characterize system uncertainty through generation of ensemble forecasts.

• Water Engineering Research
• /
• v.1 no.3
• /
• pp.233-242
• /
• 2000

A system for regularly appraising the reliability of streamflow data, KORSAS (KOwaco's Regular Streamflow Appraising System) was developed on PC based Windows for hydrological specialists and engineers working in the Korea Water Resources Corporation (KOWACO). The reliability of streamflow rates can be evaluated with KORSAS in various as pects according to the evaluation duration and method. The former being selected as short term (event based) or long term(continus based), and the latter being classified into comparison methods of flow measurement, other stations results, and simulation. Rainfall-runoff models can be used together with KORSAS in order to evaluate the reliability of observed flow data by comparing with simulated flow data. The objective of this study is to develop a systematic methodology in various aspects to evaluate the reliability of streamflow data regularly.