Evaluating the Utility of IPCC AR4 GCMs for Hydrological Application in South Korea

Understanding the uncertainty of climate models in space and time is necessary to help water resources managers and hydrologists in the selection of appropriate model for a specific application. In this paper, we use three separate methods to evaluate and compare the utility of 14 climate models for seven basins with area range of 2,656–26,355 km² on the South Korean Peninsula. On the one hand, the method of probabilistic uncertainty analysis is used to evaluate the capability of the studied General Circulation Models (GCMs) in recognizing the extreme events. On the other hand, we use two statistical tests (correlation coefficient and root mean square error) to examine the capability of the GCMs in simulating quantitatively each event. The results show that, for the first method, the performance of climate model varies depending on the number of climate model nodes used for a specific application of given basin, especially for monthly time scale. In addition, we find that, there are several GCMs showing good results for the probabilistic uncertainty test but poor results for the statistical test and conversely. Therefore, climate models should be evaluated for specific applications and specific regions. The results indicated quite clearly that, it is not easy to select an optimal climate model which can satisfy both applications using precipitation and temperature projections. However, the results of this study suggest that, there are several GCMs which are more useful than the others for general hydrological application in South Korean peninsula.     

Analyzing the Variability in Low-Flow Projections under GCM CMIP5 Scenarios

Water Resources Management - The present study considered the impacts of global climate model (GCM) selection in the Couple Model Intercomparison Phase 5 (CMIP5) scenarios on the low-flow...     

Utility of Ten-Day Climate Model Ensemble Simulations for Water Resources Applications in Korean Watersheds

We demonstrate the use of a quantitative measure of the effectiveness of using climate model simulations of surface precipitation and temperature for water resources applications involving extremes of watershed average precipitation and temperature, and watershed discharge. This diagnostic measure is considered in association with the use of climate information to condition ensemble seasonal predictions of watershed variables. Seven watersheds in the Korean peninsula constitute the application sites. The climate model effectiveness is expressed by a utility index E_P that measures the ability of the climate model simulations of an indicator variable (i.e., nodal precipitation or temperature) to discriminate observed distributions of the highs and lows of a watershed target variable (i.e., mean areal precipitation and temperature as well as outlet discharge). Monte Carlo simulations provide estimates of the significance of the E_p values. For apparently the first time, ten-member ensemble simulations of daily surface precipitation and temperature from the Korean Meteorological Agency climate model are used to evaluate the climate-model utility index E_P for a temporal interval of 10 days for each application watershed. The results show that, in spite of the high uncertainty of climate simulations, there are several Korean watersheds that can benefit from the use of climate model simulations of high temporal resolution for planning and management studies that involve precipitation, temperature and discharge. In particular, seasonal ensemble prediction of watershed variables stands to gain from conditioning on high-temporal resolution climate forecasts.     

Utilization of the Bayesian Method to Improve Hydrological Drought Prediction Accuracy

This study established a hydrological drought forecasting system based on the Bayesian method and evaluated its utilization for South Korea. The regression result between Historical Runoff (HR) and Ensemble Streamflow Prediction Runoff (ESP_R) was used as prior information in the Bayesian method. Additionally Global seasonal forecast System 5 Runoff (GS5_R) produced using a dynamic prediction method was used in a likelihood function. Bayesian Runoff (BAY_R), as posterior information, was generated and compared with the ESP_R and GS5_R results for predictive ability evaluation. The Standardized Runoff Index (SRI) was selected for the drought prediction, and the BAY_SRI, GS5_SRI and ESP_SRI were computed using BAY_R, GS5_R and ESP_R, respectively. The Correlation Coefficient (CC), Nash-Sutcliffe Efficiency (NSE) and Receiver Operating Characteristic (ROC) score of BAY_SRI were the highest, and the Root Mean Square Error (RMSE) of BAY_SRI was the lowest among the methods. The Bayesian method improved the behavioral and quantitative error of drought prediction and the predictive ability of the occurrence of drought. In particular, the simulation accuracy was significantly improved during the flood season. Additionally, BAY_SRI represented past drought scenarios better than did the other two methods. Overall, we found that the Bayesian method could be applied for hydrological drought predictions for based on 1- and 2-month lead times.     

Improving ANFIS Based Model for Long-term Dam Inflow Prediction by Incorporating Monthly Rainfall Forecasts

The necessity of long-term dam inflow forecast has been recognized for many years. Despite numerous studies, the accurate long-term dam inflow prediction is still a challenging task. This paper presents an adaptive neuro-fuzzy inference system (ANFIS) based model and evaluates the applicability of categorical rainfall forecast for improvement of monthly dam inflow prediction. In order to obtain appropriate ANFIS model configuration for dam inflow prediction, several models were trained and tested using various numbers of input variables i.e. monthly observed rainfall, relative humidity, temperature, dam inflow and categorical monthly rainfall forecast. The ANFIS based models were configured and evaluated for six major dams of South Korea i.e. Andong, Chungju, Daecheong, Guesan, Soyang and Sumjin having high, medium and low reservoir capacity. The results showed significant improvement in dam inflow prediction for all the selected dams using the ANFIS based model with categorical rainfall forecast compared to the ANFIS based model with only preceding month’s dam inflow and weather data.