As various economic activity is complicated and the industrial sectors affected by the climate is increased, the use of long-term climate prediction data for designing proactive planning in the sectors of society has increased. Direction within the water resource management was already converted to prevent the potential problems in advance rather than to correct the problems caused by extreme climate. Despite the high usability of the long-term prediction information for proactive water quality management, it is difficult to find examples of utilization of a long-term water quality prediction material to the actual decision-making both in Korea and abroad. Low utilization in the field of such aggressive water quality control is due to the low reliability caused by the high uncertainty of long-term prediction. Further, the difference in the spatio-temporal scale between the long-term climatic prediction data that is actually provided and the prediction information required for water quality management can be an indirect cause of low utilization of the long-term climate prediction data. In this study, we developed the basic technologies for spatial and temporal downscaling technology necessary for long-term prediction of integrated streamflow and water quality and evaluated the predictability of technology that has been developed. In addition, we suggested the policy as well as possible approach for utilization of long-term climate prediction information for water quality management.

Three different downscaling methods in accordance with the degree of utilizing long-term climate prediction information were developed in order to improve the predictability and refine the spatial scale. They include: 1) Simple Bias Correction (SBC) method which directly uses long-term climate prediction data, 2) Moving Window Regression (MWR) method which indirectly utilizes long-term prediction data, 3) Climate Index Regression (CIR) method which mainly uses the observation-based climate information and additionally take advantage of the long-term prediction data. Temporal downscaling can be conducted based on the monthly downscaled prediction data (e.g. monthly precipitation and average temperature) generated using SBC, MWR, and CIR methods. The temporal downscaling method was developed based on the Mahalanobis distance which determines the closest year and month from past observations data by comparing to the monthly prediction data. Among three spatial downscaling methods developed, CIR method based on the observed climate information was most reliable in predicting the monthly average temperature and precipitation during all the period.

In the evaluation of the long-term predictability of climate prediction information through the application to the water quantity and quality management, predictability for inflow of agricultural reservoir, inflow of multipurpose dam, water quality at the medium-size watershed, and water quality in the mainstream of Nakdong River was assessed. Compared to the simulated inflow using the observation climate data, simulated inflow using predicted climate data by SBC, MWR, and CIR method within the three agricultural reservoirs with the capacity of more than 1 million tons in Nakdong River basin was analyzed. It showed results similar to those of the evaluation of the predictability of three downscaling methods. That is, the CIR method showed higher and the most reliable predictability, compared to the SBC and MWR methods. In the case of inflow prediction that targets Andong multi-purpose dam, approaches including 1) observation-based method which predicts the inflow amount based on the observation data without long-term climate prediction information, 2) modeling-based method which uses the long-term climate prediction information as model input, and 3) teleconnection-based statistical method were compared. Statistical prediction method based on the teleconnection showed the highest predictability among the three methods for all periods with the exception of December, in which predicted results were similar to each other. Long-term predictability of water quality within the Wecheon medium-size watershed within the Nakdong River basin was evaluated by comparing the modeling- based approach and teleconnection-based statistical approach. In the case of statistical method, results of long-term water quality prediction for the six constituents at three stations showed appropriate predictability: 89%, 72%, 83%, 89%, 72%, 61% of all cases (12 months × 6 constituents × 3 stations) respectively for water temperature, BOD, TN, TP, SS, and chlorophyll-a showed significant correlation coefficient which is higher than the critical value at 0.05 of significant level. According to the provisions of water quality prediction and response measures, long-term water quality predictability in the Nakdong River mainstream was evaluated by using 3-month lead-time climate prediction data generated in February, May, August, and November as model input of the HSPF-EFDC modeling system . The results revealed that water temperature showed the highest predictability and followed by TN, chlorophyll-a, and TP in the order. When compared to the results predicted using the observed climate data as model input, water quality results based on climate prediction data for all constituents showed the significant correlation coefficients higher than the critical values at 0.05 of significant level. Finally, we presented an integrated approach that takes into account various climate information, downscaling methods, and water quality prediction methods for the probabilistic prediction of water quality which can be used for the decision-making.

Survey through various water-related agencies showed that percentage of actual utilization of long-term climate prediction data was lower than that of the potential utilization by showing 50% and 70%, respectively. It was revealed that low usability was due to the low reliability of the prediction data. There was no positive response to the question on reliability of the prediction data by showing equal percent of neutral and negative responses. Also, it was found that improvement of user's convenience related to the spatio-temporal scale and file format of the long-term prediction data are necessary. Services that utilize long-term climate prediction can be used for the long-term preventive measures in water quality management including 1) prevention of non-point source pollution load, 2) decision-making for the management of the water environment and quality related facilities, and 3) integrated management of water quantity and quality. Amendment of the provisions of water quality prediction and response measures regarding to the definition of water quality prediction, predicting water quality constituents, target area, announcement, and configuration of the water quality council was suggested as a legal and institutional approach for long-term water quality prediction. In addition, legal and institutional improvement in the higher level law was recommended in order to take advantage of the long- term prediction information for the preservation of ecology and water quality. It was also proposed that content on the application of various water quality management techniques which utilize the long-term climate prediction information for the prevention purpose should be added to the guidelines and plan for the water and environment management. About the construction of the long-term water quality prediction system, GIS-based prediction system was suggested for the improved utilization of spatial information such as spatial distribution of pollutant sources and past water quality trend related to water quality management as proactive response and preventive measures.

Keyboards: Seasonal Forecast, Long-term Prediction, Water Quality, Downscaling, Climate Index, Multi-Model Ensemble (MME), Teleconnection

핵심용어: 계절예측 , 수질 , 상세화 , 기후인자 , 다중모형앙상블 , 원격상관

1. 서론

1.1 연구 배경 및 필요성

1.2 연구의 목적

1.3 연구의 방법 및 범위

2. 장기 기후예측 자료 활용 국내외 사례 조사

2.1 해외 사례

2.1.1 미국

2.1.2 호주

2.2 국내 사례

2.2.1 장기 기후예측

2.2.2 장기 수량/가뭄 예보

2.2.3 장기 수질 예보

3. 연구 재료 및 방법

3.1 대상 지역

3.2 사용 모형

3.2.1 SWAT

3.2.2 EFDC

3.3 연구 방법

3.3.1 장기 기후예측 자료의 상세화 기법 개발

3.3.3 장기 수문/수질예측성 평가

3.3.4 통합 수질예측 기법 개발

3.3.5 기후정보의 사용자 수요 조사

4. 연구결과

4.1 장기 기후예측 자료의 상세화 기법 개발

4.1.1 Simple Bias Correction(SBC) Method

4.1.2 Moving Window Regression(MWR) Method

4.1.3 Climate Index Regression(CIR) Method

4.1.4 일 자료 생성을 위한 시간적 상세화

4.2 장기 기후예측 상세화 자료의 예측성 평가

4.2.1 Simple Bias Correction Method

4.2.2 Moving Window Regression Method

4.2.3 Climate Index Regression Method

4.2.4 상세화 방법에 따른 월별 예측성능 비교

4.2.5 선행예측기간에 따른 예측성 평가

4.2.6 월별 예측 정보의 누적 기간에 따른 예측성 평가

4.2.7 공간 규모에 따른 예측성 평가

4.3 장기 수문/수질예측성 평가

4.3.1 장기 예측 자료의 농업용 저수지 상류 유입량 예측성 평가

4.3.2 장기 예측 자료의 다목적 댐 유입량 예측성 평가

4.3.3 장기 예측 자료의 중권역 유역 수질예측성 평가

4.3.4 장기 예측 자료의 낙동강 본류 수질예측성 평가

4.4 통합 수질예측 기법 개발

4.5 기후 정보의 사용자 수요조사

5. 정책제언

5.1 장기 수질예보의 효용성 확산

5.2 장기 수질예보를 위한 법·제도 개선

5.3 장기 수질예보 시스템 구축

6. 결론

REFERENCES

부록 1. 기후예측 정보를 고려한 수질관리 정책제언을 위한 전문가 설문조사

1. 설문조사서

2. 기후정보의 사용자 수요 조사 결과

2.1 단기/중기 예측 정보 수요조사

2.2 장기예측 정보 수요조사

2.3 기후변화시나리오자료수요조사

부록 2. 장기 기후예측 상세화 툴

1. R기반 프로그램 사용 방법

2. 상세화 툴 Function 리스트 및 설명