Floods are one of the most common and frequent natural disaster events in the world. It can lead to very serious loss in terms of lives, agriculture, buildings, roads, and infrastructures on the national level. In recent years, flood damage has been accelerated due to the effects of global climate change and abnormal climate phenomena. However, with the advancement of computer technology in recent years, the speed of computation has accelerated, enabling the prediction of a certain level of climate over various time scales. This is called seamless prediction, which is a concept of securing a certain level of predictability from the very short term to the very long term of climate change. Well-known abnormal weather phenomena such as El Niño, La Niña and Indian Ocean Dipole can especially be used as the main information in determining the long-term variability of water resources. Due to the seamless prediction technology, the approach of flood disaster management is gradually becoming diversified and refined. Seamless predictions can serve as a new paradigm for the prediction of long-term flood risks, and are increasingly expanding the range of preventive and prepared roles

to ensure an effective lead time for flood disasters.

This study performed the mid- and long-term flood risk assessment using predicted climate information through the statistical-dynamical models, and carried out the utilization plan for Korean local governments. The study processed the Automated Synoptic Observing System (ASOS) rainfall levels from the Korean Meteorological Administration (KMA), and its weather information typically used for flood risk assessment. The study also collected and analyzed geographic information system (GIS) data for terrain and spatial information. In addition, nonlinear correlation was estimated using the teleconnection between the climate index and hydrological variables, and the possibility of time series prediction using the climate index was diagnosed. Furthermore, monthly rainfall forecasting in local areas was carried out by constructing statistical forecasting models (e.g.; AR, MA, ARMA, ARIMA, VAR)

for each region of Korea. The predictability of each model result was examined by taking ensemble members. The Multi Model Ensemble (MME) prediction information provided by the APEC Climate Center was used to predict the monthly precipitation using the dynamical climate model.

Next, for the flood risk analysis, a Delphi-Analytic Hierarchy Process (AHP) method was used from expert surveys to develop the flood risk index, and to calculate the weight for each index. In addition, the long-term flood risk was analyzed for the Korean 243-local governments, and used a number of GIS layers that were generated in a digital format to analyze flood risk and vulnerability. In order to assess flood hazard and vulnerability, the hierarchy of the integrated flood risk index was assessed by separating several criteria and indicators, which include hydrological hazard, socio-economic exposure, and socio-economic vulnerability. The indicators were classified, standardized, and weighted, and then were combined to gain the integrated flood risk index map. Different levels of flood risk were defined to be: (1) “Very High”, (2) “High”, (3) “Medium”, (4) “Low”, and (5) “Very Low”. The integrated flood risk index confirmed that there is an increasing tendency of risk in the recent period and also that there is a somewhat increased tendency of risk

in the case of the Central Pacific El Niño period. When forecasting the summer (June to September, JJAS) integrated flood risk in 2016, it is confirmed that the integrated flood risk increased from June to July, and that there is decreasing tendency of risk in August and September in Korea’s central region. However, in the southern region, there was high flood risk in June, especially in the southern

coastal region. In July, the flood risk decreased overall, with the exception of a part of Changwon in Gyeongnam, and increased again in August and September.

Based on the results of this study, several lessons were obtained regarding the enhancement of policy standards and formulation of measures, in order to reduce long-range flood risk in local governments. The results of this study are expected to provide useful data for reducing flood risk to stakeholders and decision makers. Finally, the long-range flood risk forecasting in local areas can propose the possibility of mid- and long-term planning and forecasting of flood risk for the management of water resources for one to three months of seasonal forecasting information. It can also be used as useful forecasting information for acquisition of lead time in reducing the impacts of disasters such as drought and flood.

Keyword : Integrated Flood Risk, Long-Range Forecasting, Statistical-Dynamical

Model, Korean Local Government

1. 서론

1.1 배경 및 필요성

1.2 연구 동향

1.3 연구 목적 및 범위

1.4 연구 내용 및 방법

2. 연구 내용 및 방법

2.1 국외 사례 조사

2.1.1 미국 기후 탄력 툴킷

2.1.2 미국 기상청 국가 수문 분석 시스템

2.1.3 호주 기상청 계절유량 예측 시스템

2.1.4 영국 기상청 영향예보 시스템

2.2 연구 대상지역 및 자료 분석

2.2.1 연구 대상지역

2.2.2 연구 대상 자료

2.2.3 수문기상 자료

2.2.4 티센 네트워크 구축

2.2.5 강수의 변동성/경향성 분석 및 군집화

3. 통계･역학 모형에 의한 강수예측

3.1 통계 모형에 의한 월강수량 예측

3.1.1 시계열 예측

3.1.2 모형 예측성 평가 방법

3.1.3 예측 결과

3.2 역학 모형에 의한 월 강수량 예측

3.3 해수면 온도지수 예측

4. 통합홍수위험 평가 설문 조사 분석

4.1 Delphi-AHP 분석

4.2 평가지표의 선정

4.3 평가지표 개발

4.4 설문지의 구성

4.5 전문가 집단의 구성

4.6 중･장기 예측 정보 활용성 조사 결과

4.7 Delphi 기법

4.8 AHP 분석방법

4.9 AHP 분석 계층 구조(Hierarch Tree)

4.10 신뢰도 평가 결과

4.11 중요도 평가 결과

5. 통합 홍수 위험도 평가

5.1 위험도 분석 및 평가 구간

5.2 지표별 DB구축 결과

5.2.1 위해성 지표 DB구축 결과

5.2.2 노출성 지표 DB구축 결과

5.2.3 취약성 지표 DB구축 결과

5.3 지표의 표준화

5.4 통합 홍수위험도 평가 결과

5.4.1 1995년 이전과 이후 홍수위험 평가

5.4.2 ENSO 패턴 변화에 따른 홍수위험 평가

5.4.2.1 엘니뇨 시기(CT El Niño)

5.4.2.2 중앙태평양 엘니뇨 시기(WP El Niño)

5.4.2.3 라니냐 시기(La Niña)

5.4.3 통합홍수위험도 예측 결과

6. 중장기 홍수 재난 위험도 관리 및 활용방안

6.1 이음새 없는 예측의 수자원분야 활용

6.2 재난관리의 방향

6.3 홍수발생에 따른 재난관리 및 활용방향

7. 결론 및 토의

REFERENCE

부 록

[부록1] 설문조사서

[부록2] 설문조사 결과 통계분석

[부록3] 지표관련 자료출처

[부록4] 지자체 시군구별 티센 가중치 작성 결과

[부록5] 지자체 홍수위험 지수 표출 결과