레이더 기반 도시지역 돌발성 호우의 위험성 사전 예측 : 수도권지역 사례 연구

윤성심,나카키타 에이이치,니시와키 류타,사토 히로토,Yoon, Seongsim,Nakakita, Eiichi,Nishiwaki, Ryuta,Sato, Hiroto 한국수자원학회 2016 한국수자원학회논문집 Vol.49 No.9

최근 빈번히 발생하는 도시지역에서의 돌발성 집중호우로 인한 피해를 저감하고자, 기상레이더를 통해 관측되는 자료를 바탕으로 돌발성 호우의 위험성을 사전에 예측하는 기법을 적용하였다. 본 연구에서 활용한 방법은 대기 중의 돌발성 호우를 유발할 수 있는 적란운 대류세포의 조기탐지, 탐지된 대류세포의 자동 추적, 해당 대류세포가 발달하여 돌발성 호우를 유발할 수 있는 가능성을 판단하는 위험예측이라는 3가지 단계를 결합한 것이다. 본 기법은 실제 돌발성 호우로 인해 수도권 지역 소하천에서 시민들이 고립된 사례를 포함한 집중호우 사례에 적용되었다. 그 결과, 레이더 자료만을 이용하여 지상관측망보다 사전에 강우세포를 탐지하고, 국지적 집중호우로 발달하는 현상을 위험도로 판단할 수 있음을 보여 주었다. 본 연구를 통해 제시된 위험도 예측결과를 도시 소하천 홍수대피 업무에 활용한다면 대피시간을 충분히 확보할 수 있어 인명사고를 줄이는 데 기여할 수 있을 것으로 사료된다. The aim of this study is to apply and to evaluate the radar-based risk prediction algorithm for damage reduction by sudden localized heavy rain in urban areas. The algorithm is combined with three processes such as "detection of cumulonimbus convective cells that can cause a sudden downpour", "automatic tracking of the detected convective cells", and "risk prediction by considering the possibility of sudden downpour". This algorithm was applied to rain events that people were marooned in small urban stream. As the results, the convective cells were detected through this algorithm in advance and it showed that it is possible to determine the risk of the phenomenon of developing into local heavy rain. When use this risk predicted results for flood prevention operation, it is able to secure the evacuation time in small streams and be able to reduce the casualties.

서울시 고밀도 지상강우자료 품질관리방안 도출

윤성심,이병주,최영진,Yoon, Seongsim,Lee, Byongju,Choi, Youngjean 한국수자원학회 2015 한국수자원학회논문집 Vol.48 No.4

고해상도의 정량적 실황강우장을 산정하기 위해서는 양질의 고밀도 강우관측망 정보가 필요하다. 이를 위해 본 연구에서 정량적 실황강우장 산정을 위한 입력자료로 SK 플래닛의 고밀도 복합기상센서 관측망과 기존 기상청 관측망을 이용하고자 하였다. 이를 위해 서울지역에 위치한 SK 플래닛의 복합기상센서 관측망을 소개하고, 2013년 7~9월 3개월 동안의 관측자료의 품질을 분석하였다. 품질분석 결과, SK 플래닛 관측소가 일부 관측소를 제외하고 대부분 기존 관측망과 유사하게 강우를 관측하는 것을 확인할 수 있었다. 다만, 일시적인 기계 및 자료 전송 오류로 인해 발생할 수 있는 결측치 및 이상치가 미치는 영향을 최대한 저감하기 위해서 오자료를 실시간으로 보정할 수 있는 품질보정 기법을 개발하였으며, 개발된 기법이 적절히 강우를 보정하는 것을 확인하였다. 이를 통해 결측률이 20% 미만이면서 오자료의 영향이 최소가 되는 190개소(기상청 34개소, SK 플래닛 156 개소)를 정량적 실황강우장 산정에 활용하였다. 또한, 약 $3km^2$의 밀도를 갖는 고해상도 관측망을 이용하여 산정된 강우분포장의 재현성을 기존 기상청 관측망의 결과비교를 통해 평가한 결과, 고밀도 관측망을 통해 산정된 강우분포장의 빈도곡선이 레이더 공간분포장과 유사하며, 기존 기상청 관측망의 공백을 보완할 수 있음을 확인하였다. 특히, 이 결과를 통해 고밀도의 강우관측 결과를 활용한다면 레이더 참강우장에 근사한 공간분포된 강우를 산정할 수 있다는 것을 확인할 수 있었다. This study used high density network of integrated meteorological sensor, which are operated by SK Planet, with KMA weather stations to estimate the quantitative precipitation field in Seoul area. We introduced SK Planet network and analyzed quality of the observed data for 3 months data from 1 July to 30 September 2013. As the quality analysis result, we checked most SK Planet stations observed similar with previous KMA stations. We developed the real-time quality check and adjustment method to reduce the error effect for hydrological application by missing and outlier value and we confirmed the developed method can be corrected the missing and outlier value. Through this method, we used the 190 stations(KMA 34 stations, SK Planet 156 stations) that missing ratio is less than 20% and the effect of the outlier was the smallest for quantitative precipitation estimation. Moreover, we evaluated reproducibility of rainfall field high density rain gauge network has $3km^2$/gauge. As the result, the spatial relative frequency of rainfall field using SK Planet and KMA stations is similar with radar rainfall field. And, it supplement the blank of KMA observation network. Especially, through this research we will take advantage of the density of the network to estimate rainfall field which can be considered as a very good approximation of the true value.

심층신경망을 이용한 레이더 영상 학습 기반 초단시간 강우예측

윤성심,박희성,신홍준,Yoon, Seongsim,Park, Heeseong,Shin, Hongjoon 한국수자원학회 2020 한국수자원학회논문집 Vol.53 No.12

This study applied deep convolution neural network based on U-Net and SegNet using long period weather radar data to very short-term rainfall prediction. And the results were compared and evaluated with the translation model. For training and validation of deep neural network, Mt. Gwanak and Mt. Gwangdeoksan radar data were collected from 2010 to 2016 and converted to a gray-scale image file in an HDF5 format with a 1km spatial resolution. The deep neural network model was trained to predict precipitation after 10 minutes by using the four consecutive radar image data, and the recursive method of repeating forecasts was applied to carry out lead time 60 minutes with the pretrained deep neural network model. To evaluate the performance of deep neural network prediction model, 24 rain cases in 2017 were forecast for rainfall up to 60 minutes in advance. As a result of evaluating the predicted performance by calculating the mean absolute error (MAE) and critical success index (CSI) at the threshold of 0.1, 1, and 5 mm/hr, the deep neural network model showed better performance in the case of rainfall threshold of 0.1, 1 mm/hr in terms of MAE, and showed better performance than the translation model for lead time 50 minutes in terms of CSI. In particular, although the deep neural network prediction model performed generally better than the translation model for weak rainfall of 5 mm/hr or less, the deep neural network prediction model had limitations in predicting distinct precipitation characteristics of high intensity as a result of the evaluation of threshold of 5 mm/hr. The longer lead time, the spatial smoothness increase with lead time thereby reducing the accuracy of rainfall prediction The translation model turned out to be superior in predicting the exceedance of higher intensity thresholds (> 5 mm/hr) because it preserves distinct precipitation characteristics, but the rainfall position tends to shift incorrectly. This study are expected to be helpful for the improvement of radar rainfall prediction model using deep neural networks in the future. In addition, the massive weather radar data established in this study will be provided through open repositories for future use in subsequent studies.

산지지역 수재해 대응을 위한 레이더 기반 돌발성 호우 위험성 사전 탐지 기술 적용성 평가

윤성심,손경환,Yoon, Seongsim,Son, Kyung-Hwan 한국수자원학회 2020 한국수자원학회논문집 Vol.53 No.4

There is always a risk of water disasters due to sudden storms in mountainous regions in Korea, which is more than 70% of the country's land. In this study, a radar-based risk prediction technique for sudden downpour is applied in the mountainous region and is evaluated for its applicability using Mt. Biseul rain radar. Eight local heavy rain events in mountain regions are selected and the information was calculated such as early detection of cumulonimbus convective cells, automatic detection of convective cells, and risk index of detected convective cells using the three-dimensional radar reflectivity, rainfall intensity, and doppler wind speed. As a result, it was possible to confirm the initial detection timing and location of convective cells that may develop as a localized heavy rain, and the magnitude and location of the risk determined according to whether or not vortices were generated. In particular, it was confirmed that the ground rain gauge network has limitations in detecting heavy rains that develop locally in a narrow area. Besides, it is possible to secure a time of at least 10 minutes to a maximum of 65 minutes until the maximum rainfall intensity occurs at the time of obtaining the risk information. Therefore, it would be useful as information to prevent flash flooding disaster and marooned accidents caused by heavy rain in the mountainous area using this technique.

하천 수위예측을 위한 레이더와 수위 관측자료를 이용한 멀티모달 딥러닝 모형 적용

윤성심(Yoon, Seong Sim),김세용(Kim, Seyong),배상민(Bae, Sangmin) 대한토목학회 2023 대한토목학회 학술대회 Vol.2023 No.10

후회없는 선택

윤성심(Seong-Sim Yoon) 대한토목학회 2014 대한토목학회지 Vol.62 No.2

청계천 홍수예측 시스템 구축

윤성심(Yoon Seong-sim),배덕효(Bae Deg-hyo),정창삼(Jeong Chang Sam),이경도(Lee Kyuong Do),이종대(Lee Jong Dae) 대한토목학회 2009 대한토목학회 학술대회 Vol.2009 No.10

고밀도 지상강우관측망을 활용한 서울지역 정량적 실황강우장 산정

윤성심(Seong-sim Yoon),이병주(Byongju Lee),최영진(Youngjean Choi) 한국기상학회 2015 대기 Vol.25 No.2

For urban flash flood simulation, we need the higher resolution radar rainfall than radar rainfall of KMA, which has 10 min time and 1km spatial resolution, because the area of subbasins is almost below 1 km2. Moreover, we have to secure the high quantitative accuracy for considering the urban hydrological model that is sensitive to rainfall input. In this study, we developed the quantitative precipitation estimation (QPE), which has 250 m spatial resolution and high accuracy using KMA AWS and SK Planet stations with Mt. Gwangdeok radar data in Seoul area. As the results, the rainfall field using KMA AWS (QPE1) is showed high smoothing effect and the rainfall field using Mt. Gwangdeok radar is lower estimated than other rainfall fields. The rainfall field using KMA AWS and SK Planet (QPE2) and conditional merged rainfall field (QPE4) has high quantitative accuracy. In addition, they have small smoothedarea and well displayed the spatial variation of rainfall distribution. In particular, the quantitative accuracy of QPE4 is slightly less than QPE2, but it has been simulated well the non-homogeneity of the spatial distribution of rainfall.

도시홍수재해해석을 위한 중랑천 시험유역 운영

윤성심(Yoon Seong Sim),이경도(Lee Kyoung Do),김정회(Kim Jeong Hoi),배덕효(Bae Deg Hyo),이종태(Lee Jong Tae) 대한토목학회 2004 대한토목학회 학술대회 Vol.2004 No.10

수문관측 기반의 청계천 홍수예측모델 개발 연구

윤성심(Yoon Seong Sim),정창삼(Jeong Chang Sam),배덕효(Bae Deg Hyo) 대한토목학회 2008 대한토목학회 학술대회 Vol.2008 No.10