• Proceedings of the Korea Water Resources Association Conference
• /
• 2011.05a
• /
• pp.72-76
• /
• 2011

본 연구는 레이더 강우 영상에서 추출된 강우 패턴을 인공신경망으로 처리하여 단기 강우 예측을 수행하는 방안을 제시한 것이다. 본 연구에 활용한 CAPPI 영상자료로는 편차 보정과 품질 관리가 이루어지고 있으며 획득이 용이한 기상청 자료를 이용하였으며 CAPPI의 PNG 영상으로부터 강우 패턴을 추출하고, 이를 역전파 알고리즘의 인공신경망 강우 예측 모형에 학습시켜 단기 강우를 예측하기 위한 절차를 제시하였다. 이를 위하여 강우의 시공간적 변화 패턴 추출을 위한 영상 처리와 GIS 자료처리 기법을 제시하였고 이를 인공신경망의 단기 강우 예측 학습과 검증에 적용하여 본 연구에서 제시된 기법의 타당성을 검토하였다.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.128-128
• /
• 2022

최근 호우의 빈도와 규모는 증가하는 추세이며 이에 따른 홍수 피해는 많은 피해를 야기하고 있다. 이러한 관점에서 홍수 피해에 대한 선제적 대응을 위한 요소로써 초단시간 강우예측 정보의 중요성은 매우 높다. 특히, 레이더 자료 기반의 강우예측은 수치예보모델과 비교하여 3시간 이내의 짧은 선행시간 이내의 높은 정확도를 갖고 있어 홍수예보에 다수 활용되고 있다. 최근에는 강우자료의 복잡한 관계와 특징을 고려하기 위해 딥러닝 기반의 강우예측 활용 사례가 증가하고 있으나 국내 적용 사례는 적어 관련 연구가 요구되는 실정이다. 본 연구에서는 레이더 강우를 활용한 딥러닝 기반의 강우예측 기법을 제안하고 이에 대한 적용성을 평가하고자 한다. 2차원 레이더 강우자료의 특징과 시계열 특성을 고려하기 위한 심층신경망 구조를 제안하였으며 기존 딥러닝 모형과의 비교를 통해 활용 가능성을 제시하고자 하였다. 적용 대상지역은 한강 유역으로 선정하였다. 정성적 평가를 위해 임계성공지수(CSI)를 활용하여 예측 강우에 대한 정확도를 평가하였으며 정량적 평가를 위해 예측 강우와 관측 강우의 상관관계를 분석하였다. 평가 결과, 제안하는 방법이 기존 모형과 비교하여 예측오차의 범위가 적고 강우의 위치 변화를 잘 반영하는 것으로 나타났다. 본 연구결과는 초단기간 강우예측 자료를 활용하는 홍수예보의 정확도 향상에 기여할 것으로 기대된다.

• Journal of Korea Water Resources Association
• /
• v.53 no.12
• /
• pp.1159-1172
• /
• 2020

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.

• Journal of Wetlands Research
• /
• v.13 no.2
• /
• pp.189-197
• /
• 2011

This paper reviewed application of data-driven method, distance-weighted method(IDWM, IEWM, CCWM, ANN), and radar data method estimated of missing raifall data. To evaluate these methods, statistics was compared using radar and station rainfall data from Imjin-river basin. The range of RMSE values calculated for CCWM, ANN was 1.4 to 1.79mm, and the range of RMSE values estimated data used for radar rainfall data was 0.05 to 2.26mm. Spatial characteristics is considered to Radar rainfall data rather than station rainfall data. Result suggest that estimated data used for radar data can impove estimation of missing raifall data.

• Journal of Korea Water Resources Association
• /
• v.42 no.12
• /
• pp.1039-1052
• /
• 2009

Radar measurement with high temporal and spatial resolutions can be a valuable source of data, especially in the areas where rain gauge installment is not practical. However, this kind of data brings with it many errors. The objective of this paper is to propose a method to evaluate statistically the quantitative and qualitative accuracy at different radar ranges, temporal intervals and raingage densities and use a bias adjustment technique to improve the quality of radar rainfall for the purpose of hydrological application. The method is tested with the data of 2 storm events collected at Jindo (S band) and Kwanak (C band) radar stations. The obtained results show that the accuracy of radar rainfall estimation increases when time interval rises. Radar data at the shorter range seems to be more accurate than the further one, especially for C-band radar. Using the Monte Carlo simulation experiment, we find out that the sampling error of the bias between radar and gauge rainfall reduces nonlinearly with increasing raingage density. The accuracy can be improved considerably if the real-time bias adjustment is applied, making adjusted radar rainfall to be adequately good to apply for hydrological application.

• Journal of Korea Water Resources Association
• /
• v.56 no.8
• /
• pp.471-484
• /
• 2023

Deep learning models based on generative adversarial neural networks are specialized in generating new information based on learned information. The deep generative models (DGMR) model developed by Google DeepMind is an generative adversarial neural network model that generates predictive radar images by learning complex patterns and relationships in large-scale radar image data. In this study, the DGMR model was trained using radar rainfall observation data from the Ministry of Environment, and rainfall prediction was performed using an generative adversarial neural network for a heavy rainfall case in August 2021, and the accuracy was compared with existing prediction techniques. The DGMR generally resembled the observed rainfall in terms of rainfall distribution in the first 60 minutes, but tended to predict a continuous development of rainfall in cases where strong rainfall occurred over the entire area. Statistical evaluation also showed that the DGMR method is an effective rainfall prediction method compared to other methods, with a critical success index of 0.57 to 0.79 and a mean absolute error of 0.57 to 1.36 mm in 1 hour advance prediction. However, the lack of diversity in the generated results sometimes reduces the prediction accuracy, so it is necessary to improve the diversity and to supplement it with rainfall data predicted by a physics-based numerical forecast model to improve the accuracy of the forecast for more than 2 hours in advance.

• Magazine of the Korean Society of Hazard Mitigation
• /
• v.5 no.2 s.17
• /
• pp.82-88
• /
• 2005

• Magazine of the Korean Society of Hazard Mitigation
• /
• v.5 no.3 s.18
• /
• pp.54-60
• /
• 2005

• The Magazine of the IEIE
• /
• v.40 no.2
• /
• pp.72-79
• /
• 2013

• The Magazine of the IEIE
• /
• v.40 no.2
• /
• pp.32-40
• /
• 2013