• 한국멀티미디어학회논문지
• /
• 제22권9호
• /
• pp.1069-1079
• /
• 2019

In this study, we developed a $PM_{10}$ forecasting model using DNN and Membership Function, and improved the forecasting performance. The model predicts the $PM_{10}$ concentrations of the next 3 days in the Seoul area by using the weather and air quality observation data and forecast data. The best model(RM14)'s accuracy (82%, 76%, 69%) and false alarm rate(FAR:14%,33%,44%) are good. Probability of detection (POD: 79%, 50%, 53%), however, are not good performance. These are due to the lack of training data for high concentration $PM_{10}$ compared to low concentration. In addition, the model dose not reflect seasonal factors closely related to the generation of high concentration $PM_{10}$. To improve this, we propose Julian date membership function as inputs of the $PM_{10}$ forecasting model. The function express a given date in 12 factors to reflect seasonal characteristics closely related to high concentration $PM_{10}$. As a result, the accuracy (79%, 70%, 66%) and FAR (24%, 48%, 46%) are slightly reduced in performance, but the POD (79%, 75%, 71%) are up to 25% improved compared with those of the RM14 model. Hence, this shows that the proposed Julian forecast model is effective for high concentration $PM_{10}$ forecasts.

• 대기
• /
• 제16권2호
• /
• pp.125-139
• /
• 2006

Precipitation forecasts from MM5 have been verified for the period 1989-2001 over Yeongdong region to show a tendency of model forecast. We select 57 events which are related with the heavy snowfall in Yeongdong region. They are classified into three precipitation types; mountain type, cold-coastal type, and warm type. The threat score (TS), the probability of detection (POD), and the false-alarm rate (FAR) are computed for categorical verification and the mean squared error (MSE) is also computed for scalar accuracy measures. In the case of POD, warm, mountain, and cold-coastal precipitation type are 0.71, 0.69, and 0.55 in turn, respectively. In aspect of quantitative verification, mountain and cold-coastal type are relatively well matched between forecasts and observations, while for warm type MM5 tends to overestimate precipitation. There are 12 events for the POD below 0.2, mountain, cold-coastal, warm type are 2, 7, 3 events, respectively. Most of their precipitation are distributed over the East Sea nearby Yeongdong region. These events are also shown when there are no or very weak easterlies in the lower troposphere. Even in the case that we use high resolution sea surface temperature (about 18 km) for the boundary condition, there are not much changes in the wind direction to compare that with low resolution sea surface temperature (about 100 km).

• 대한원격탐사학회지
• /
• 제38권6_2호
• /
• pp.1653-1661
• /
• 2022

본 연구에서는 천리안위성 2A호 1일 평균 표층수온영상을 대상으로 합성곱신경망(convolution neural network, CNN) 딥러닝 기법을 적용하여 냉수대 발생 여부를 분류하는 연구를 수행하였다. 이를 위하여, 2019년부터 2022년까지 1,155장의 영상을 사용하였으며, 국립수산과학원 제공 냉수대 발생 주의보 및 경보자료로부터 냉수대 발생 영상과 그 외 영상으로 분류하여 학습을 수행하였다. 학습 결과로 82.5%의 probability of detection (POD)와 54.4%의 false alarm ratio (FAR) 지수를 획득하였다. 오분류 분석을 통해 냉수대 분류에 실패한 경우의 대부분은 구름의 영향 때문이며, 비냉수대를 오분류한 경우의 대부분은 실제 영상에 냉수대가 존재함을 확인하였다.

• International Journal of Reliability and Applications
• /
• 제5권1호
• /
• pp.25-36
• /
• 2004

Ultrasonic inspection system consist of the examination procedures, equipment, and operators. The reliability of nondestructive testing is influenced by the inspection environment, materials and types of defect. It is very difficult to estimate the reliability of NDT due to the various factors. Piping inspection round robin was conducted to quantify the capability of ultrasonic inspection during in-service. In this study, the models for predicting the ultrasonic NDE reliability by logistic model and linear regression model are discussed. The utility of the NDT reliability assessment is verified by the analysis of the data from round robin test with these models.

• 대한토목학회논문집
• /
• 제28권5B호
• /
• pp.485-493
• /
• 2008

한반도 영역을 대상으로 RDAPS모형의 수치예보자료, AWS의 관측강수, 상층기상관측(upper-air sounding)의 관측자료를 이용하여 권역별 강수발생확률을 예측할 수 있는 인공신경망 모형을 제시하였다. 사용된 자료의 기간은 2001년 7, 8월과 2002년 6월로 홍수기를 대상으로 하였다. 500/750/1000 hPa에서의 지위고도, 500-1000 hPa에서의 층후(thickness), 500 hPa에서의 X와 Y방향 바람성분, 750 hPa에서의 X와 Y방향 바람성분, 표면풍속, 500/750 hPa/표면에서의 온도, 평균해면기압, 3시간 누적 강수, AWS관측소에서 관측된 RDAPS모형 실행전의 6시간과 12시간동안의 누적강수, 가강수량, 상대습도등을 신경망의 예측인자로 사용하였다. 신경망의 구조는 3층 MLP(Multi Layer Perceptron)로 구성하여 역전파알고리즘(Back-propagation)을 학습방법으로 사용하였다. 신경망예측결과 한반도전체에 대한 예측성과의 개선은 H가 6.8%상승하였고, 특히 TS와 POD는 각각 99.2%와 148.1% 상승함으로서 강수예측에 대한 신경망모형이 효과적인 도구가 될 수 있음을 확인하였다. KSS 역시 92.8% 개선됨으로서 RDAPS 예측에 비하여 뚜렷이 개선된 결과를 보여주고 있다.

• 천문학회보
• /
• 제42권2호
• /
• pp.61.1-61.1
• /
• 2017

As the application of deep-learning methods has been succeeded in various fields, they have a high potential to be applied to space weather forecasting. Convolutional neural network, one of deep learning methods, is specialized in image recognition. In this study, we apply the AlexNet architecture, which is a winner of Imagenet Large Scale Virtual Recognition Challenge (ILSVRC) 2012, to the forecast of daily solar flare occurrence using the MatConvNet software of MATLAB. Our input images are SOHO/MDI, EIT $195{\AA}$, and $304{\AA}$ from January 1996 to December 2010, and output ones are yes or no of flare occurrence. We consider other input images which consist of last two images and their difference image. We select training dataset from Jan 1996 to Dec 2000 and from Jan 2003 to Dec 2008. Testing dataset is chosen from Jan 2001 to Dec 2002 and from Jan 2009 to Dec 2010 in order to consider the solar cycle effect. In training dataset, we randomly select one fifth of training data for validation dataset to avoid the over-fitting problem. Our model successfully forecasts the flare occurrence with about 0.90 probability of detection (POD) for common flares (C-, M-, and X-class). While POD of major flares (M- and X-class) forecasting is 0.96, false alarm rate (FAR) also scores relatively high(0.60). We also present several statistical parameters such as critical success index (CSI) and true skill statistics (TSS). All statistical parameters do not strongly depend on the number of input data sets. Our model can immediately be applied to automatic forecasting service when image data are available.

• 대한원격탐사학회지
• /
• 제39권6_3호
• /
• pp.1779-1790
• /
• 2023

위성 자료의 성능이 크게 개선됨에 따라 최근에는 위성을 이용하여 광범위한 영역에 대한 실시간 안개 탐지 알고리즘들이 개발되고 있다. 한반도 주변을 관측하는 기상위성 중 관측주기가 10분으로 시간해상도가 가장 우수한 GEO-KOMPSAT-2A/Advanced Meteorological Imager (GK2A/AMI)는 공간해상도가 500 m이다. 반면 GEO-KOMPSAT-2B/Geostationary Ocean Color Imager-II (GK2B/GOCI-II)는 해상도가 250 m지만, 1시간 주기로 관측하고 가시채널만 보유하고 있다. 따라서 본 연구에서는 한반도 주변에서 발생하는 안개를 10분 및 250 m 해상도로 탐지하기 위해 GK2AB 융합 안개 탐지 알고리즘(Fog Detection Algorithm, FDA)인 GK2AB FDA를 개발하였다. GK2AB FDA는 세 파트로 구성된다. 첫 번째로 현업 운용중인 GK2A 안개 탐지 알고리즘(GK2A FDA)으로 10분 및 500 m 해상도로 안개를 탐지한다. 두 번째 단계에서는 두 위성 자료 간 시공간 일치, 태양천정각과 파장역 차이를 보정한 GK2A normalized visible (NVIS)의 10분 변화량을 이용하여 GK2B NVIS를 10분 간격으로 외삽한다. 마지막 단계에서는 외삽된 GK2B NVIS, 태양천정각, GK2A FDA 산출물 등을 입력자료로 기계학습(의사결정나무)을 이용하여 개발된 GK2AB FDA로 지리적위치에 따라 안개를 탐지(250 m, 10분)한다. GK2AB FDA의 훈련에는 6개 사례, 검증에는 4개 사례가 이용되었다. GK2AB FDA의 정량적 검증에는 지상관측 시정, 풍속 그리고 상대습도 자료를 이용하였다. GK2AB FDA는 GK2A FDA에 비해 공간해상도가 4배 증가함에 따라 안개 및 비안개 화소가 보다 자세히 구분되었다. 또한 검증방법에 관계없이 GK2A FDA에 비해 probability of detection (POD)은 높고 Hanssen-Kuiper Skill score (KSS)는 높거나 비슷함을 보여 안개 탐지 수준이 개선된 것으로 보인다. 하지만 일부 사례에서는 GK2AB FDA의 false alarm ratio (FAR)와 Bias가 크게 나타나 안개를 과대탐지하는 문제를 보이고 있다.

• 대기
• /
• 제25권2호
• /
• pp.367-374
• /
• 2015

Korean Peninsula has high potential for occurrence of aviation turbulence. A Korean aviation Turbulence Guidance (KTG) system focused on the Korean Peninsula, named Korean-Peninsula KTG (KP-KTG) system, is developed using the high resolution (horizontal grid spacing of 1.5 km) Local Data Assimilation and Prediction System (LDAPS) of the Korea Meteorological Administration (KMA). The KP-KTG system is constructed first by selection of 15 best diagnostics of aviation turbulence using the method of probability of detection (POD) with pilot reports (PIREPs) and the LDAPS analysis data. The 15 best diagnostics are combined into an ensemble KTG predictor, named KP-KTG, with their weighting scores computed by the values of area under curve (AUC) of each diagnostics. The performance of the KP-KTG, represented by AUC, is larger than 0.84 in the recent two years (June 2012~May 2014), which is very good considering relatively small number of PIREPs. The KP-KTG can provide localized turbulence forecasting in Korean Peninsula, and its skill score is as good as that of the operational-KTG conducting in East Asia.

• 한국멀티미디어학회논문지
• /
• 제24권10호
• /
• pp.1414-1424
• /
• 2021

In this paper, the importance of input factors of a DNN (Deep Neural Network) PM_2.5 forecasting model using LRP(Layer-wise Relevance Propagation) is analyzed, and forecasting performance is improved. Input factor importance analysis is performed by dividing the learning data into time and PM_2.5 concentration. As a result, in the low concentration patterns, the importance of weather factors such as temperature, atmospheric pressure, and solar radiation is high, and in the high concentration patterns, the importance of air quality factors such as PM_2.5, CO, and NO₂ is high. As a result of analysis by time, the importance of the measurement factors is high in the case of the forecast for the day, and the importance of the forecast factors increases in the forecast for tomorrow and the day after tomorrow. In addition, date, temperature, humidity, and atmospheric pressure all show high importance regardless of time and concentration. Based on the importance of these factors, the LRP_DNN prediction model is developed. As a result, the ACC(accuracy) and POD(probability of detection) are improved by up to 5%, and the FAR(false alarm rate) is improved by up to 9% compared to the previous DNN model.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2022년도 학술발표회
• /
• pp.143-143
• /
• 2022

Precipitation plays an essential role in water resources management and disaster prevention. Therefore, the understanding related to spatiotemporal characteristics of rainfall is necessary. Nowadays, highly accurate precipitation is mainly obtained from gauge observation systems. However, the density of gauge stations is a sparse and uneven distribution in mountainous areas. With the proliferation of technology, satellite-based precipitation sources are becoming increasingly common and can provide rainfall information in regions with complex topography. Nevertheless, satellite-based data is that it still remains uncertain. To overcome the above limitation, this study aims to take the strengthens of machine learning to generate a new reanalysis of precipitation data by fusion of multiple satellite precipitation products (SPPs) with gauge observation data. Several machine learning algorithms (i.e., Random Forest, Support Vector Regression, and Artificial Neural Network) have been adopted. To investigate the robustness of the new reanalysis product, observed data were collected to evaluate the accuracy of the products through Kling-Gupta efficiency (KGE), probability of detection (POD), false alarm rate (FAR), and critical success index (CSI). As a result, the new precipitation generated through the machine learning model showed higher accuracy than original satellite rainfall products, and its spatiotemporal variability was better reflected than others. Thus, reanalysis of satellite precipitation product based on machine learning can be useful source input data for hydrological simulations in ungauged river basins.