A Study on the GK2A/AMI Image Based Cold Water Detection Using Convolutional Neural Network SCOPUS KCI

Title
A Study on the GK2A/AMI Image Based Cold Water Detection Using Convolutional Neural Network
Alternative Title
합성곱신경망을 활용한 천리안위성 2A호 영상 기반의 동해안 냉수대 감지 연구
Author(s)
Park, Sung Hwan; Kim, Dae Sun; Kwon, Jae Il
KIOST Author(s)
Park, Sung Hwan(박숭환)Kim, Dae Sun(김대선)Kwon, Jae Il(권재일)
Alternative Author(s)
박숭환; 김대선; 권재일
Publication Year
2022-12
Abstract
In this study, the classification of cold water and normal water based on Geo-Kompsat 2A images was performed. Daily mean surface temperature products provided by the National Meteorological Satellite Center (NMSC) were used, and convolution neural network (CNN) deep learning technique was applied as a classification algorithm. From 2019 to 2022, the cold water occurrence data provided by the National Institute of Fisheries Science (NIFS) were used as the cold water class. As a result of learning, the probability of detection was 82.5% and the false alarm ratio was 54.4%. Through misclassification analysis, it was confirmed that cloud area should be considered and accurate learning data should be considered in the future.

본 연구에서는 천리안위성 2A호 1일 평균 표층수온영상을 대상으로 합성곱신경망(convolution neural network, CNN) 딥러닝 기법을 적용하여 냉수대 발생 여부를 분류하는 연구를 수행하였다. 이를 위하여, 2019 년부터 2022년까지 1,155장의 영상을 사용하였으며, 국립수산과학원 제공 냉수대 발생 주의보 및 경보자료로 부터 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 실패 한 경우의 대부분은 구름의 영향 때문이며, 비냉수대를 오분류한 경우의 대부분은 실제 영상에 냉수대가 존재 함을 확인하였다.
ISSN
1225-6161
URI
https://sciwatch.kiost.ac.kr/handle/2020.kiost/43801
DOI
10.7780/kjrs.2022.38.6.2.7
Bibliographic Citation
Korean Journal of Remote Sensing, v.38, no.6, pp.1653 - 1661, 2022
Publisher
대한원격탐사학회
Keywords
Geo-Kompsat 2A; Cold water; Upwelling; Korean peninsula; Deep learning; Convolution neural network
Type
Article
Language
Korean
Files in This Item:
There are no files associated with this item.

qrcode

Items in ScienceWatch@KIOST are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse