There are no files associated with this item.
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.citation.startPage | 113814 | - |
dc.citation.title | REMOTE SENSING OF ENVIRONMENT | - |
dc.citation.volume | 298 | - |
dc.contributor.author | Kang, Yoojin | - |
dc.contributor.author | Sung, Taejun | - |
dc.contributor.author | Im, Jungho | - |
dc.date.accessioned | 2023-12-19T11:13:30Z | - |
dc.date.available | 2023-12-19T11:13:30Z | - |
dc.date.created | 2023-10-31 | - |
dc.date.issued | 2023-12 | - |
dc.description.abstract | As the majority of active fire detection algorithms have been developed for worldwide applications using only satellite data without considering observing conditions and environmental factors, their performance varies regionally. This study investigates the viability of an adaptable active fire detection model that is applicable to diverse environmental and observing conditions by fusing numerical model data and satellite images. The model was developed for various land cover and climate types using commonly utilized brightness temperature-related variables (key variables) and supporting variables (sub-variables), including solar zenith angle, satellite zenith angle (SAZ), relative humidity (RH), and skin temperature. A dual-module (DM) convolutional neural network (CNN) structure was adopted to consider the different properties of key variables and sub-variables, and a control without sub-variables was used to assess the impact of observing and environmental variables. The proposed model was further evaluated using existing polar-orbiting and geostationary satellite-based active fire products. The recall and precision of the control model were 0.80 and 0.98, respectively, and the standard deviation of recall for the five focus sites was 0.140. However, the DM CNN model was notable for its higher recall and robustness compared to the control model (recall of 0.84, precision of 0.97, and standard deviation of recall of 0.126). High RH and SAZ, and the day-night transition period contributed to the poor performance of the control model which was mitigated by the DM CNN model. In particular, the use of RH improved the recall of the model, and SAZ contributed to the reduction of performance variation. Our model also outperformed the two geostationary satellite-based active fire products in terms of detection capacity, resulting in a spatial distribution of active fires similar to that of polar-orbiting satellite-based active fire products. | - |
dc.identifier.bibliographicCitation | REMOTE SENSING OF ENVIRONMENT, v.298, pp.113814 | - |
dc.identifier.doi | 10.1016/j.rse.2023.113814 | - |
dc.identifier.issn | 0034-4257 | - |
dc.identifier.scopusid | 2-s2.0-85171187176 | - |
dc.identifier.uri | https://scholarworks.unist.ac.kr/handle/201301/66069 | - |
dc.identifier.wosid | 001080329300001 | - |
dc.language | 영어 | - |
dc.publisher | ELSEVIER SCIENCE INC | - |
dc.title | Toward an adaptable deep-learning model for satellite-based wildfire monitoring with consideration of environmental conditions | - |
dc.type | Article | - |
dc.description.isOpenAccess | FALSE | - |
dc.relation.journalWebOfScienceCategory | Environmental Sciences; Remote Sensing; Imaging Science & Photographic Technology | - |
dc.relation.journalResearchArea | Environmental Sciences & Ecology; Remote Sensing; Imaging Science & Photographic Technology | - |
dc.type.docType | Article | - |
dc.description.journalRegisteredClass | scie | - |
dc.description.journalRegisteredClass | scopus | - |
dc.subject.keywordAuthor | Active fire detection | - |
dc.subject.keywordAuthor | Convolutional neural network | - |
dc.subject.keywordAuthor | Robust to environmental changes | - |
dc.subject.keywordAuthor | Geostationary satellite | - |
dc.subject.keywordAuthor | Numerical and satellite data fusion | - |
dc.subject.keywordPlus | ACTIVE FIRE DETECTION | - |
dc.subject.keywordPlus | ADVANCED HIMAWARI IMAGER | - |
dc.subject.keywordPlus | DETECTION ALGORITHM | - |
dc.subject.keywordPlus | MODIS | - |
dc.subject.keywordPlus | VALIDATION | - |
dc.subject.keywordPlus | PRODUCT | - |
dc.subject.keywordPlus | AFRICA | - |
dc.subject.keywordPlus | ASTER | - |
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Tel : 052-217-1404 / Email : scholarworks@unist.ac.kr
Copyright (c) 2023 by UNIST LIBRARY. All rights reserved.
ScholarWorks@UNIST was established as an OAK Project for the National Library of Korea.