COMS/GOCI Marine Fog Detection Algorithm Combined with Himawari-8/AHI

Title
COMS/GOCI Marine Fog Detection Algorithm Combined with Himawari-8/AHI
Author(s)
김동희; 박명숙; 박영제; 김원국
KIOST Author(s)
Park, Myung Sook(박명숙)Park, Young Je(박영제)
Alternative Author(s)
김동희; 박명숙; 박영제
Publication Year
2019-07-29
Abstract
Marine fog is one of the most dangerous weather hazards that threaten coastal and marine areas.
However, marine fog monitoring is difficult because of the lack of in-situ observation data. Accordingly,
satellite data has been considered a valuable information for monitoring fog above the ocean. In this
study, COMS/GOCI marine fog detection algorithm combined with Himawari-8/AHI was developed
based on Decision Tree technique. Hourly available and fine visible channel measurement of GOCI and
spatial pattern index were mainly used to detect marine fog in the daytime. In addition, cloud top height
data from Himawari-8 was used as a supplementary data because visible channel alone is not enough to
remove clouds. Decision Tree method was selected for developing our algorithm because the result is
easy to interpret marine fog detection process. The resulting algorithm consists of main marine fog
detection process and two post-processing including cloud removal and marine fog boundary detection.
Post-processing could provide more stable performance than GOCI-only algorithm. Validation was
performed for 2107 and 2018 by comparing the detection results with in-situ visibility data and CALIPSO
Vertical Feature Mask(VFM) data. Results show that marine fog can be well detected with this algorithm
when there is no cloud above the marine fog.
URI
https://sciwatch.kiost.ac.kr/handle/2020.kiost/22545
Bibliographic Citation
AOGS 2019, pp.1, 2019
Publisher
Asia Oceania Geosciences Society
Type
Conference
Language
English
Publisher
Asia Oceania Geosciences Society
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