Multi-Temporal Satellite Images with Bathymetry Correction For Mapping and Assessing Seagrass Bed Changes In Dongsha Atoll / 應用遙測影像之水深校正於東沙環礁海草棲地變遷

碩士 / 國立中央大學 / 遙測科技碩士學位學程 / 104 / In the benthic environment, seagrass was considered as the ecological stabilization index of the marine habitat. Because of the low cost and frequent observations for multispectral satellite images, they are suitable to assess seagrass habitat change. To investigate bottom material directly from spectral information, sun glint, atmospheric and water depth correction are necessary.
In this study, the sun glint and water depth correction will be conducted first to remove the surface and bathymetric effects. Meanwhile, multi Retinex with color restoration method technique which integrate Homomorphic filter and white balance enhancement are implemented. After the processes, image dynamic range is improved which achieve atoll enhancement and cloudy pixel attenuation. In Order to minimize atmospheric effect between images, relative calibration Whitening and De Whitening method is also implemented among multi-temporal image by matching target image with reference image in spectrum statically. Then, water depth correction is applied by pairs band of habitat bottom reflectance.
A huge event of seagrass disappearance has been observed in summer 2014 and 2015. Study area 1 reduced about 12 km2 within three months also found study area 2 lose 3 km2 seagrass habitat. Since most habitat and atoll attenuation condition has been changed, depth invariant index from Lyzenga equation might need recent bathymetry information. In future work, we will try to analysis satellite sea surface temperature product relationship when occur anomaly sea water and extreme weather influence at Dongsha marine resource. .

Identiferoai:union.ndltd.org:TW/104NCU05105110
Date January 2016
CreatorsChih-Yuan Lin, 林致遠
Contributors任玄
Source SetsNational Digital Library of Theses and Dissertations in Taiwan
Languageen_US
Detected LanguageEnglish
Type學位論文 ; thesis
Format70

Page generated in 0.0091 seconds