Spectral mixture of intertidal marine macroalgae around the island of Helgoland (Germany, North Sea)
详细信息 查看全文
- 作者：Florian Uhl ; Natascha Oppelt ; Inka Bartsch
- 关键词：ASD ; Analytical Spectral Devices ; FOV ; field of view ; AP ; actual proportion ; ASC ; average slope classifier ; Ascnod ; Ascophyllum nodosum ; BS-Plot ; [B]est [S]lopes for unmixing plot ; Chocri ; Chondrus crispus ; CP ; calculated proportion ; Fucser ; Fucus serratus ; Fucves ; Fucus vesiculosus ; Lamdig ; Laminaria digitata ; Masste ; Mastocarpus stellatus ; MDa ; rithmetic mean of the deviations ; MTMF ; Mixture Tuned Matched Filtering ; QA ; overall accuracy of quarter classifier ; QC ; quarter classifier ; R2 ; Pearso
- 刊名：Aquatic Botany
- 出版年：November, 2013
- 年：2013
- 卷：111
- 期：Complete
- 页码：112-124
- 全文大小：2707 K
文摘
The monitoring of emersed and submersed aquatic macrophytes using airborne hyperspectral remote sensing is an innovative method to identify and quantify algal populations at a landscape scale. In general, a high spectral resolution means that one has to accept a lower spatial resolution. This is often considered problematic for the mapping of patchily distributed intertidal macroalgal vegetation. Analysis of sub-pixel information of visually dominant algal species and vegetation units and their respective coverage is therefore of great interest. In this context, it is necessary to analyze the spectral and spatial mixture of macroalgae. This paper presents an experimental laboratory approach to analyze the spectral reflectance characteristics as well as the spectral mixture behaviour of common biomass-dominant macroalgae growing at the intertidal rocky shores of Helgoland (North Sea, Germany). It became evident that unmixing of spectral signatures can hardly be performed between species of the same genus or family due to their strong spectral analogies. At a higher taxonomic level, however, red-, green- and brown algae can be distinguished as groups via derivative analysis as well as with spectral slope calculation between specific wavelengths in the 400-680 nm spectral region. Both methods described the spectral mixing behaviour with high accuracy (Pearson's R2 > 0.8). The newly introduced linear slope unmixing achieved best unmixing results in comparison to two state-of-the-art unmixing approaches.