ABSTRACT In the past decade, floating macroalgae blooms have been increasing on a global scale. Sub-pixel coverage of floating macroalgae in a remote-sensing image is a crucial parameter for the estimation of biomass. In this study, in situ spectra of green macroalgae (Ulva prolifera), brown macroalgae (Sargassum horneri), and sea water were collected, and they were used to simulate the spectra of macroalgae–seawater mixtures in a linear mixing way. Three algae indices, normalized difference of vegetation index (NDVI), difference of vegetation index (DVI), and virtual-baseline reflectance height for floating algae (VB-FAH) derived from the spectra, were examined with the coverage of macroalgae. The results show that all three indices increase monotonically with increasing sub-pixel coverage of macroalgae: VB-FAH and DVI increase linearly, while NDVI shows a logarithmic increase. Based on this characteristic, two sub-pixel coverage models were proposed (i.e., a linear model based on VB-FAH (or DVI) and an exponential model based on NDVI). These models were then applied to the multiple-spectral GaoFen-1 (GF-1, 16 m resolution) satellite image to examine the sub-pixel coverage of green tide in the Yellow Sea caused by the bloom of floating green macroalgae (U. prolifera). The results show that the relative differences between the two models are no more than 5%, indicating good consistency between the two models. Taking into account the sensitivity of these models (or indices) to the coverage of macroalgae, as well as atmospheric and sea surface conditions and their simplicity, we suggest using the linear model based on VB-FAH, DVI, or a similar band-difference index to estimate sub-pixel coverage of floating macroalgae.

中文翻译：

---

一种估计海面漂浮巨藻亚像素覆盖的光谱混合模型

摘要 在过去的十年中，漂浮的大型藻华在全球范围内不断增加。遥感图像中漂浮巨藻的亚像素覆盖率是估算生物量的关键参数。本研究收集了绿色大型藻类 (Ulva prolifera)、棕色大型藻类 (Sargassum horneri) 和海水的原位光谱，并将它们用于以线性混合方式模拟大型藻类-海水混合物的光谱。三种藻类指数，即归一化植被指数差异 (NDVI)、植被指数差异 (DVI) 和来自光谱的漂浮藻类的虚拟基线反射高度 (VB-FAH)，在大型藻类的覆盖范围内进行了检查。结果表明，随着巨藻亚像素覆盖率的增加，所有三个指数均单调增加：VB-FAH和DVI线性增加，而 NDVI 呈对数增长。基于这一特点，提出了两种亚像素覆盖模型（即基于VB-FAH（或DVI）的线性模型和基于NDVI的指数模型）。然后将这些模型应用于多光谱 GaoFen-1（GF-1，16 m 分辨率）卫星图像，以检查由漂浮的绿色大型藻类（U. prolifera ）。结果表明，两个模型之间的相对差异不超过5%，表明两个模型之间具有良好的一致性。考虑到这些模型（或指数）对大型藻类覆盖范围的敏感性，以及大气和海面条件及其简单性，我们建议使用基于 VB-FAH、DVI、