Abstract
Since the outbreak of a large-scale Ulva prolifera bloom in the Yellow Sea during the Qingdao Olympic Sailing Competition in summer 2008, Ulva blooms have been a marine hazard every summer. Accurate and timely information on Ulva areal coverage and biomass is of critical importance for governmental responses, decision making, and field studies. Previous studies have shown that satellite remote sensing is the most effective method for this purpose, yet Ulva areal coverage has been estimated in different ways with significantly different results. The objective of this paper is to determine the lower and upper bounds (T0 and T1) of algae-containing pixels in Floating Algae Index images with an objective method that accurately estimates the Ulva areal coverage in individual images, and then converts coverage to biomass using a previously established conversion equation. First, a seawater background image, FAIsw, is constructed to determine T0, which varies for different algae patches. Then, T1 is determined from water tank and in situ measurements as well as radiative transfer simulations to account for different sensor configurations, solar/viewing geometry, and atmospheric conditions. Such determined T1 for MODIS 250-m resolution data is validated using concurrent and collocated 2-m resolution WorldView-2 data. Finally, Ulva areal coverage derived from MODIS data using this method are compared with those from the high-resolution data (OLI/Landsat, WFV/GaoFen-1), with a mean relative difference of 9.6%. Furthermore, an analysis of 17 same-day MODIS/Terra and MODIS/Aqua image pairs shows that large viewing angles, atmospheric turbidity, and sunglint can lead to an underestimation of Ulva coverage of up to 45% under extreme conditions.
Highlights
Since the outbreak of a large-scale green tide in the Yellow Sea (YS) during the Qingdao Olympic Sailing Competition in summer 2008, green tides have been a marine hazard every summer in China (China SOA, 2016)
Where S is the Ulva areal coverage in km2; Spixel is the pixel size; αi is Ulva areal density in pixel i; n is the number of algae-containing pixels, T is the total biomass in kg or metric tons, and σ0 is a calibration constant determined from water tank and in situ experiments
This suggests that Ulva coverage calculated from NDVI using a linear unmixing model can be overestimated significantly (Fig. 9b), while FAI and Difference Vegetation Index (DVI) do not suffer from this problem
Summary
Since the outbreak of a large-scale green tide in the Yellow Sea (YS) during the Qingdao Olympic Sailing Competition in summer 2008, green tides have been a marine hazard every summer in China (China SOA, 2016). It is generally accepted that Ulva blooms originate from the Subei Shoal, as evidenced by satellite remote sensing (Hu and He, 2008; Liu et al, 2009), numerical model simulations (Lee et al, 2011), or in situ experiments (Liu et al, 2010; Huo et al, 2013). Subei Shoal is most likely the ‘seed’ source of the Ulva blooms The disposed fragments of Ulva may drift to offshore waters and grow rapidly, forming large-scale blooms in the YS (Liu et al, 2016)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
