Statellite Optical Remote Sensing Of Ulva Prolifera In Coastal Waters Off Qingdao And The Yellow Sea Of China

Similar to red tides, "green tides" (Ulva prolifera blooms) have become an increasingly frequent marine ecological phenomenon in the world. The abnormal aggregation of Ulva prolifera in the Yellow Sea in 2008 and 2009 has drawn extensive attention from national and international research communities and from the public. The work here attempts to conduct research on estimating Ulva prolifera coverage by comparing various methods and satellite data products. First, the spectral characteristics of green tides are investigated. In the near-infrared bands a reflectance peak is observed, which is similar to the spectral properties of terrestrial vegetation. In the short-wave infrared band (1240 nm) the reflectance is reduced due to strong water absorption. Secondly, based on these spectral characteristics, the spatial extent and area coverage of green tides are estimated from MODIS and MERIS data using a new index, namely the floating Ulva prolifera index or floating algae index (FAI). These estimates are validated with limited in-situ data. The results are used directly to support a remote sensing based monitoring system that provides timely information to the management agencies and researchers.. Further, the conventional methods of color composite (RGB images) and normalized difference vegetation index (NDVI) are also evaluated. They are found to contain higher noise than FAI due to changes in the atmosphere and also in observing angles. The ability of the Chinese FY-3A/MERSI in detecting and quantifying Ulva prolifera coverage is also evaluated and compared with that of MODIS. Under cloudfree conditions, their estimates are comparable. However, MERSI can supplement MODIS observations by reducing cloud cover, showing the advantage of the multi-sensor approach.. Finally, Sea Surface Temperature (SST) and chlorophyll a concentration (Chl_a) estimated from satellite measurements are found to correlate with Ulva prolifera coverage, yet interpretation of such correlation requires further research on potential algorithm artifacts.