Adaptive Threshold Model in Google Earth Engine: A Case Study of Ulva prolifera Extraction in the South Yellow Sea, China

An outbreak of Ulva prolifera poses a massive threat to coastal ecology in the Southern Yellow Sea, China (SYS). It is necessity extract its area and monitor development accurately. At present, monitoring by remote sensing imagery mostly based on fixed threshold or artificial visual interpretation for selection, which has large errors. In this paper, an adaptive model Google Earth Engine (GEE) proposed applied U. SYS. The first applies Floating Algae Index (FAI) Normalized Difference Vegetation (NDVI) algorithm preprocessed images then uses Canny Edge Filter Otsu segmentation automatically. Landsat8/OLI Sentinel-2/MSI images, confusion matrix cross-sensor comparison are used evaluate accuracy applicability model. verification results show that extraction FAI higher (R2 = 0.99, RMSE 5.64) better robustness. However, when average cloud cover more than 70% image (based statistical multi-year information), NDVI can algae distributed at edge cloud. When algorithm, it named FAI-COM (model FAI, Filter, thresholding). And NDVI-COM NDVI, Therefore, final generated supplementing basis paper. F1-score extracted above 0.85. spatiotemporal distribution South Sea from 2016 2020 obtained through calculation. Overall, coverage shows decreasing trend over five years. found delay recovery time Porphyra yezoensis culture facilities Northern Jiangsu Shoal manual salvage cleaning-up May among reasons smaller interannual scale 2017 2018.