Framework for predicting the spatial distribution of green algae blooms utilizing historical GOCI and MODIS imagery

Green algal blooms (GABs) have become a global concern and are particularly serious in the southern Yellow Sea (SYS). Remote sensing (RS) images are the most popular and effective data source for detecting GABs; however, in case of unfavorable weather conditions (e.g., cloud cover), extracting GABs from RS images is impossible. To compensate for the shortage of RS images for GAB detection, this study proposes an effective GAB prediction framework based on cellular automata (CA-GPM) that utilizes historical Geostationary Ocean Color Imager (GOCI) and Moderate Resolution Imaging Spectroradiometer (MODIS) imagery. CA-GPM first extracts the GAB coverage and chlorophyll-a (Chl-a) concentration from the last available RS images and establishes a competitive association between GAB events and Chl-a concentration. This study used structural equation model analysis to evaluate the impact weights of all marine environment data (MED) on GAB growth and drift. Eventually, the cellular automata method was used to predict GABs based on the competition pattern, drift pattern, and growth-and-decline rules. Through the comparison of the CA-GPM results with GABs identified in RS images from 2011 to 2022, the precision, missing alarm rate, and false alarm rate were found to be 0.6909 ± 0.5001, 0.2980 ± 0.6975, and 0.3091 ± 0.5001 respectively. The designed CA-GPM can be effectively applied for emergency management of GABs. In addition, based on CA-GPM, this study revealed the competition, physical, and biological factors affecting GABs in the SYS, with the biological and physical conditions being equally important during 2011-2020. Since 2021, the physical conditions have had a negative effect on GABs. In summary, the proposed method can facilitate the detection of GABs when RS images cannot be used.