Chlorophyll concentration forecasts during tropical cyclones using satellite remote sensing imagery

Phytoplankton pigment concentrations such as chlorophyll-a (Chl-a) provide a measure of the biological state of the surface ocean. The Chl-a concentration, a proxy for phytoplankton abundance, is a valuable indicator of the marine ecosystem, and satellite remote sensing is the only way at present to take frequent measurements of Chl-a at regional and ocean-basin scales. Tropical cyclones when passing over land may have devastating effects on human lives, but over the ocean they can strongly enhance another form of life-ocean primary (phytoplankton) production. Researchers indicated that the passing of typhoon in the open ocean can induce the decreasing of sea surface temperature and Chl-a concentration increasing. Meanwhile, the typhoon-induced SST and Chl-a changes is related to the typhoon intensity, moving speed, and days of influence on the ocean. This study adopted the machine learning algorithms in forecasting the Chl-a concentrations from 1- to 5-day ahead by using datasets from the climatologic characteristics of typhoons from Central Weather Bureau (CWB), the typhoon path from Joint Typhoon Warning Center (JTWC), and the satellite observations from Moderate Resolution Imaging Spectroradiometer (MODIS) during typhoon attacks. The study collected the typhoon tracks from JTWC, the typhoon climatologic data issued by CWB, and the atmosphere and ocean color products from MODIS/Aqua satellites. The Chl-a concentration forecast models are constructed by machine learning, namely, multilayer perceptron neural networks and classification and regression tree. The study area was the Taiwan Strait which is between Taiwan and China. This study collected a total of 36 typhoon events affecting the Taiwan Strait over years 2002-2014. The results showed that the proposed methodology was promising to improve the typhoon Chl-a forecast efficiency by prediction models using the typhoon path, climatological, and MODIS/Aqua remote sensing data.