Spatio-Temporal Variation of Diatom Blooms along the Singapore Coasts from Multispectral Imagery
Amihan Yson Manuel (a*), Ryan Tan (a), Chenguang Hou (a), Soo Chin Liew (a), Chee Yew Sandric Leong (b)

a) Centre for Remote Imaging Processing and Sensing, National University of Singapore
*aymanuel[at]nus.edu.sg
b) Tropical Marine Science Institute, National University of Singapore

Abstract

Recent incidences of massive fish kill events and the constant presence of toxic species in Singapore waters have been a rising cause of concern especially for our aquaculture industry. While government and research sectors are intensifying efforts to mitigate the threat of harmful algal blooms, current monitoring methods lack the frequency and spatial coverage needed to help better characterize how these blooms develop and spread through time. Remote sensing addresses this gap by providing a panoramic view of bloom extents with sufficient revisit times to bridge information in between field data collection times. Field surveys were conducted from January 2024 to March 2025 in Singapore coastal waters. Phytoplankton community analyses across monsoon seasons revealed diatoms as the dominant group, driving the blooms observed during the sampling period. In this study, we review and apply different diatom abundance indicators such as combining the backscatter signature with the red band ratio, and other spectral indices to map diatom blooms from Sentinel-2 images. Bio-optical models for Singapore coastal waters were developed from the field survey data that included measurements of absorption and backscattering coefficients. We have good results in the retrieval of inherent optical properties with machine learning algorithms on multispectral images, and we are also currently investigating the feasibility of deriving relative diatom fraction from these results. In addition to providing insights vital for monitoring and early detection of blooms, the results from this study may also aid in identifying diatom bloom hotpots throughout the straits of Singapore.

Keywords: diatoms, algae blooms, machine learning, spectral indices

Topic: Topic B: Applications of Remote Sensing