Detection and Quantification of Coral Bleaching using Sentinel-2 Multispectral Imagery

Abstract

Warming oceans and marine heatwaves disrupt coral-zooxanthellae symbiosis, causing coral bleaching globally. The Indian Ocean has warmed two to three times faster than the Pacific in the past fifty years, thereby leading to an increase in marine heatwaves and subsequent coral bleaching. Given this context, it is critical to study coral bleaching in the Indian Ocean. However, studies on coral bleaching in the Indian Ocean are limited and field surveys are difficult due to the limitations in the accessibility of reefs. This study demonstrates the use of satellite images for detecting bleaching in the Indian Ocean, by using Sentinel-2 satellite images over Kavaratti island as a case study. A workflow for the detection of coral bleaching is first implemented on a test case over the northwestern coast of Hainan island, successfully reproducing the previously reported bleaching patterns of the coral bleaching event that occurred in 2020. Then the workflow is implemented over Kavaratti island to detect the coral bleaching event that occurred in 2024. Coral bleaching is also quantified for the case study region of Kavaratti. The workflow to detect bleaching comprises sun glint correction, pseudo-invariant radiation normalisation, band differencing, and finally creating false composite images from the difference bands. Quantification is done by calculating the percentage of bleached pixels using a suitable percentile as a threshold.

Bleaching patterns, depicted by golden-yellow hues in the false composites, can be observed around Kavaratti island, showing slight bleaching in March 2024, becoming widespread in May in the same year. These bleaching patterns are consistent with field survey observations. The 60th percentile of the percentage change values from each green band of healthy images, with respect to an ‘averaged healthy green band’, over coral reef pixels is found to be a suitable threshold for quantification. The estimated bleaching percentage in March is ~25%, which increases to being ~63% in May, indicating widespread bleaching, along with the patterns seen in the composite images. The bleaching percentage estimated for September is ~26% indicating recovery from bleaching. A temporal progression from slight to peak bleaching and then later recovery is captured here. Visualisation of the spatial extent and intensity of bleaching and estimation of bleaching percentage can be captured using satellite images. This study highlights the potential of satellite imagery for detection and quantification of coral bleaching, which can be used as a complementary tool to field surveys, contributing to improved assessment and management of coral reefs.