ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume X-5-2024
https://doi.org/10.5194/isprs-annals-X-5-2024-165-2024
https://doi.org/10.5194/isprs-annals-X-5-2024-165-2024
11 Nov 2024
 | 11 Nov 2024

Coral Bleaching Detection Using PRISMA Hyperspectral Satellite Imagery in Calatagan, Batangas

Francesca Deighl Rivera, Ginell Elyza Buenavista, Ayin Tamondong, and Ariel Blanco

Keywords: coral bleaching, PRISMA, hyperspectral satellite image, SAM, SST

Abstract. Coral bleaching arises due to stress from warming seas and pollution. Several studies on bleaching detection utilize multispectral imagery, due to data availability. In this study, the usage of PRISMA hyperspectral data imagery was assessed for coral bleaching detection in Calatagan, Batangas, and sea surface temperature (SST) for confirming bleaching occurrence. In-situ measurements were gathered for spectral data of live and bleached corals, and a collaborative qualitative data interview with a coral bleach patroller and municipal office was conducted to gain further information about bleaching details in the area. Sentinel-2 L2A was also used for comparison during classification. Spectral Angle Mapper (SAM) was utilized for image classification, while using field data and data from related studies for spectral library. A total of 27.12% relative to the coral reef area was classified as bleached corals. T-test statistics were done to analyze the live and bleached corals, which exhibited that only two (2) bands in the green region were found to be not separable. Using box plots, bleached corals showed to have generally lower reflectance values than live corals for RGB region, as compared to Sentinel, with bleached having higher values, which implies that spatial resolution can affect the spectral response of different features. Using ground truth data, 60.6% user’s accuracy for bleached corals, and 63.4% overall accuracy is achieved. Hence, hyperspectral satellite imagery can be utilized in coral bleaching detection, despite its limitations. Depth variables and diverse sample points are recommended for the betterment of coral bleaching detection.