Detecting coral bleaching using high-resolution satellite data analysis and 2-dimensional thermal model simulation in the Ishigaki fringing reef, Japan

In 2007, high-temperature-induced mass coral mortality was observed in a well-developed fringing reef area on the southeastern coast of Ishigaki Island, Japan. To analyze the response of the corals to thermal stress, the coral cover was examined using Quickbird data, taken across the reef flat just before and after the bleaching event and performing a reef scale horizontal 2-dimensional thermal model simulation. The Quickbird data consisted of multispectral (MSS) imagery, which had a spatial resolution of 2.4 m, and panchromatic (PAN)-fused multispectral imagery, which had a 0.6-m spatial resolution. The observed changes in coral cover implied that the delineation of partially bleached coral was more precise with PAN + MSS. The classification accuracy achieved using PAN + MSS (93%) was superior to that obtained using MSS (88%). The in situ water temperature observations and 2-dimensional thermal model simulation results indicated that the water temperature fluctuated greatly in the inner reef area in late July 2007. Different thermal stress indices, including daily average temperature, daily maximum excess temperature, and daily accumulated temperature, were examined to define a suitable index that represented the severity of the thermal stress on coral cover. The results suggested that the daily accumulated temperature that occurred during the maximum sea surface temperature period of the bleaching season provided the best predictor of bleaching. The changes in water temperature, bathymetry, and coral patch size affected the severity of bleaching; therefore, the spatial dependence of these variables was examined using Moran’s I and Lagrange multiplier tests. An investigation of the effect of coral patch sizes on coral bleaching indicated that large coral patches were less affected than the small patches, which were more likely to suffer bleaching and coral mortality.