Mapping the bathymetry of Bonaire through the use of satellite data

By Kasper Slagter

Abstract
Bonaire is home to a wide range of biodiversity, and especially in the shallow coastal waters where coral reefs occur. Coral reefs provide many important ecosystem services and should therefore be protected. However, they are threatened due to many causes like global warming and diseases. Therefore, knowledge about their habitat, the shallow coastal waters, is crucial in order to ensure the conservation of this organism. This knowledge is attainable through the use of satellite data, which is called satellite-derived bathymetry (SDB). The basic principle behind SDB is the relationship between the attenuation of radiance, on one hand, and the depth and wavelength in the water column on the other. This research aims to investigate the possibilities of satellite-derived bathymetry for the island of Bonaire. Furthermore, it explores which method achieves the most accurate bathymetric models and to what extent accurate estimation of bathymetry is possible.

For this research, the bathymetry was calculated with an empirical approach that makes use of in-situ measurements and a ratio between the green and blue band. However, in order to be able to apply this formula, the data first had to be preprocessed. These preprocessing steps included the masking of land/clouds and a sun glint correction. The masking was done through thresholds of reflectance values in the visible bands. The masked images were then deglinted. After deglinting, the data was ready for the calculation of the bathymetry. This was done using two formulas: SDBA and SDBB. The formula of SDBA was calibrated using all in-situ depths less than 30 metres whereas SDBB was calibrated using all depths less than 20 metres.

The results were validated by determining the Root Mean Square Error (RMSE) for the different depth classes: 1-10m, 10-20m, and 20-30m. However, since this research mainly focuses on shallow coastal waters, especially the 1-10m depth class was interesting. The results showed that the models created with SDBB were more accurate for depth class 1-10m, compared to the results with SDBA. The average RMSE of depth class 1-10m for the most accurate method was 4.11m. The most accurate bathymetric model had an RMSE of 3.58m for depth class 1-10m. However, the RMSEs for the other depth classes showed that this method is not applicable for accurate results in deeper depth classes. This research showed that there are possibilities for the island of Bonaire regarding satellite-derived bathymetry. However, more research needs to be done in order to create more accurate results and be able to circumvent limitations.