Developing and assessing the inversion model of water color for retrieving the
coastal water quality and the properties of benthic coral reefs.
Hsiao-Wei Chung1, Cheng-Chien Liu2*, Chih-Hua Chang3, Long-Jeng Lee4, Edward Chen5,
Wen-Chang Yang6
1
Institute of Satellite Informatics and Earth Environment, National Cheng Kung University, Tainan,
TAIWAN 701 R.O.C
2
Department of Earth Sciences National Cheng Kung University No 1, Ta-Hsueh Road, Tainan,
TAIWAN 701 R.O.C.E-mail：ccliu88@mail.ncku.edu.tw
3
Department of Environmental Engineering National Cheng Kung University No 1, Ta-Hsueh Road,
Tainan, TAIWAN 701 R.O.C.
4
Instrument Technology Research Center, National Applied Research Laboratories 20 R&D Road VI,
Hsinchu Science Park, Hsinchu 300, Taiwan
5
Ocean Exploration Division, Taiwan Ocean Research Institute, National Applied Research
Laboratories
6
Taiwan Ocean Research Institute, National Applied Research Laboratories
Abstract
Coral reefs prefer to reside in warm, clean, clear waters with high oxygen contents.
Any deterioration of environment would pose fatal threat to the living of coral reefs.
Therefore, coral reefs serve as an important indicator of environment. Kenting
National Park enjoys the most abundant resources of coral reefs in Taiwan area.
However, the recent extreme weather event, such as Typhoon Morakot in 2009, had
caused 50% of coral reefs destroyed in this area. The technique of water color remote
sensing is promising in assessing the status of coral reefs at both high spatial and high
temporal resolutions. To retrieve the water quality and the properties of benthic coral
reefs directly from the water color signal, however, required a robust algorithm that
has been validated against a comprehensive dataset of in situ measurements or model
simulations. In this research, we improve the GA-SA(genetic algorithm and
semi-analytical) model by taking the properties of benthic coral reefs into account,
with the intention to classify the bottom into six different types, including sea grass,
algae (green algae, red algae and brown algae), coral reefs, and sand. The spectral
library of bottom reflectance is established from the in situ data measured in Kenting
National Park or simulated by the HydroLight radiative transfer model. Our new
model is able to iterate for the optimized solution of the water quality and the property
of benthic coral reefs from the input of bottom reflectance spectrum. These solutions
are then compared to the conditions of water quality and benthic coral reefs property,
under which the bottom reflectance spectra are measured/simulated. The results
demonstrated that our new model is able to achieve a satisfied accuracy to as high as
80%. This new model would be employed to process the data collected by the
shipborne hyperspectral scanner to assess the status of coral reefs in Kenting National
Park.