Sunday, 22 January 2017
4E (Washington State Convention Center )
Dinoflagellate blooms of Karenia brevis (KB) are of great concern along the Gulf of Mexico due to its different levels of toxicity. In order to detect and monitor the occurrence of these blooms ocean color data available by NASA’s Goddard Space Flight Center have for long been studied. In this research, sensor bands of MODIS-Aqua and VIIRS satellite instruments are considered to study the performance of detection techniques consisting of structured algorithms based on reflectance, fluorescence and chlorophyll values. Our final outcome is to evaluate retrieval accuracies of KB blooms. For such, detection techniques were propitious to pixel-based analysis against in-shore and off-shore KB in situ cell counts to study their propensity for false positives and false negatives. Results showed good agreements with the majority of the methods used, but these demonstrated to be more accurate when using VIIRS data over MODIS-A data. However we believe that refined thresholds will lessen retrieval errors and more in situ measurements match-ups will be necessary to make a precise statement about the accuracy of the techniques sampled. As a final component of this research, a comparison between NASA’s OCX algorithm and Neural Networks inverse algorithm (El-habashi, 2016) for in-shore and off-shore waters was also of our interest to better picture spatial arrangement of KB blooms.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner