Title: Understading Ocean Chlorophyll-a from Ocean Color Remote Sensing
Abstract:
We have developed a new machine learning model to estimate Chlorophyll-a concentration the in the global waters with four ocean color remote sensing wavebands that are present in most existing ocean color sensors. The Multi-Layer Perceptron Neural Network (MLPNN) was trained on NASA’s NOMAD and tested on three validation datasets, namely, SeaWiFS and MODIS matchupdatasets and a simulated dataset with measurements from the Red Sea. The proposed MLPNN showed promising results over traditionalalgorithms with blue and green bands over a wide range of chl-a concentration levels including oligotrophic and eutrophic waters.Hyperparameter tuning of the MLPNN and extensive tests with different types of input configuration (using only four bands) resulted in thebest configuration to estimate chl-a. Implemented the MLPNN on satellite data from SeaWiFS, MODIS, VIIRS, and SeaHAWK sensorsindicated the ability of MLPNN to identify ocean color features in global open ocean, coastal waters and in phytoplankton blooms inArabian Sea and the Red Sea. The MLPNN developed in the study has direct implications for the future multi- and hyperspectral sensors,such as PACE and OceanSat-3, to understand phytoplankton distribution, primary productivity, and other processes at regional and globalscales.
Biography:
Dr. Surya Prakash Tiwari received a Ph.D. deegree in Ocean Engineering at Indian Institute of Technology Madras in 2013. He worked at the City College City University of New York, USA as a postdoctoral fellow during 2013-2014. From 2013-2017, Dr. Tiwari worked as a postdoctoral fellow atthe King Abdullah University of Science & Technology (KAUST), Saudi Arabia. At present, he is associate professor at King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia. He has published 23 ISI research papers in the top-ranked international journals
