AN IMPROVED BIO-OPTICAL MODEL FOR REMOTE SENSING OF PHYTOPLANKTON ABSORPTION COEFFICIENT IN GLOBAL WATERS

A new empirical model has been developed and validated for determining phytoplankton absorption coefficient a_ph(λ) in global waters using high-quality in-situ hyperspectral remote sensing reflectance R_rs(λ) and MODIS-Aqua data. The strong correlation between the a_ph(λ) and the ratio of R_rs(λ) at 490 and 665 nm has been exploited to model the data. An extensive in-situ data set that has 1290 independent stations with co-existing R_rs(λ) and a_ph(λ) values that spans global oceans is utilized for the training and validation of the model. The developed model delivers acceptable results in terms of statistical measures. The model gives a maximum mean absolute percent error of 0.22, R² value not less than 0.72, and slope estimates very close to 1 for all visible spectral bands. The model has been further extended to analyze the variability of a_ph(443) and a_ph(665) over the global oceans using MODIS-Aqua images acquired from the NASA Ocean Color repository.