Accuracy Assessment of Bio-optical Models to Retrieve Backscattering Coefficients in Case 1 Waters of the Bay of Bengal

• Published in: Journal of the Indian Society of Remote Sensing 2019-03, Vol.47 (3), p.527-535
• Main Authors: Baranval, Nikhil Kumar, Nagamani, P. V., Rama Rao, P., Choudhury, S. B.
• Format: Article
• Language: English
• Subjects: Accuracy; Algorithms; Backscattering; Bias; Coefficients; Earth and Environmental Science; Earth Sciences; Empirical analysis; Indian space program; Indian spacecraft; Infrared imaging; Infrared radiometers; Mathematical models; Model accuracy; MODIS; Ocean models; Optical properties; Optimization; Optimization techniques; Radiometry; Reflectance; Remote sensing; Remote Sensing/Photogrammetry; Research Article; Root-mean-square errors; Satellite imagery; Satellites; Sensors; Spectroradiometers; Statistical analysis
• ISSN: 0255-660X; 0974-3006
• DOI: 10.1007/s12524-019-00937-3

The backscattering coefficient ( b b ) has been obtained either in situ observations/measurements or semi-analytical/analytical and empirical algorithms that depend on the relationship between ‘ b b ’ and the remote sensing reflectance ( R rs ). Several models were developed to estimate backscattering coefficient from the satellite imagery. The present study aims to assess the accuracy of bio-optical models and infer the best suitable model for Ocean Colour Monitoring (OCM-2) and for the upcoming OCM-3 sensors of Indian Space Research Organization. For this purpose, the bio-optical algorithms/models such as quasi-analytical algorithm (QAA), QAA version 5, Generalized Inherent Optical Properties (GIOP) and optimization of semi-analytical models in case 1 waters of the Bay of Bengal (BoB) were considered. For this analysis, three satellite sensors Moderate Resolution Imaging Spectroradiometer (MODIS)-Aqua, Visible Infrared Imaging Radiometer Suite (VIIRS), and OCM-2 datasets has been used to assess the model’s accuracy. The results showed that the optimization technique performs better (bias = 0.002, 0.179 and root-mean-square error (RMSE) = 8%, 20%) for MODIS-A and OCM-2, respectively, whereas GIOP performs better (bias = 0.08, RMSE = 14%) for VIIRS in case 1 waters of the BoB. From the statistical analysis of each model for all stations, it is recommended to use the optimization technique rather than GIOP technique for estimation of satellite based ‘ b b ’ in Indian waters.