Effect of inclusion of new variables in inversion of semi – analytical model for synthetic hyperspectral data of shallow waters

In the analysis of Inherent Optical Properties (IOPs which are a - total absorption coefficient and bb - the total backscattering coefficient) of water bodies, the parameters generally retrieved using semi-analytical model for inversion of hyperspectral remotely sensed reflectance from shallow waters are mainly based on five variables related to absorption due to phytoplankton, gelbstoff and detritus (combined), backscattering due to particulate matter, bottom reflectance and geometric depth. Optimization techniques, both local and global are used in inversion for retrieval of optimized values of aforementioned parameters. In semi-analytical models, two empirically fitted constants, g0 and g1 are often used to relate the subsurface remote sensing reflectance (rrs) to IOP. These two constants vary with particle phase function for a given remote sensing setting and hence need to be predetermined. In the present study, a local optimization technique for inversion is proposed, considering g0 and g1 as variables and its effect on improvement of accuracy of IOP retrieval is reported. A synthetic dataset with 3000 hyper spectra in 400-720 nanometre range with 5nm bandwidth and corresponding to sandy bottom type with three levels of depth with 20 concentrations of Chlorophyll - a is created to study the effect. We suggest that, processing of Hyperspectral Imagery from shallow waters can include these additional variables for better accuracies for retrieval of IOPs.