An open-source Hydrolight-based framework for fast inverse modelling of hyperspectral data

We present a fast and flexible python framework for forward and inverse modelling of hyperspectral data based on the formerly developed HYDROPT algorithm. The forward model is based on Hydrolight numerical solutions of the radiative transfer equations. Computation time is greatly reduced by the use of polynomial interpolation of the radiative transfer solutions, while at the same time maintains high accuracy. Additional features of HYDROPT are specification of sensor viewing geometries, sun zenith angle and IOP spectral models. Uncertainty estimates and goodness-of-fit metrics are simultaneously derived for the inversion routines. The use of HYDROPT on MERIS bands, in conjunction with regional specific IOPs, resulted in considerable improvements of retrievals in coastal and continental shelf waters. This illustrates the need for flexible retrieval algorithms that allow for the configuration of IOP models characteristic for the region of interest. The spectral absorption model for water within HYDROPT has been revised according to Mason et al. (2016) with substantial changes at shorter wavelengths. Moreover, our updated HYDROPT framework can be used for the inter-comparison of retrievals using different sensor band settings including coupling to full spectral coverage, as would be needed for NASA’s PACE. The flexible configuration of IOP spectral models within the HYDROPT framework allow for more than three components to be fitted, such as multiple phytoplankton types with distinct absorption and backscatter ratio characteristics. We showcase our model by evaluating its performance for retrievals in European waters with a strong transition between extreme case-2 and case-1 waters. We also discuss opportunities and limitations for the retrieval of phytoplankton community composition from hyperspectral observations.