A Bidirectional Subsurface Remote Sensing Reflectance Model Explicitly Accounting for Particle Backscattering Shapes

The subsurface remote sensing reflectance ( r rs , sr −1 ), particularly its bidirectional reflectance distribution function (BRDF), depends fundamentally on the angular shape of the volume scattering functions (VSFs, m −1 sr −1 ). Recent technological advancement has greatly expanded the collection, and the knowledge of natural variability, of the VSFs of oceanic particles. This allows us to test the Zaneveld's theoretical r rs model that explicitly accounts for particle VSF shapes. We parameterized the r rs model based on HydroLight simulations using 114 VSFs measured in three coastal waters around the United States and in oceanic waters of North Atlantic Ocean. With the absorption coefficient ( a ), backscattering coefficient ( b b ), and VSF shape as inputs, the parameterized model is able to predict r rs with a root mean square relative error of ∼4% for solar zenith angles from 0 to 75°, viewing zenith angles from 0 to 60°, and viewing azimuth angles from 0 to 180°. A test with the field data indicates the performance of our model, when using only a and b b as inputs and selecting the VSF shape using b b , is comparable to or slightly better than the currently used models by Morel et al. and Lee et al. Explicitly expressing VSF shapes in r rs modeling has great potential to further constrain the uncertainty in the ocean color studies as our knowledge on the VSFs of natural particles continues to improve. Our study represents a first effort in this direction.