Fusing Multiple Ocean Color Sensors for Coastal Water Clarity Monitoring

Water clarity is an excellent first-order indicator for overall water quality and is commonly represented with the Secchi disk depth (Z_SD, m). With Z_SD measurable from spaceborne platforms, retrieving satellite Z_SD from multiple ocean color sensors would increase water clarity observations. However, it is essential to achieve consistent Z_SD from the different spectral band settings of various satellite sensors. Thus, this study applied the Lee et al. (2015) Z_SD theoretical model to a hyperspectral synthetic dataset (N = 720) to determine the consistency between Landsat 8 (L8), Sentinel 2 (S2), and Sentinel 3 (S3) Z_SD. Since this model of Z_SD depends on the minimum diffuse attenuation coefficient (K_d), L8, S2, and S3 minimum K_d were compared and minimal variations were observed. However, L8, S2, and S3 minimum K_d underestimated minimum K_d estimated from hyperspectral information. To improve the estimation of minimum K_d and cross-sensor consistency, hyperspectral transfer coefficient tables were developed from L8, S2, and S3 multispectral information. As K_d is primarily a function of absorption (a) and backscattering (b_b) coefficients, hyperspectral transfer a and b_b coefficient tables were developed and tested on a separate synthetic dataset. With promising a and b_b matchups (N = 500, MAPE = < 4%), the hyperspectral transfer coefficient tables were applied to L8, S2, and S3 imagery over coastal Massachusetts waters and were validated with in situ citizen scientist Z_SD datasets. This work presents an effective system that enables the fusion of multiple ocean color sensors for coastal water clarity monitoring.