Uncertainties in Coastal Ocean Color Products: Impacts of Spatial Sampling

With increasing demands for ocean color (OC) products with improved accuracy and well characterized, per-retrieval uncertainty budgets, it is vital to decompose overall estimated errors into their primary components. Amongst various contributing elements (e.g., instrument calibration, atmospheric correction, inversion algorithms) in the uncertainty of an OC observation, less attention has been paid to uncertainties associated with spatial sampling. In this paper, we simulate MODIS and VIIRS OC products from 30m resolution OC products derived from the Operational Land Imager (OLI) aboard Landsat-8, to examine impacts of spatial sampling on both cross-sensor product intercomparisons and in-situ validations of R_rs products in coastal waters. The simulations were carried out for OLI scenes "scanned" for one full orbital-repeat cycle of each ocean color satellite. While some view-angle dependent differences in simulated Aqua-MODIS and VIIRS were observed, the average uncertainties (absolute) in product intercomparisons (due to differences in spatial sampling) at regional scales are found to be 1.8%, 1.9%, 2.4%, 4.3%, 2.7%, 1.8%, and 4% for the R_rs(443), R_rs(482), R_rs(561), R_rs(655), [Chla], K_d(482), and b_bp(655) products, respectively. It is also found that, depending on in-water spatial variability and the sensor's footprint size, the errors for an in-situ validation location in coastal areas can reach as high as ±18%. We conclude that a) expected biases induced by the spatial sampling in product intercomparisons are mitigated when products are averaged over at least 7km×7km windows, b) VIIRS observations, with improved consistency in cross-track spatial sampling yields more precise calibration/validation results than MODIS, and c) use of a single pixel centered on in-situ coastal sites provides an optimal sampling size for validation efforts. These findings will have implications for enhancing our understanding of uncertainties in ocean color retrievals and for planning of future calibration/validation exercises.