How well do we estimate particulate backscatter from satellites as compared to autonomous profiling floats in the open ocean?

Kelsey Bisson1, Emmanuel Boss2, Toby Kolohe Westberry1 and Michael Behrenfeld1, (1)Oregon State University, Corvallis, OR, United States, (2)University of Maine, Orono, ME, United States
Particulate backscatter (bbp) has been used to as a proxy of particulate organic carbon, phytoplankton carbon, and as an input to estimate particle size, particle export, and net primary production on global scales using satellites. Satellite derived bbp has been historically difficult to validate due to sparse in situ observations, but it is important (especially for global ecosystem studies) to know where, when, and how satellite derived bbp may be biased. The ongoing deployment of BGC-Argo floats with backscattering sensors (at 700 nm) makes it now possible to compare satellite estimates of bbp with a large number of coincident in situ bbp matchups. Depending on the sensor, location, or bbp algorithm used, there is a wide spread in performance (Spearman’s rank correlation varies from r = 0.06 to r = 0.79). The performance differences allow us to 1) investigate the influence that Rrs noise, spatial sampling bias, and the assumed bbp spectral slope have on bbp retrievals across the satellite sensors, and 2) provide recommendations for future BGC-Argo sampling and satellite validation efforts. Our results broadly demonstrate the utility of using BGC-Argo floats for validation of ocean satellite products.