Performance of Algorithms for Retrieving Chlorophyll a Concentrations in the Arctic Ocean: Impact on Primary Production Estimates

Abstract

Chlorophyll a concentration (Chl) is a key variable for estimating primary production (PP) through ocean-color remote sensing (OCRS). Accurate Chl estimates are crucial for better understanding of the spatio-temporal trends in PP in recent decades as a consequence of climate change. However, a number of studies have reported that currently operational chlorophyll a algorithms perform poorly in the Arctic Ocean (AO), largely due to the interference of colored and detrital material (CDM) with the phytoplankton signal in the visible part of the spectrum. To determine how and to what extent CDM biases the estimation of Chl, we evaluated the performances of eight currently available ocean-color algorithms: OC4v6, OC3Mv6, OC3V, OC4L, OC4P, AO.emp, GSM01 and AO.GSM. Our results suggest that the empirical AO.emp algorithm performs the best overall, but, for waters with high CDM acdm(443) > 0.067 m−1), a common scenario in the Arctic, the two semi-analytical GSM models yield better performance. In addition, sensitivity analyses using a spectrally and vertically resolved Arctic primary-production model show that errors in Chl mostly propagate proportionally to PP estimates, with amplification of up to 7%. We also demonstrate that, the higher level of CDM in relation to Chl in the water column, the larger the bias in both Chl and PP estimates. Lastly, although the AO.GSM is the best overall performer among the algorithms tested, it tends to fail for a significant number of pixels (16.2% according to the present study), particularly for waters with high CDM. Our results therefore suggest the ongoing need to develop an algorithm that provides reasonable Chl estimates for a wide range of optically complex Arctic waters.