Verification of a simple band ratio algorithm for retrieving Great Lakes open water surface chlorophyll concentrations from satellite observations

We compared in situ surface chlorophyll concentration values measured between 2012 and 2015 as part of the U.S. Environmental Protection Agency's Great Lakes National Program Office (GLNPO) annual monitoring program with corresponding concentration estimates obtained by applying our previously published (Lesht et al., 2013) Great Lakes Fit (GLF) band ratio algorithm to data from the Moderate-resolution Imaging Spectroradiometer (MODIS) sensor. Coefficients used in the original GLF algorithm were derived from similarly matched GLNPO and satellite observations collected between 2002 and 2011. The Model II linear relationship between the original GLF-predicted log-transformed values and the new set (2012-2015) of field observations yielded intercept = 0.036, slope = 1.063, and r2 = 0.830. Residuals for modeled chlorophyll concentrations below ~ 8.0 mg m − 3 were unbiased and normally distributed, but positively biased at higher modeled concentrations. When applied to the entire dataset (2002-2015), the linear relationship between the GLF-modeled and the observed values had intercept = 0.000, slope = 0.999, and r 2 = 0.820. New model coefficients derived from the entire (2002-2015) dataset were very similar to those obtained from the 2002-2011 data. Continual testing and assessment of any empirical model are desirable especially when the model is designed to be employed by a broad community. We conclude that this comparison of the GLF algorithm with the additional four years of independent data further validates its use for estimating surface chlorophyll concentrations from satellite observations of the open waters of the Great Lakes.