## Abstract

Using a data set of 1333 samples, we assess the spectral absorption relationships
of different wave bands for phytoplankton (ph) and particles. We find that a
nonlinear model (second-order quadratic equations) delivers good performance in
describing their spectral characteristics. Based on these spectral
relationships, we develop a method for partitioning the total absorption
coefficient into the contributions attributable to phytoplankton
$[{a}_{\mathrm{ph}}(\lambda )]$, colored dissolved organic material [CDOM;
${a}_{\mathrm{CDOM}}(\lambda )$], and nonalgal particles [NAP;
${a}_{\mathrm{NAP}}(\lambda )$]. This method is validated using a data set
that contains 550 simultaneous measurements of phytoplankton, CDOM, and NAP from
the NASA bio-Optical Marine Algorithm Dataset. We find that our method is highly
efficient and robust, with significant accuracy: the relative root-mean-square
errors (RMSEs) are 25.96%, 38.30%, and 19.96% for ${a}_{\mathrm{ph}}(443)$, ${a}_{\mathrm{CDOM}}(443)$, and the CDOM exponential slope, respectively.
The performance is still satisfactory when the method is applied to water
samples from the northern South China Sea as a regional case. The computed and
measured absorption coefficients (167 samples) agree well with the RMSEs, i.e.,
18.50%, 32.82%, and 10.21% for ${a}_{\mathrm{ph}}(443)$, ${a}_{\mathrm{CDOM}}(443)$, and the CDOM exponential slope, respectively.
Finally, the partitioning method is applied directly to an independent data set
(1160 samples) derived from the Bermuda Bio-Optics Project that contains
relatively low absorption values, and we also obtain good inversion accuracy
[RMSEs of 32.37%, 32.57%, and 11.52% for ${a}_{\mathrm{ph}}(443)$, ${a}_{\mathrm{CDOM}}(443)$, and the CDOM exponential slope, respectively].
Our results indicate that this partitioning method delivers satisfactory
performance for the retrieval of ${a}_{\mathrm{ph}}$, ${a}_{\mathrm{CDOM}}$, and ${a}_{\mathrm{NAP}}$. Therefore, this may be a useful tool for
extracting absorption coefficients from *in situ* measurements or
remotely sensed ocean-color data.

© 2013 Optical Society of America

Full Article | PDF Article