Estimation of Secondary Phytoplankton Pigments From Satellite Observations Using Self‐Organizing Maps (SOMs)

Abstract

This study presents a method for estimating secondary phytoplankton pigments from satellite ocean color observations. We first compiled a large training data set composed of 12,000 samples; each sample is composed of 10 in situ phytoplankton high‐performance liquid chromatography (HPLC)‐measured pigment concentrations, GlobColour products of chlorophyll‐a concentration, and remote sensing reflectance (Rrs(λ)) data at different wavelengths, in addition to advanced very high resolution radiometer sea surface temperature measurements. The resulting data set regroups a large variety of encountered situations between 1997 and 2014. The nonlinear relationship between the in situ and satellite components was identified using a self‐organizing map, which is a neural network classifier. As a major result, the self‐organizing map enabled reliable estimations of the concentration of chlorophyll‐a and of nine different pigments from satellite observations. A cross‐validation procedure showed that the estimations were robust for all pigments (R2 > 0.75 and an average root‐mean‐square error = 0.016 mg/m3). A consistent association of several phytoplankton pigments indicating phytoplankton group specific dynamic was shown at a global scale. We also showed the uncertainties for the estimation of each pigment.

Radiometric approach for the detection of picophytoplankton assemblages across oceanic fronts

“…1) [8,14,52]. This Prochlorococcus distribution pattern agrees with predictions of other ocean color-based models, such as Alvain et al [75], El-Hourany et al [76], and Xi et al [31], and the model of Lange et al [28] which combines ocean color information with environmental variables. In turn, a model based solely on environmental variables (SST, photosynthetically-active radiation -PAR) -i.e.…”

Phytoplankton Diversity in the Mediterranean Sea From Satellite Data Using Self‐Organizing Maps

“…Comparing the PFTs given by the present method to these obtained by processing in situ data with the identification criteria proposed by Alvain et al (), we found some differences in identifying Nano and Syne classes. These errors that are of 23% and 13%, respectively, can be associated with cumulative uncertainties during the process of the method; some errors can be due to the pigment estimation in the first step of the approach when using the SOM‐Pigments (El Hourany et al, ), where 19HF or Zea concentrations can be misestimated with high uncertainties. The use of SOM‐Pigment highly depends on the quality of the satellite data (GlobColour and AVHRR) and therefore controls the pigment estimation errors.…”

“…The PFTs can be retrieved from the secondary pigments. Since Med‐Pigments data set is sparse in time and space, we built a high coverage secondary pigment database by processing the GlobColour and AVHRR SST data on the Mediterranean Sea region with the SOM‐Pigments as did El Hourany et al (). Level 3 mapped 4‐km daily images of SST, Chla, and Rrs at four wavelength (412, 443, 490, and 555 nm) were used as input of SOM‐Pigments.…”

“…The SOMs are unsupervised neural classifiers commonly used when dealing with environmental studies (Cavazos & Cavazos, ; Hewitson & Crane, ; Liu et al, ; Liu & Weisberg, ; Niang et al, ; Reusch et al, ; Richardson et al, ). In El Hourany et al (), a global approach was built to estimate phytoplankton pigment concentration data using SOM (SOM‐Pigments) and optical satellite data. This approach was calibrated and efficiently validated using a global database of High‐performance liquid chromatography (HPLC) measured pigments collocated with GlobColour satellite data (Chla, Rrs at four wavelengths; 412, 443, 490, and 555 nm) alongside the AVHRR sea surface temperature (SST).…”

“…While observing the availability of in situ HPLC measurements in the Mediterranean Sea, the heterogeneity of the sampling locations, which are significantly more abundant in the western basin, limits the development of a regional algorithm. Therefore, our method was based on estimating the phytoplankton pigments composition of the Mediterranean Sea from satellite data by using the SOM‐Pigments developed by El Hourany et al (). Then the dominant PFTs were estimated by using a dedicated SOM (the SOM‐PFT), which was calibrated on the output of the SOM‐Pigments.…”