Phytoplankton
Description: Phytoplankton products - Chlorophyll a, Primary Production, and Phytoplankton Size Class
Found in: State of the Ecosystem - Gulf of Maine & Georges Bank (2018+), State of the Ecosystem - Mid-Atlantic (2018+)
Indicator category: Database pull; Database pull with analysis; Published methods
Contributor(s): Kimberly Hyde
Data steward: Kimberly Hyde, kimberly.hyde@noaa.gov
Point of contact: Kimberly Hyde, kimberly.hyde@noaa.gov
Public availability statement: Source data used in these analyses are publicly available.
Current Methods
Data sources
Daily Level 3 mapped (4km resolution, sinusoidally projected) satellite ocean color data are acquired from the European Space Agency’s Ocean Colour Climate Change Initiative (OC-CCI; version 6.0) and GlobColour Project. The OC-CCI data is the primary ocean color data source, however the data latency is approximately 6-12 months. GlobColour ocean color data are used to supplement the OC-CCI data to complete the time series for the current year. Sea Surface Temperature (SST) data include the 4 km nighttime NOAA Advanced Very High Resolution Radiometer (AVHRR) Pathfinder (Casey et al. (2010); Saha et al. (2018)) and the Group for High Resolution Sea Surface Temperature (GHRSST) Multiscale Ultrahigh Resolution (MUR, version 4.1) Level 4 (Chin, Vazquez-Cuervo, and Armstrong (2017); Project (2015)) data. AVHRR Pathfinder data are used as the SST source until 2002 and MUR SST in subsequent years.
Data analysis
The L3 OC-CCI products merge data from multiple ocean color sensors (SeaWiFS, MODIS Aqua, MERIS, VIIRS, Sentinel 3A and 3B OLCI) and include chlorophyll a (CHL-CCI), remote sensing reflectance \((R_{rs}(\lambda))\), and several inherent optical property (IOPs) products. The CHL-CCI blended algorithm attempts to weight the outputs of the best-performing chlorophyll algorithms based on the water types present, which improves performance in nearshore water compared to open-ocean algorithms. The L3 GlobColour products use data from the same ocean color sensors as the OC-CCI, but the chlorophyll a product is derived from the Garver, Siegel, and Maritorena (GSM) algorithm, which is a semi-analytical bio-optical model (O’Reilly et al. (1998)). GlobClolour also provides a photosynthetic available radiation (PAR) product, which is the mean daily photon flux density in the visible range (400 to 700 nm) that are used in the primary production calculations. The global OC-CCI, GlobColour, and the SST data are mapped to the same sinusoidal map projection and subset to the east coast region (SW longitude=82.5\(^\circ\)W, SW latitude=22.5\(^\circ\)N, NE longitude=51.5\(^\circ\)W, NE latitude=48.5\(^\circ\)N).
Data Interpolation
For use in the primary production model, the daily CHL and AVHRR SST data are temporally interpolated and smoothed (CHLINT and SSTINT respectively). The interpolation increases the data coverage and is necessary to better match data collected from different sensors and different times. The daily PAR data are not affected by cloud cover and MUR SST data is a blended/gap free product so these parameters were not interpolated. Daily data at each pixel location are linearly interpolated based on days in the time series using interpx.pro. Prior to interpolation, the CHL data are log-transformed to account for the log-normal distribution of chlorophyll data (Campbell (1995)). The time series are processed in one-year chunks, with each yearly series including 60 days from the previous year and 60 days from the following year to improve the interpolation at the beginning and end of the year. Following interpolation, the data are smoothed with a tri-cube filter (width=7) using IDL’s CONVOL program. In order to avoid over interpolating data when there were several days of missing data in the time series, the interpolated data were removed and replaced with blank data if the window of interpolation spanned more than 7 days for CHL or 10 days for SST.
Primary Productivity
The Vertically Generalized Production Model (VGPM) estimates net primary production (PP) as a function of chlorophyll a, photosynthetically available radiation (PAR), and photosynthetic efficiency (Behrenfeld and Falkowski (1997)). In the VGPM-Eppley version, the original temperature-dependent function to estimate the chlorophyll-specific photosynthetic efficiency is replaced with the exponential “Eppley” function (Equation 14.1) as modified by Morel (1991). The VGPM calculates the daily amount of carbon fixed based on the maximum rate of chlorophyll-specific carbon fixation in the water column, sea surface daily photosynthetically available radiation, the euphotic depth (the depth where light is 1% of that at the surface), chlorophyll a concentration, and the number of daylight hours (Equation (9.1)).
\[\begin{equation}
P_{max}^{b}(SST) = 4.6 * 1.065^{SST-20^{0}}
\tag{9.1}
\end{equation}\]
Where \(P_{max}^{b}\) is the maximum carbon fixation rate and SST is sea surface temperature.
\[\begin{equation}
PP_{eu} = 0.66125 * P_{max}^{b} * \frac{I_{0}}{I_{0}+4.1} * Z_{eu} * \textrm{CHL} * \text{DL}
\tag{9.2}
\end{equation}\]
Where \(PP_{eu}\) is the daily amount of carbon fixed integrated from the surface to the euphotic depth (mgC m-2 day-1), \(P_{max}^{b}\) is the maximum carbon fixation rate within the water column (mgC mgChl-1 hr-1), \(I_{0}\) is the daily integrated molar photon flux of sea surface PAR (mol quanta m-2 day-1), Zeu is the euphotic depth (m), CHL is the daily interpolated CHLINT-CCI (mg m-3), and DL is the photoperiod (hours) calculated for the day of the year and latitude according to Kirk (1994). The light dependent function \((I_{0}/(I_{0}+4.1))\) describes the relative change in the light saturation fraction of the euphotic zone as a function of surface PAR (\(I_0\)). Zeu is derived from an estimate of the total chlorophyll concentration within the euphotic layer (CHLeu) based on the Case I models of Morel and Berthon (1989):
- For \(\textrm{CHL}_{eu} > 10.0\;\;\;\;\;Z_{eu} = 568.2 * \textrm{CHL}_{eu}^{-0.746}\)
- For \(\textrm{CHL}_{eu} \leq 10.0\;\;\;\;\;Z_{eu} = 200.0 * \textrm{CHL}_{eu}^{-0.293}\)
- For \(\textrm{CHL}_{0} \leq 1.0\;\;\;\;\;\textrm{CHL}_{eu} = 38.0 * \textrm{CHL}_{0}^{0.425}\)
- For \(\textrm{CHL}_{0} > 1.0\;\;\;\;\;\textrm{CHL}_{eu} = 40.2 * \textrm{CHL}_{0}^{0.507}\)
Where \(\textrm{CHL}_0\) is the surface chlorophyll concentration.
Phytoplankton Size Class
Phytoplankton size classes (PSC) are calculated according to Turner et al. (2021). The regionally tuned abundance-based model is based on the three-component model of Brewin et al. (2010) that varies as a function of SST (Brewin et al. (2017), Moore and Brown (2020)). The model uses a look-up table with parameters indexed by SST, developed using a local data set of HPLC diagnostic pigment-derived phytoplankton size fractions matched with coincident satellite SST.
Stastistics and Anomalies
Statistics, including the arithmetic mean, geometric mean, median, standard deviation, and coefficient of variation are calculated at daily (3 and 8-day running means), weekly, monthly, and annual time steps, and for several climatological periods. Annual statistics used the monthly means as inputs to avoid a summer time bias when more data are available due to reduced cloud cover. The daily, weekly, monthly and annual climatological statistics include the entire time series for each specified period. For example, the climatological January uses the monthly mean from each January in the time series and the climatological annual uses the annual mean from each year. The CHL and PP climatological statistics include data from both SeaWiFS (1997-2007) and MODIS (2008-2017).
Weekly, monthly and annual anomalies are calculated for each product by taking the difference between the mean of the input time period (i.e. week, month, year) and the climatological mean for the same period. Because bio-optical data are typically log-normally distributed Campbell (1995), the CHL and PP data were first log-transformed prior to taking the difference and then untransformed, resulting in an anomaly ratio.
The ecological production unit (EPU) shapefile that excludes the estuaries was used to spatially extract all data located within an ecoregion from the statistic and anomaly files. The median values, which are equivalent to the geometric mean, were used for the CHL and PP data.
Data processing
CHL and PPD time series were formatted for inclusion in the ecodata R package using the R code found here.
Code used to process the phytoplankton size class inidcator can be found in the ecodata package here.
2018-2020 Methods
Data sources
Level 1A ocean color remote sensing data from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) (NASA Ocean Biology Processing Group 2018) on the OrbView-2 satellite and the Moderate Resolution Imaging Spectroradiometer (MODIS) (NASA Ocean Biology Processing Group 2017) on the Aqua satellite were acquired from the NASA Ocean Biology Processing Group (OBPG). Sea Surface Temperature (SST) data included the 4 km nighttime NOAA Advanced Very High Resolution Radiometer (AVHRR) Pathfinder (Casey et al. 2010; Saha et al. 2018) and the Group for High Resolution Sea Surface Temperature (GHRSST) Multiscale Ultrahigh Resolution (MUR, version 4.1) Level 4 (Chin, Vazquez-Cuervo, and Armstrong 2017; Project 2015) data. Prior to June 2002, AVHRR Pathfinder data are used as the SST source and MUR SST in subsequent years.
Data analysis
The SeaWiFS and MODIS L1A files were processed using the NASA Ocean Biology Processing Group SeaDAS software version 7.4. All MODIS files were spatially subset to the U.S. East Coast (SW longitude=-82.5, SW latitude=22.5, NE longitude=-51.5, NE latitude=48.5) using L1AEXTRACT_MODIS. SeaWiFS files were subset using the same coordinates prior to begin downloaded from the Ocean Color Web Browser. SeaDAS’s L2GEN program was used to generate Level 2 (L2) files using the default settings and optimal ancillary files, and the L2BIN program spatially and temporally aggregated the L2 files to create daily Level 3 binned (L3B) files. The daily files were binned at 2 km resolution that are stored in a global, nearly equal-area, integerized sinusoidal grids and use the default L2 ocean color flag masks. The global SST data were also subset to the same East Coast region and remapped to the same sinusoidal grid.
The L2 files contain several ocean color products including the default chlorophyll a; product (CHL-OCI), photosynthetic available radiation (PAR), remote sensing reflectance \((R_{rs}(\lambda))\), and several inherent optical property products (IOPs). The CHL-OCI product combines two algorithms, the O’Reilly band ratio (OCx) algorithm (O’Reilly et al. 1998) and the Hu color index (CI) algorithm (Hu, Lee, and Franz 2012). The SeaDAS default CHL-OCI algorithm diverges slightly from Hu, Lee, and Franz (2012) in that the transition between CI and OCx occurs at 0.15 < CI < 0.2 mg m-3 to ensure a smooth transition. The regional chlorophyll a algorithm by Pan et al. (2008) was used to create a second chlorophyll product (CHL-PAN). CHL-PAN is an empirical algorithm derived from in situ sampling within the Northeast Large Marine Ecosystem (NE-LME) and demonstrated significant improvements from the standard NASA operational algorithm in the NES-LME (Pan et al. 2010). A 3rd-order polynomial function (Equation (9.3)) is used to derive [CHL-PAN] from Rrs band ratios (RBR):
\[\begin{equation}
log[\textrm{CHL-PAN}] = A_{0} + A_{1}X + A_{2}X^{2} + A_{3}X^{3},
\tag{9.3}
\end{equation}\]
where \(X = log(R_{rs}(\lambda_{1})/R_{rs}(\lambda_{2}))\) and \(A_{i} (i = 0, 1, 2, \textrm{or } 3)\) are sensor and RBR specific coefficients:
- If SeaWiFS and RBR is \(R_{rs}(490)/R_{rs}(555)(R_{^3{\mskip -5mu/\mskip -3mu}_5})\) then: \(A_0=0.02534, A_1=-3.033, A_2=2.096, A_3=-1.607\)
- If SeaWiFS and RBR is \(R_{rs}(490)/R_{rs}(670)(R_{^3{\mskip -5mu/\mskip -3mu}_6})\) then: \(A_0=1.351, A_1=-2.427, A_2=0.9395, A_3=-0.2432\)
- If MODIS and RBR is \(R_{rs}(488)/R_{rs}(547)(R_{^3{\mskip -5mu/\mskip -3mu}_5})\) then: \(A_0=0. 03664, A_1=-3.451, A_2=2.276, A_3=-1.096\)
- If MODIS and RBR is \(R_{rs}(488)/R_{rs}(667)(R_{^3{\mskip -5mu/\mskip -3mu}_6})\) then: \(A_0=1.351, A_1=-2.427, A_2=0.9395, A_3=-0.2432\)
C
3/5 and C
3/6 were calculated for each sensor specific RBR (R
3/5 and R
3/6 respectively) and then the following criteria were used to determine to derive CHL-PAN:
-
If \(R_{^3{\mskip -5mu/\mskip -3mu}_5}>0.15\) or \(R_{6} <0.0001\) then \(\textrm{CHL-PAN} = C_{^3{\mskip -5mu/\mskip -3mu}_5};\)
-
Otherwise, \(\textrm{CHL-PAN} = \textrm{max}(C_{^3{\mskip -5mu/\mskip -3mu}_5}, C_{^3{\mskip -5mu/\mskip -3mu}_6})\),
where \(R_6\) is \(R_{rs}(670)\) (SeaWiFS) or \(R_{rs}(667)\) (Pan et al. 2010).
catalog link
https://noaa-edab.github.io/catalog/chl_pp.html
https://noaa-edab.github.io/catalog/phyto_size.html
Behrenfeld, Michael J., and Paul G. Falkowski. 1997.
“Photosynthetic Rates Derived from Satellite-Based Chlorophyll Concentration.” Limnology and Oceanography 42 (1): 1–20.
https://doi.org/10.4319/lo.1997.42.1.0001.
Brewin, Robert J. W., Stefano Ciavatta, Shubha Sathyendranath, Thomas Jackson, Gavin Tilstone, Kieran Curran, Ruth L. Airs, et al. 2017.
“Uncertainty in Ocean-Color Estimates of Chlorophyll for Phytoplankton Groups.” Frontiers in Marine Science 4: 104.
https://doi.org/10.3389/fmars.2017.00104.
Brewin, Robert J. W., Shubha Sathyendranath, Takafumi Hirata, Samantha J. Lavender, Rosa M. Barciela, and Nick J. Hardman-Mountford. 2010.
“A Three-Component Model of Phytoplankton Size Class for the Atlantic Ocean.” Ecological Modelling 221 (11): 1472–83. https://doi.org/
https://doi.org/10.1016/j.ecolmodel.2010.02.014.
Campbell, Janet W. 1995.
“The Lognormal Distribution as a Model for Bio-Optical Variability in the Sea.” Journal of Geophysical Research: Oceans 100 (C7): 13237–54.
https://doi.org/10.1029/95JC00458.
Casey, Kenneth S., Tess B. Brandon, Peter Cornillon, and Robert Evans. 2010.
“The past, present, and future of the AVHRR pathfinder SST program.” In
Oceanography from Space: Revisited, 273–87.
https://doi.org/10.1007/978-90-481-8681-5_16.
Chin, Toshio Michael, Jorge Vazquez-Cuervo, and Edward M. Armstrong. 2017.
“A Multi-Scale High-Resolution Analysis of Global Sea Surface Temperature.” Remote Sensing of Environment 200: 154–69.
https://doi.org/10.1016/j.rse.2017.07.029.
Hu, Chuanmin, Zhongping Lee, and Bryan Franz. 2012.
“Chlorophyll a Algorithms for Oligotrophic Oceans: A Novel Approach Based on Three-Band Reflectance Difference.” Journal of Geophysical Research 117 (C1): C01011.
https://doi.org/10.1029/2011JC007395.
Kirk, John T. O. 1994. Light and Photosynthesis in Aquatic Ecosystems. Cambridge, UK: Cambridge University Press.
Moore, Timothy S., and Christopher W. Brown. 2020.
“Incorporating Environmental Data in Abundance-Based Algorithms for Deriving Phytoplankton Size Classes in the Atlantic Ocean.” Remote Sensing of Environment 240: 111689. https://doi.org/
https://doi.org/10.1016/j.rse.2020.111689.
Morel, André. 1991.
“Light and Marine Photosynthesis: A Spectral Model with Geochemical and Climatological Implications.” Progress in Oceanography 26 (3): 263–306.
https://doi.org/10.1016/0079-6611(91)90004-6.
Morel, André, and Jean-François Berthon. 1989.
“Surface Pigments, Algal Biomass Profiles, and Potential Production of the Euphotic Layer: Relationships Reinvestigated in View of Remote-Sensing Applications.” Limnology and Oceanography 34 (8): 1545–62.
https://doi.org/10.4319/lo.1989.34.8.1545.
NASA Ocean Biology Processing Group. 2017.
“MODIS-Aqua Level-3 Binned Garver, Siegel, Maritorena Model (GSM) Data Version R2018.0.” NASA Ocean Biology DAAC.
https://doi.org/10.5067/AQUA/MODIS/L3B/GSM/2018.
———. 2018.
“SEAWIFS-ORBVIEW-2 Level 3 Mapped Chlorophyll Data Version R2018.0.” NASA Ocean Biology DAAC.
https://doi.org/10.5067/ORBVIEW-2/SEAWIFS/L3M/CHL/2018.
O’Reilly, John E., Stéphane Maritorena, B. Greg Mitchell, David A. Siegel, Kendall L. Carder, Sara A. Garver, Mati Kahru, and Charles McClain. 1998.
“Ocean Color Chlorophyll Algorithms for SeaWiFS.” Journal of Geophysical Research: Oceans 103 (C11): 24937–53.
https://doi.org/10.1029/98JC02160.
Pan, Xiaoju, Antonio Mannino, Mary E. Russ, and Stanford B. Hooker. 2008.
“Remote Sensing of the Absorption Coefficients and Chlorophyll a Concentration in the United States Southern Middle Atlantic Bight from SeaWiFS and MODIS-Aqua.” Journal of Geophysical Research 113 (C11): C11022.
https://doi.org/10.1029/2008JC004852.
Pan, Xiaoju, Antonio Mannino, Mary E. Russ, Stanford B. Hooker, and Lawrence W. Harding Jr. 2010.
“Remote Sensing of Phytoplankton Pigment Distribution in the United States Northeast Coast.” Remote Sensing of Environment 114 (11): 2403–16.
https://doi.org/10.1016/j.rse.2010.05.015.
Project, JPL MUR MEaSUREs. 2015.
“GHRSST Level 4 MUR Global Foundation Sea Surface Temperature Analysis (V4.1).” CA, USA: NASA PO.DAAC.
https://doi.org/10.5067/GHGMR-4FJ04.
Saha, Korak, Xuepeng Zhao, Huai-min Zhang, Kenneth S. Casey, Dexin Zhang, Sheekela Baker-Yeboah, Katherine A. Kilpatrick, Robert H. Evans, Thomas Ryan, and John M. Relph. 2018.
“AVHRR Pathfinder version 5.3 level 3 collated (L3C) global 4km sea surface temperature for 1981-Present.” NOAA National Centers for Environmental Information.
https://doi.org/doi:10.7289/V52J68XX Accessed in 2016 and 2018.
Turner, Kyle J., Colleen B. Mouw, Kimberly J. W. Hyde, Ryan Morse, and Audrey B. Ciochetto. 2021.
“Optimization and Assessment of Phytoplankton Size Class Algorithms for Ocean Color Data on the Northeast u.s. Continental Shelf.” Remote Sensing of Environment 267: 112729. https://doi.org/
https://doi.org/10.1016/j.rse.2021.112729.