1. Satellite Retrieval of Phytoplankton Community Size Structure in the Global Ocean Colleen Mouw University of Wisconsin-Madison In collaboration with Jim Yoder Woods Hole Oceanographic Institution Photo David Doubilet

2. Ecological Importance of Cell Size Chisholm, 2000 Small cells: recycled within euphotic zone utilizing regenerated nutrients Prefer stratified high light conditions Large cells: sink out of the euphotic zone utilize new nutrients efficiently Prefer turbulent, low light conditions Many biogeochemical processes are directly related to the distribution of phytoplankton size class (Longhurst 1998), and is a major biological factor that governs the functioning of pelagic food webs (Legendre and Lefevre 1991).

3. Optical Importance of Cell Size Despite the physiological and taxonomic variability, variation in spectral shape can be defined by changes in the dominant size class. (Ciotti et al. 2002) a * ph ( )= [(1-S f )  a * pico ( )] + [S f  a * micro ( )] Package effect

4. Motivation R rs ( ) imagery also contains information about cell size in addition to chlorophyll and CDM concentration. O’Reilly et al. 1998 Chl (mg m -3 ) R=log{(R rs 443 > R rs 490 > R rs 510)/R rs 555} SeaWiFS standard chlorophyll algorithm (OC4v4).

5. Effect of Phytoplankton Concentration on R rs ( ) O’Reilly et al. 1998 Maximum band shifts from 443 to 490 to 510 nm with increasing chlorophyll concentration Effect of [Chl] on water-leaving radiance Spectral shift

6. Effect of Cell Size on R rs ( ) Magnitude shift! S f varying Constant [Chl] = 0.5 mg m -3 Constant a CDM/NAP (443) = 0.002 m -1 Wavelength (nm) R rs (sr -1 ) Hydrolight simulations

7. Effect of CDM/NAP on R rs ( ) In addition to the magnitude shift of cell size, effects of CDM/NAP must be considered. a CDM/NAP (443) varying Constant Chl = 0.5 mg m -3 Constant S f = 50% Wavelength (nm) R rs (sr -1 )

8. How can phytoplankton cell size be retrieved from satellite imagery? Mouw & Yoder (2009) Remote Sensing of Environment, submitted

9. HPLC in situ observations n=4,564 Percent microplankton Log 10 in situ [Chl] (mg m -3 ) The relative biomass proportions of pico-, nano-, and microplankton can be estimated from the concentrations of pigments which have a taxonomic significance and associated to a size class (Bricaud et al. 2004; Vidussi et al. 1996).

10. Look-up-table Construction Full factorial design Independently varied [Chl], S f, & a CDM/NAP over expected ranges for the global ocean For a given combination of IOPs, AOPs are calculated via radiative transfer

11. Look-up-table Construction Full Factorial Design: Chl, S f, a CDM/NAP (443) Optical model Hydrolight R rs ( ) Log 10 in situ [Chl] (mg m -3 ) Log 10 GSM01 [Chl] (mg m -3 ) Percent Microplankton GSM01 a CDM/NAP (443) m -1 n = 44,343

12. Wavelength (nm) Detectable Ranges If LUT ∆nRrs(443) > SeaWiFS NE∆nR rs (443) –Beyond detection R rs (sr -1 ) Chlorophyll (mg m -3 ) a CDM/NAP (443) (m -1 ) SeaWiFS has the sensitivity to retrieve S f... chlorophyll 0.05 - 1.75 mg m -3 a CDM/NAP (443) < 0.17 m -1 Of decadal mean imagery, 84% of [Chl] 99.7% of a CDM/NAP (443) fall within thresholds

13. LUT Retrieval SeaWiFS R rs ( ) imagery If a CDM (443) > threshold  Mask If a CDM (443) < threshold  Continue    Hydrolight Normalized R rs (443) (S f range) GSM01 Chl GSM01 a CDM/NAP (443) SfSf SeaWiFS Normalized & Corrected R rs (443) If [Chl] above/below threshold <  Mask If [Chl] within threshold  Continue 0.05 - 1.75 mg m -3 < 0.17 m -1 (443/555) Guide search space in LUT

14. Land/Cloud Size Retrieval Masked regions that are outside of thresholds for S f retrieval. Estimated S f for May 2006 High CDM/Chl Low Chl No flag

15. Validation 85% within 1 standard deviation 11%, 2 std. dev. 4%, 3 std. dev. S f in situ S f retrieval

16. Comparison with other functional type retrievals Uitz et al. 2006 June 2000

17. S f - SeaWiFS first 10 years

18. How do the S f temporal and spatial patterns compare with [Chl]?

19. S f and [Chl] Decadal Climatology

20. Individual EOF – Mode 1 [Chl] - adjustments to seasonal cycle S f - ENSO relations –Smaller S f deviations until until 2002 (Equatorial Pacific) when deviations become negative

21. Joint EOF – Mode 1 Amplitude time series – mirror over zero of individual S f mode 1 - Variance driven by S f

22. Summary Satellite S f estimates agree well with previous observations Regions of the ocean where S f and [Chl] are decoupled ENSO variability more apparent in S f than [Chl] Non-linear response between S f & [Chl] points to the importance of additional ecological information in the interpretation of [Chl] distributions

23. Moving Forward Much more to investigate with S f time series… –Further investigation of S f changes over the decadal record –Flux estimates with assistance from numerical models –Production estimates considering cell size (Mouw & Yoder 2005) –Other suggestions/ideas…

24. Acknowledgements Jim Yoder (WHOI) Jay O’Reilly (NOAA, NMFS) Tatiana Rynearson (URI, GSO) Benjamin Beckmann (MSU) Maureen Kennelly (URI, GSO) Kim Hyde (NOAA, NMFS) Primary Funding –RI Space Grant/Vetlesen Climate Change Fellowship –NASA Earth and Space Science Fellowship –URI GSO Alumni Fellowship