• Alvera-Azcárate, A., A. Barth, J.-M. Beckers, and R. H. Weisberg, 2007: Multivariate reconstruction of missing data in sea surface temperature, chlorophyll and wind satellite fields. J. Geophys. Res., 112, C03008, doi:10.1029/2006JC003660.

    • Search Google Scholar
    • Export Citation
  • Ashok, K., S. K. Behera, S. A. Rao, H. Weng, and T. Yamagata, 2007: El Niño Modoki and its possible teleconnection. J. Geophys. Res., 112, C11007, doi:10.1029/2006JC003798.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bailey, S. W., and P. J. Werdell, 2006: A multi-sensor approach for the on-orbit validation of ocean color satellite data products. Remote Sens. Environ., 102, 1223, doi:10.1016/j.rse.2006.01.015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ballabrera-Poy, J., R. G. Murtugudde, J. R. Christian, and A. J. Busalacchi, 2003: Signal-to-noise ratios of observed monthly tropical ocean color. Geophys. Res. Lett., 30, 1645, doi:10.1029/2003GL016995.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barber, R. T., and F. P. Chavez, 1983: Biological consequences of El Niño. Science, 222, 12031210, doi:10.1126/science.222.4629.1203.

  • Barnett, T. P., and R. Preisendorfer, 1987: Origins and levels of monthly and seasonal forecast skill for United States surface air temperatures determined by canonical correlation analysis. Mon. Wea. Rev., 115, 18251850, doi:10.1175/1520-0493(1987)115<1825:OALOMA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Behrenfeld, M. J., and et al. , 2006: Climate-driven trends in contemporary ocean productivity. Nature, 444, 752755, doi:10.1038/nature05317.

  • Boyce, D. G., M. R. Lewis, and B. Worm, 2010: Global phytoplankton decline over the past century. Nature, 466, 591596, doi:10.1038/nature09268.

  • Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate, 5, 541560, doi:10.1175/1520-0442(1992)005<0541:AIOMFF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Carton, J. A., and B. S. Giese, 2008: A reanalysis of ocean climate using Simple Ocean Data Assimilation (SODA). Mon. Wea. Rev., 136, 29993017, doi:10.1175/2007MWR1978.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Carton, J. A., H. F. Seidel, and B. S. Giese, 2012: Detecting historical ocean climate variability. J. Geophys. Res., 117, C02023, doi:10.1029/2011JC007401.

    • Search Google Scholar
    • Export Citation
  • Chavez, F. P., K. R. Buck, K. H. Coale, J. H. Martin, G. R. DiTullio, N. A. Welschmeyer, A. C. Jacobson, and R. T. Barber, 1991: Growth rates, grazing, sinking, and iron limitation of equatorial Pacific phytoplankton. Limnol. Oceanogr., 36, 18161833, doi:10.4319/lo.1991.36.8.1816.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chavez, F. P., P. G. Strutton, G. E. Friederich, R. A. Feely, G. A. Feldman, D. Foley, and M. J. McPhaden, 1999: Biological and chemical response of the equatorial Pacific Ocean to the 1997 and 1998 El Niño. Science, 286, 21262131, doi:10.1126/science.286.5447.2126.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chavez, F. P., J. Ryan, S. E. Lluch-Cota and M. C. Niquen, 2003: From anchovies to sardines and back: Multidecadal change in the Pacific Ocean. Science, 299, 217221, doi:10.1126/science.1075880.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Church, J. A., N. J. White, R. Coleman, K. Lambeck, and J. X. Mitrovica, 2004: Estimates of the regional distribution of sea level rise over the 1950–2000 period. J. Climate, 17, 26092625, doi:10.1175/1520-0442(2004)017<2609:EOTRDO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dandonneau, Y., 1992: Surface chlorophyll concentrations in the tropical Pacific: An analysis of data collected by merchant ships from 1978 to 1989. J. Geophys. Res., 97, 35813591, doi:10.1029/91JC02848.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dave, A. C., and M. S. Lozier, 2015: The impact of advection on stratification and chlorophyll variability in the equatorial Pacific. Geophys. Res. Lett., 42, 45234531, doi:10.1002/2015GL063290.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Deser, C., A. S. Phillips, and J. W. Hurrell, 2004: Pacific interdecadal climate variability: Linkages between the tropics and the North Pacific during boreal winter since 1900. J. Climate, 17, 31093124, doi:10.1175/1520-0442(2004)017<3109:PICVLB>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Drenkard, E. J., and K. B. Karnauskas, 2014: Strengthening of the Pacific Equatorial Undercurrent in the SODA reanalysis: Mechanisms, ocean dynamics, and implications. J. Climate, 27, 24052416, doi:10.1175/JCLI-D-13-00359.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fedorov, A. V., and S. G. Philander, 2000: Is El Niño changing? Science, 288, 19972002, doi:10.1126/science.288.5473.1997.

  • Feldman, G. C., D. Clark, and D. Halpern, 1984: Satellite color observations of the phytoplankton distribution in the eastern equatorial Pacific during the 1982-1983 El Niño. Science, 226, 10691071, doi:10.1126/science.226.4678.1069.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Follows, M. J., S. Dutkiewicz, S. Grant, and S. W. Chisholm, 2007: Emergent biogeography of microbial communities in a model ocean. Science, 315, 18431846, doi:10.1126/science.1138544.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gent, P. R., and M. A. Cane, 1989: A reduced gravity, primitive equation model of the upper equatorial ocean. Comput. Phys., 81, 444480, doi:10.1016/0021-9991(89)90216-7.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gierach, M. M., T. Lee, D. Turk, and M. J. McPhaden, 2012: Biological response to the 1997-98 and 2009-10 El Niño events in the equatorial Pacific Ocean. Geophys. Res. Lett., 39, L10602, doi:10.1029/2012GL051103.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gordon, H. R., 1997: Atmospheric correction of ocean color imagery in the Earth Observing System era. J. Geophys. Res., 102, 17 08117 106, doi:10.1029/96JD02443.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gu, D., and S. G. H. Philander, 1997: Interdecadal climate fluctuations that depend upon exchanges between the tropics and extratropics. Science, 275, 805807, doi:10.1126/science.275.5301.805.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hare, S. R., and N. J. Mantua, 2000: Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr., 47, 103145, doi:10.1016/S0079-6611(00)00033-1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • IPCC, 2014: Climate Change 2014: Synthesis Report. R. K. Pachauri and L. A. Meyer, Eds., IPCC, 151 pp.

  • Jackson, D., 1993: Stopping rules in principal components analysis: A comparison of heuristical and statistical approaches. Ecology, 74, 22042214, doi:10.2307/1939574.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Johnson, D. R., T. P. Boyer, H. E. Garcia, R. A. Locarnini, O. K. Baranova, and M. M. Zweng, 2009: World Ocean Database 2009 documentation. S. Levitus, Ed., NODC Internal Report 20, 175 pp.

  • Kahru, M., S. T. Gille, R. Murtugudde, P. G. Strutton, M. Manzano-Sarabia, H. Wang, and B. G. Mitchell, 2010: Global correlations between winds and ocean chlorophyll. J. Geophys. Res., 115, C12040, doi:10.1029/2010JC006500.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and et al. , 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kao, H.-Y., and J.-Y. Yu, 2009: Contrasting eastern-Pacific and central-Pacific types of ENSO. J. Climate, 22, 615632, doi:10.1175/2008JCLI2309.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kaplan, A., M. A. Cane, Y. Kushnir, A. C. Clement, M. B. Blumenthal, and B. Rajagopalan, 1998: Analyses of global sea surface temperature 1856–1991. J. Geophys. Res., 103, 18 56718 589, doi:10.1029/97JC01736.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kaplan, A., Y. Kushnir, and M. A. Cane, 2000: Reduced space optimal interpolation of historical marine sea level pressure: 1854–1992. J. Climate, 13, 29873002, doi:10.1175/1520-0442(2000)013<2987:RSOIOH>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Karnauskas, K., 2013: Can we distinguish canonical El Niño from Modoki? Geophys. Res. Lett., 40, 52465251, doi:10.1002/grl.51007.

  • Kennedy, J. J., 2014: A review of uncertainty in in situ measurements and data sets of sea surface temperature. Rev. Geophys., 52, 132, doi:10.1002/2013RG000434.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kessler, W. S., M. J. McPhaden, and K. M. Weickmann, 1995: Forcing of intraseasonal Kelvin waves in the equatorial Pacific. J. Geophys. Res. Oceans, 100, 10 61310 631, doi:10.1029/95JC00382.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Le Borgne, R., R. T. Barber, T. Delcroix, H. Y. Inoue, D. J. Mackey, and M. Rodier, 2002: Pacific warm pool and divergence: Temporal and zonal variations on the equator and their effects on the biological pump. Deep-Sea Res. II, 49, 24712512, doi:10.1016/S0967-0645(02)00045-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lee, T., and M. J. McPhaden, 2010: Increasing intensity of El Niño in the central-equatorial Pacific. Geophys. Res. Lett., 37, L14603, doi:10.1029/2010GL044007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lehodey, P., M. Bertignac, J. Hampton, A. Lewis, and J. Picaut, 1997: El Niño Southern Oscillation and tuna in the western Pacific. Nature, 389, 715718, doi:10.1038/39575.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lehodey, P., and et al. , 2006: Climate variability, fish, and fisheries. J. Climate, 19, 50095030, doi:10.1175/JCLI3898.1.

  • Leonard, C. L., and C. R. McClain, 1996: Assessment of interannual variation (1979–1986) in pigment concentrations in the tropical Pacific using the CZCS. Int. J. Remote Sens., 17, 721732, doi:10.1080/01431169608949040.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • L’Heureux, M. L., D. C. Collins, and Z.-Z. Hu, 2013: Linear trends in sea surface temperature of the tropical Pacific Ocean and implications for the El Niño–Southern Oscillation. Climate Dyn., 40, 12231236, doi:10.1007/s00382-012-1331-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, W., and et al. , 2015: Extended reconstructed sea surface temperature version 4 (ERSST.v4): Part II. Parametric and structural uncertainty estimations. J. Climate, 28, 931951, doi:10.1175/JCLI-D-14-00007.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Maes, C., J. Sudre, and V. Garçon, 2010: Detection of the eastern edge of the equatorial Pacific warm pool using satellite-based ocean color observations. SOLA, 6, 129132, doi:10.2151/sola.2010-033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mann, M. E., R. S. Bradley, and M. K. Hughes, 1998: Global-scale temperature patterns and climate forcing over the past six centuries. Nature, 392, 779787, doi:10.1038/33859.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mantua, N. J., and S. R. Hare, 2002: The Pacific decadal oscillation. J. Oceanogr., 58, 3544, doi:10.1023/A:1015820616384.

  • Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78, 10691079, doi:10.1175/1520-0477(1997)078<1069:APICOW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Martin, J. H., 1990: Glacial-interglacial CO2 change: The iron hypothesis. Paleoceanography, 5, 113, doi:10.1029/PA005i001p00001.

  • Martinez, E., D. Antoine, F. D’Ortenzio, and B. Gentili, 2009: Climate-driven basin-scale decadal oscillations of oceanic phytoplankton. Science, 326, 12531256, doi:10.1126/science.1177012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McClain, C. R., 2009: A decade of satellite ocean color observations. Annu. Rev. Mar. Sci., 1, 1942, doi:10.1146/annurev.marine.010908.163650.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McClain, C. R., G. C. Feldman, and S. B. Hooker, 2004: An overview of the SeaWiFS project and strategies for producing a climate research quality global ocean bio-optical time series. Deep-Sea Res. II, 51, 542, doi:10.1016/j.dsr2.2003.11.001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and D. Zhang, 2002: Slowdown of the meridional overturning circulation in the upper Pacific Ocean. Nature, 415, 603608, doi:10.1038/415603a.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., T. Lee, and D. McClurg, 2011: El Niño and its relationship to changing background conditions in the tropical Pacific Ocean. Geophys. Res. Lett., 38, L15709, doi:10.1029/2011GL048275.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Messié, M., and M.-H. Radenac, 2006: Seasonal variability of the surface chlorophyll in the western tropical Pacific from SeaWiFS data. Deep-Sea Res. I, 53, 15811600, doi:10.1016/j.dsr.2006.06.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Messié, M., and F. P. Chavez, 2013: Physical-biological synchrony in the global ocean associated with recent variability in the central and western equatorial Pacific. J. Geophys. Res. Oceans, 118, 37823794, doi:10.1002/jgrc.20278.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Messié, M., M.-H. Radenac, J. Lefèvre, and P. Marchesiello, 2006: Chlorophyll bloom in the western Pacific at the end of the 1997–1998 El Niño: The role of the Kiribati Islands. Geophys. Res. Lett., 33, L14601, doi:10.1029/2006GL026033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Murtugudde, R., R. Seager, and A. Busalacchi, 1996: Simulation of the tropical oceans with an ocean GCM coupled to an atmospheric mixed layer model. J. Climate, 9, 17951815, doi:10.1175/1520-0442(1996)009<1795:SOTTOW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Murtugudde, R., S. Signorini, J. Christian, A. Busalacchi, C. McClain, and J. Picaut, 1999: Ocean color variability of the tropical Indo-Pacific basin observed by SeaWiFS during 1997–1998. J. Geophys. Res., 104, 18 35118 366, doi:10.1029/1999JC900135.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • NASA OB.DAAC CZCS, 2010: Coastal Zone Color Scanner (CZCS) ocean color data, 2010.0 reprocessing. NASA Ocean Biology Distributed Active Archive Center, Goddard Space Flight Center. Accessed 26 July 2011, doi:10.5067/NIMBUS-7/CZCS_OC.2010.0.

    • Crossref
    • Export Citation
  • NASA OB.DAAC MODIS, 2010: Moderate Resolution Imaging Spectroradiometer (MODIS) Ocean Color Data, 2010.0 Reprocessing. NASA Ocean Biology Distributed Active Archive Center, Goddard Space Flight Center. Accessed 9 April 2010, doi:10.5067/AQUA/MODIS_OC.2010.0.

    • Crossref
    • Export Citation
  • NASA OB.DAAC SeaWiFS, 2010: Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Ocean Color Data, 2010.0 Reprocessing. NASA Ocean Biology Distributed Active Archive Center, Goddard Space Flight Center, Greenbelt MD. Accessed 29 September 2010, doi:10.5067/ORBVIEW-2/SEAWIFS_OC.2010.0.

    • Crossref
    • Export Citation
  • Newman, M., G. P. Compo, and M. A. Alexander, 2003: ENSO-forced variability of the Pacific decadal oscillation. J. Climate, 16, 38533857, doi:10.1175/1520-0442(2003)016<3853:EVOTPD>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Newman, M., S.-I. Shin, and M. A. Alexander, 2011: Natural variation in ENSO flavors. Geophys. Res. Lett., 38, L14705, doi:10.1029/2011GL047658.

  • Overland, J. E., and R. W. Preisendorfer, 1982: A significance test for principal components applied to a cyclone climatology. Mon. Wea. Rev., 110, 14, doi:10.1175/1520-0493(1982)110<0001:ASTFPC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Picaut, J., M. Ioualalen, C. Menkes, T. Delcroix, and M. J. McPhaden, 1996: Mechanism of the zonal displacements of the Pacific warm pool: Implications for ENSO. Science, 274, 14861489, doi:10.1126/science.274.5292.1486.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, 1992: Statistical description of data. Numerical Recipes in C: The Art of Scientific Computing. 2nd ed. Cambridge University Press, 609–655.

  • Radenac, M.-H., and M. Rodier, 1996: Nitrate and chlorophyll distributions in relation to thermohaline and current structures in the western tropical Pacific during 1985–1989. Deep-Sea Res. II, 43, 725752, doi:10.1016/0967-0645(96)00025-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Radenac, M.-H., F. Leger, A. Singh, and T. Delacroix, 2012: Sea surface chlorophyll signature in the tropical Pacific during eastern and central Pacific ENSO events. J. Geophys. Res., 117, C04007, doi:10.1029/2011JC007841.

    • Search Google Scholar
    • Export Citation
  • Robock, A., 2002: Volcanic eruption, El Chichón. The Earth System: Physical and Chemical Dimensions of Global Environmental Change, Vol. 1, M. C. MacCracken and J. S. Perry, Eds., John Wiley & Sons, 736–736.

  • Rousseaux, C. S., and W. W. Gregg, 2012: Climate variability and phytoplankton composition in the Pacific Ocean. J. Geophys. Res., 117, C10006, doi:10.1029/2012JC008083.

    • Search Google Scholar
    • Export Citation
  • Ryan, J. P., P. S. Polito, P. G. Strutton, and F. P. Chavez, 2002: Unusual large-scale phytoplankton blooms in the equatorial Pacific. Prog. Oceanogr., 55, 263285, doi:10.1016/S0079-6611(02)00137-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryan, J. P., I. Ueki, Y. Chao, H. Zhang, P. S. Polito, and F. P. Chavez, 2006: Western Pacific modulation of large phytoplankton blooms in the central and eastern equatorial Pacific. J. Geophys. Res., 111, G02013, doi:10.1029/2005JG000084.

    • Search Google Scholar
    • Export Citation
  • Rykaczewski, R., and J. P. Dunne, 2011: A measured look at ocean chlorophyll trends. Nature, 472, E5E6, doi:10.1038/nature09952.

  • Sapiano, M. R., C. W. Brown, S. Schollaert Uz, and M. Vargas, 2012: Establishing a global climatology of marine phytoplankton phenological characteristics. J. Geophys. Res., 117, C08026, doi:10.1029/2012JC007958.

    • Search Google Scholar
    • Export Citation
  • Shen, S. S. P., W. K. M. Lau, K.-M. Kim, and G. Li, 2001: A canonical ensemble correlation prediction model for seasonal precipitation anomaly. NASA/TM-2001-209989, 53 pp.

  • Shen, S. S. P., N. Tafolla, T. M. Smith, and P. A. Arkin, 2014: Multivariate regression reconstruction and its sampling error for the quasi-global annual precipitation from 1900 to 2011. J. Atmos. Sci., 71, 32503268, doi:10.1175/JAS-D-13-0301.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Signorini, S. R., C. R. McClain, and Y. Dandonneau, 1999: Mixing and phytoplankton bloom in the wake of the Marquesas Island. Geophys. Res. Lett., 26, 31213124, doi:10.1029/1999GL010470.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smerdon, J. E., A. Kaplan, D. Chang, and M. N. Evans, 2011: Corrigendum: A pseudoproxy evaluation of the CCA and RegEM methods for reconstructing climate fields of the last millennium. J. Climate, 24, 12841309, doi:10.1175/2010JCLI4110.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., and R. W. Reynolds, 2005: A global merged land–air–sea surface temperature reconstruction based on historical observations (1880–1997). J. Climate, 18, 20212036, doi:10.1175/JCLI3362.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., A. G. Barnston, M. Ji, and M. Chelliah, 1995: The impact of Pacific Ocean subsurface data on operational prediction of tropical Pacific SST at the NCEP. Wea. Forecasting, 10, 708714, doi:10.1175/1520-0434(1995)010<0708:TIOPOS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., R. W. Reynolds, R. E. Livezey, and D. C. Stokes, 1996: Reconstruction of historical sea surface temperatures using empirical orthogonal functions. J. Climate, 9, 14031420, doi:10.1175/1520-0442(1996)009<1403:ROHSST>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., R. E. Livezey, and S. S. P. Shen, 1998: An improved method for analyzing sparse and irregularly distributed SST data on a regular grid: The tropical Pacific Ocean. J. Climate, 11, 17171729, doi:10.1175/1520-0442(1998)011<1717:AIMFAS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., P. A. Arkin, and M. R. P. Sapiano, 2009: Reconstruction of near-global annual precipitation using correlations with sea surface temperature and sea level pressure. J. Geophys. Res., 114, D12107, doi:10.1029/2008JD011580.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., P. A. Arkin, L. Ren, and S. S. P. Shen, 2012: Improved reconstruction of global precipitation since 1900. J. Atmos. Oceanic Technol., 29, 15051517, doi:10.1175/JTECH-D-12-00001.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., S. S. P. Shen, L. Ren, and P. A. Arkin, 2013: Estimating monthly precipitation uncertainty beginning 1900. J. Atmos. Oceanic Technol., 30, 11071122, doi:10.1175/JTECH-D-12-00197.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sprintall, J., and M. Tomczak, 1992: Evidence of the barrier layer in the surface layer of the tropics. J. Geophys. Res., 97, 73057316, doi:10.1029/92JC00407.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • SSALTO/DUACS, 2014: SSALTO/DUACS User Handbook: MSLA and (M)ADT Near-Real Time and Delayed Time Products (reference CLS-DOS-NT-06-034). CNES, 29 pp. Accessed 15 November 2014. [Available online at http://www.aviso.altimetry.fr/fileadmin/documents/data/tools/hdbk_duacs.pdf.]

  • Trenberth, K. E., and D. P. Stepaniak, 2001: Indices of El Niño evolution. J. Climate, 14, 16971701, doi:10.1175/1520-0442(2001)014<1697:LIOENO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Turk, D., M. J. McPhaden, A. J. Busalacchi, and M. R. Lewis, 2001: Remotely sensed biological production in the equatorial Pacific. Science, 293, 471, doi:10.1126/science.1056449.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Turk, D., C. S. Meinen, D. Antoine, M. J. McPhaden, and M. R. Lewis, 2011: Implications of changing El Niño patterns for biological dynamics in the equatorial Pacific Ocean. Geophys. Res. Lett., 38, L23603, doi:10.1029/2011GL049674.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Uitz, J., H. Claustre, B. Gentili, and D. Stramski, 2010: Phytoplankton class-specific primary production in the world’s oceans: Seasonal and interannual variability from satellite observations. Global Biogeochem. Cycles, 24, GB3016, doi:10.1029/2009GB003680.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, X. J., R. Le Borgne, R. Murtugudde, A. J. Busalacchi, and M. Behrenfeld, 2008: Spatial and temporal variations in dissolved and particulate organic nitrogen in the equatorial Pacific: Biological and physical influences. Biogeosciences, 5, 17051721, doi:10.5194/bg-5-1705-2008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, X. J., M. Behrenfeld, R. Le Borgne, R. Murtugudde, and E. Boss, 2009: Regulation of phytoplankton carbon to chlorophyll ratio by light, nutrients and temperature in the equatorial Pacific Ocean: A basin-scale model. Biogeosciences, 6, 391404, doi:10.5194/bg-6-391-2009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wilson, C. and D. Adamec, 2001: Correlations between surface chlorophyll and sea surface height in the tropical Pacific during the 1997–1999 El Niño–Southern Oscillation event. J. Geophys. Res., 106, 3117531188, doi:10.1029/2000JC000724.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Woodruff, S. D., and et al. , 2011: ICOADS release 2.5: Extensions and enhancements to the surface marine meteorological archive. Int. J. Climatol., 31, 951967, doi:10.1002/joc.2103.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, J.-Y., H.-Y. Kao, T. Lee, and S. T. Kim, 2011: Subsurface ocean temperature indices for central-Pacific and eastern-Pacific types of El Niño and La Niña events. Theor. Appl. Climatol., 103, 337344, doi:10.1007/s00704-010-0307-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, J.-Y., Y. Zou, S. T. Kim, and T. Lee, 2012: The changing impact of El Niño on US winter temperatures. Geophys. Res. Lett., 39, L15702, doi:10.1029/2012GL052483.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal variability: 1900–93. J. Climate, 10, 10041020, doi:10.1175/1520-0442(1997)010<1004:ELIV>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Cross-correlations for the training period 9/1997–2008 between (a) chlorophyll (CHL) and sea surface temperature (SST), and (b) CHL and sea surface height (SSH). Contours show areas of significant correlation (r > 0.35, p < 0.01 for N* = 45).

  • View in gallery

    CCA spatial functions for (left) mode 1, which explains 18% of the total variance in CHL and (right) mode 2, which explains 13%, for (top to bottom) SST, SSH, CHL, and time-varying amplitude.

  • View in gallery

    Fraction of variance resolved (FVR) equals the cumulative variance of the first 10 modes divided by the total variance.

  • View in gallery

    Deseasonalized CHL during an east Pacific El Niño in December 1997: (top) SeaWiFS and (bottom) CCA reconstruction.

  • View in gallery

    Leave-five-out cross-validations for reconstructed CHL during the dependent SeaWiFS period (September 1997–2008). Contours indicate bins with significant correlation, p < 0.01 (69% of total area). Boxes delineate Niño-1&2 (0°–10°S, 90°–80°W), Niño-3 (5°S–5°N, 150°–90°W), Niño-3.4 (5°S–5°N, 170°–120°W), and Niño-4 (5°S–5°N, 160°E–150°W).

  • View in gallery

    Reconstructed CHL (blue) ± RMSECV (dotted blue) and SeaWiFS CHL (black) ± RMSEobs (dotted gray) averaged over (a) Niño-1&2, (b) Niño-3, (c) Niño-3.4, and (d) Niño-4. RMSEobs was calculated during the dependent period between SeaWiFS observations and cross-validated CHL reconstructions. RMSECV was estimated using the cross-validations described in the methods section (Section 3). Note that (a) has a larger range than (b)–(d).

  • View in gallery

    Reconstructed CHL across the equator as (a) longitude–time plots (5°S–5°N). The black contour, 0.1 mg m−3, demarcating the east edge of the warm pool is correlated to ONI (r = 0.85). Niño-4 (160°E–150°W) is between the dotted and solid lines; Niño-3 (150°–90°W) is eastward of the solid line. Strong La Niñas (ONI < −1) are indicated by blue arrows on the y axis, strong east Pacific El Niños (ONI > 1, EP > CP) by red arrows, and strong central Pacific El Niños (ONI > 1, CP > EP), by gold arrows, as in (b) climate indices (ONI, EP, CP). All data are smoothed over 3 months.

  • View in gallery

    Strong ENSO event zonal composites along the equator (2°S–2°N) for the east Pacific El Niños (red: ONI > 1 and EP > CP), central Pacific El Niños (gold: ONI > 1 and CP > EP), and La Niñas (blue: ONI < −1) averaged between November and January for years indicated in Table 2 with deseasonalized (a) reconstructed CHL, (b) SST, and (c) SSH. Error bars are . Niño-4 (160°E–150°W) is between the dotted and dashed lines; Niño-3 (150°–90°W) is eastward of the dashed line.

  • View in gallery

    Warm ENSO CHL anomaly composites for (a) strong east Pacific El Niños (ONI > 1 and EP > CP), (b) weak east Pacific El Niños (1 > ONI > 0.5 and EP > CP), (c) strong central Pacific El Niños (ONI > 1 and CP > EP), and (d) weak central Pacific El Niños (1 > ONI > 0.5 and CP > EP) averaged between November and January for years indicated in Table 2.

  • View in gallery

    Longitude–time plots along the equator (2°S–2°N) of deseasonalized (a) zonal wind stress (positive = westerly), (b) zonal current at 5 m (positive = eastward), and (c) 20°C isotherm depth (positive = deep). Contour, lines, and arrows as in Fig. 7. All data are smoothed over 3 months.

  • View in gallery

    Decadal average CHL anomalies for (a) 1998–2007, (b) 1988–97, (c) 1978–87, (d) 1968–77, and (e) 1958–67.

  • View in gallery

    PDO cool phase (blue: 1958–75) and warm phase (red: 1977–97) deseasonalized zonal composites along the equator (2°S–2°N) averaged between November and January for (a) CHL, (b) SST, and (c) SSH with error bars: . Niño-4 (160°E–150°W) is between the dotted and dashed lines; Niño-3 (150°–90°W) is eastward of the dashed line.

  • View in gallery

    Average differences between the warm (1977–97) and cool (1958–75) PDO regimes for (a) CHL (mg m−3), (b) SST (°C), (c) SSH (cm), (d) τx (N m−2), and (e) τ [N m−2 (104 km)−1]. Significant differences (red and blue) exceed the combined error of both regimes, + . The change in τy was insignificant and is not shown here.

  • View in gallery

    Multidecadal subsurface temperature change along the equator (2°S–2°N): PDO warm (1977–97) minus cool (1958–75), with the equatorial under current (EUC) depth-averaged over each phase and superimposed for cool (solid) and warm (dashed).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 86 86 3
PDF Downloads 64 64 11

Interannual and Decadal Variability in Tropical Pacific Chlorophyll from a Statistical Reconstruction: 1958–2008

View More View Less
  • 1 Earth System Science Interdisciplinary Center, University of Maryland, College Park, College Park, Maryland
  • | 2 Center for Satellite Applications and Research, NOAA, and Cooperative Institute for Climate and Satellites, Earth System Science Interdisciplinary Center, University of Maryland, College Park, College Park, Maryland
  • | 3 Department of Geology and Earth System Science Interdisciplinary Center, University of Maryland, College Park, College Park, Maryland
  • | 4 Center for Satellite Applications and Research, NOAA, and Cooperative Institute for Climate and Satellites, Earth System Science Interdisciplinary Center, University of Maryland, College Park, College Park, Maryland
  • | 5 NASA Goddard Space Flight Center, Greenbelt, Maryland
© Get Permissions
Full access

Abstract

Historical understanding of marine biological dynamics has been limited by sparse in situ observations and the fact that dedicated ocean color satellite remote sensing only began in 1997. From these observations, it has become clear that physical oceanography controls biological variability over seasonal to interannual time scales. To quantify how multidecadal, climate-scale patterns impact biological productivity, the strong correlation with sea surface temperature and sea surface height is utilized to reconstruct a retrospective 51-yr time series of surface chlorophyll, the pigment measured by ocean color satellites. The canonical correlation analysis statistical reconstruction demonstrates greatest skill away from land and within about 10° of the equator where chlorophyll variance is greatest and predominantly associated with El Niño–Southern Oscillation dynamics. Differences in chlorophyll patterns between east or central Pacific El Niño events are observed, with larger declines east of 180° for east Pacific events and west of 180° for central Pacific events. Additionally, small but significant decadal variations in chlorophyll patterns are observed corresponding to the Pacific decadal oscillation. Decadal changes in chlorophyll west of 180° are consistent with increased stratification, whereas changes between 110°–140°W may be related to long-term shoaling of the nutrient-bearing equatorial undercurrent.

Current affiliation: Global Science and Technology, Inc., Greenbelt, Maryland.

Current affiliation: University Corporation for Atmospheric Research, Boulder, Colorado.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Stephanie Schollaert Uz, stephanie.uz@nasa.gov

Abstract

Historical understanding of marine biological dynamics has been limited by sparse in situ observations and the fact that dedicated ocean color satellite remote sensing only began in 1997. From these observations, it has become clear that physical oceanography controls biological variability over seasonal to interannual time scales. To quantify how multidecadal, climate-scale patterns impact biological productivity, the strong correlation with sea surface temperature and sea surface height is utilized to reconstruct a retrospective 51-yr time series of surface chlorophyll, the pigment measured by ocean color satellites. The canonical correlation analysis statistical reconstruction demonstrates greatest skill away from land and within about 10° of the equator where chlorophyll variance is greatest and predominantly associated with El Niño–Southern Oscillation dynamics. Differences in chlorophyll patterns between east or central Pacific El Niño events are observed, with larger declines east of 180° for east Pacific events and west of 180° for central Pacific events. Additionally, small but significant decadal variations in chlorophyll patterns are observed corresponding to the Pacific decadal oscillation. Decadal changes in chlorophyll west of 180° are consistent with increased stratification, whereas changes between 110°–140°W may be related to long-term shoaling of the nutrient-bearing equatorial undercurrent.

Current affiliation: Global Science and Technology, Inc., Greenbelt, Maryland.

Current affiliation: University Corporation for Atmospheric Research, Boulder, Colorado.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Stephanie Schollaert Uz, stephanie.uz@nasa.gov

1. Introduction

Primary production in the ocean plays an important role in the function of marine ecosystems and Earth’s carbon cycle with a need for a long-term synoptic view of its distribution and variation. Global ocean surface chlorophyll (CHL), a proxy for phytoplankton standing stock, has been continuously derived from ocean color satellites for almost two decades (McClain 2009). Satellite retrievals of CHL have indicated a strong connection between ocean physics and biology in the tropical Pacific and provided new insights into the factors driving seasonal and interannual biological processes, especially those associated with El Niño–Southern Oscillation (ENSO) activity (Chavez et al. 1999; Turk et al. 2001; Wilson and Adamec 2001; Ryan et al. 2002; Chavez et al. 2003; Behrenfeld et al. 2006; Follows et al. 2007; Martinez et al. 2009; Kahru et al. 2010). There are, however, many unanswered questions about the variability of large-scale phytoplankton distributions at time scales longer than a decade because long-term observations are discontinuous and subject to substantial uncertainty (e.g., Martinez et al. 2009; Boyce et al. 2010; Rykaczewski and Dunne 2011). In this study, we wish to place recent biological variability and trends into a multidecadal context.

The coupling of nutrient availability to strong atmospheric and oceanic forcing in the tropical Pacific makes it an ideal region to exploit the correlation between physics and biology to develop a multidecadal reconstruction and study slow changes that may be happening at the base of the marine food web. There, primary production is not limited by photosynthetically active radiation but primarily by upwelling of nutrient-rich waters. The nutrient content of equatorial Pacific surface waters is characterized by two climatological mean features: a mesotrophic, upwelling cold tongue that extends from the east toward the center and an oligotrophic, low-salinity, surface warm pool in the west (e.g., Le Borgne et al. 2002). The west Pacific warm pool is a stable feature above a salinity barrier layer that inhibits exchange between the isothermal, isohaline mixed layer and water below (Sprintall and Tomczak 1992). Variability in tropical Pacific CHL primarily arises from the basin-scale modification of the west–east thermocline slope during ENSO that impacts nutrient supply and biological productivity across the basin (Barber and Chavez 1983; Radenac and Rodier 1996; Leonard and McClain 1996; Chavez et al. 1999; Ryan et al. 2002; Kahru et al. 2010) as well as phytoplankton community composition (Feldman et al. 1984; Lehodey et al. 1997; Chavez et al. 1999; Uitz et al. 2010; Rousseaux and Gregg 2012). Wind-driven upwelling supplies nutrients to the equatorial surface layer along with other mechanisms, such as vertical and horizontal advection (Dave and Lozier 2015), turbulent vertical mixing (Ryan et al. 2002), the turbulent effect of islands (Signorini et al. 1999; Messié and Radenac 2006; Messié et al. 2006), changes in the New Guinea coastal currents that supply iron to the equatorial undercurrent (Ryan et al. 2002, 2006), and the erosion of the west Pacific warm pool barrier layer (Murtugudde et al. 1999).

During the ENSO cool phase (La Niña), stronger easterly winds push surface water westward, increasing the zonal thermocline slope across the basin. Enhanced equatorial upwelling lifts deep nutrients into the photic zone in the east, resulting in higher biological productivity throughout the cold tongue. During the ENSO warm phase (El Niño), westerly wind bursts in the west Pacific result in anomalous Ekman convergence and generate equatorial Kelvin waves (Kessler et al. 1995; Picaut et al. 1996): surface currents advect the nutrient-poor warm pool eastward, deepening the thermocline and reducing biological productivity in the east (Murtugudde et al. 1999; Ryan et al. 2002; Kahru et al. 2010). Simultaneously, a shoaling of the thermocline in the west results in higher surface CHL there. El Niño events with strong east Pacific expression cause a flattening of the zonal slope in the thermocline depth (Ashok et al. 2007; Kao and Yu 2009; Lee and McPhaden 2010; Yu et al. 2012). By contrast, central Pacific El Niño events have more localized anomalies with limited eastward extent (Ashok et al. 2007; McPhaden et al. 2011; Yu et al. 2012; Karnauskas 2013). Study of the biological impact of diverse ENSO events has been limited to the few events since continuous ocean color measurements began in 1997, but it is implied that central Pacific events have a localized effect on primary productivity associated with a weakened vertical nutrient supply while east Pacific events experience a reduction of biological productivity across the cold tongue (Turk et al. 2011; Radenac et al. 2012; Gierach et al. 2012; Messié and Chavez 2013).

Additionally, several studies have suggested that decadal climate oscillations strengthen or weaken ENSO-driven biological changes (Mantua et al. 1997; Hare and Mantua 2000; Trenberth and Stepaniak 2001; Mantua and Hare 2002; Chavez et al. 2003; Lehodey et al. 2006). Basinwide cool conditions evident in the decades before 1976 were characterized by stronger easterly trade winds, a steeper zonal pycnocline slope along the equator, enhanced upper ocean meridional overturning circulation, and greater upwelling along the equator (Mantua et al. 1997; McPhaden and Zhang 2002), similar to ENSO cold phase conditions. Conversely, the basinwide warm conditions after 1976 were characterized by weakened easterly trade winds, a relaxed zonal pycnocline slope, slower meridional overturning circulation, and decreased equatorial upwelling (McPhaden and Zhang 2002). However, we lack a continuous ocean color record that is long enough to resolve corresponding decadal and secular changes in phytoplankton distributions.

The objective of this study is to reconstruct multidecadal, basinwide CHL to improve our understanding of the ways in which synoptic-scale variability impacts ocean ecosystems through bottom-up controls, meaning food supply as opposed to predation. We use canonical correlation analysis (CCA) to identify patterns linking physical variables to phytoplankton blooms and extend the surface CHL record back over five decades. Statistical reconstruction techniques quantify robust linear relationships among two or more fields of data and can be used either to fill gaps in a record (e.g., Dandonneau 1992; Alvera-Azcárate et al. 2007) or to extend a record using longer proxy variables (Bretherton et al. 1992; Barnett and Preisendorfer 1987; Smith et al. 1996; Kaplan et al. 1998, 2000; Church et al. 2004; Smith and Reynolds 2005; Smith et al. 2009). While statistical reconstructions are increasingly applied to extend physical variables back in time by more than a century, the method has not yet been used to reconstruct marine biological data back in time due to the complex, nonlinear dynamics of ecosystems and the episodic nature of phytoplankton blooms. In the tropical Pacific where physics exerts the primary large-scale, low-frequency control on biology, we show that a skillful statistical reconstruction of a biological variable is feasible. We describe the training and validation data in section 2. The methodology is described and the reconstructed CHL fields are evaluated and their uncertainty is quantified in section 3. In section 4 we use the reconstructed fields to study the interannual and decadal variability of CHL patterns in the tropical Pacific during 1958–2008.

2. Data

The CCA reconstruction method requires complete simultaneous fields of CHL and physical proxies with which to train the method over the dependent period. The longer proxy record is then applied to generate CHL fields over the independent period for which CHL is not available. In section 2a we provide background on the satellite-derived CHL and ocean data assimilation datasets used to train the reconstruction method and detail their pretreatment. In section 2b we select the physical proxies. In section 2c we describe historically observed, remotely sensed, and simulated CHL estimates outside of the statistical training interval used to validate the reconstructed CHL. All datasets used in this study are listed in Table 1.

Table 1.

Datasets used in this study, their details, and sources. An asterisk (*) indicates primary datasets used for training and reconstruction. Other datasets were used to validate, check for consistency, or interpret the results of the reconstruction. Note that dataset duration years include all 12 months, except SeaWiFS, which began in September (1997) and MODIS/Aqua, July (2002).

Table 1.

a. Training data

Sea-viewing Wide Field-of-View Sensor (SeaWiFS) and Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua level 3 data of monthly CHL at 9-km resolution from September 1997 through December 2008 were obtained from NASA’s Ocean Biology Processing Group (Table 1). The 2010 Ocean Color reprocessing included corrections and intercalibrations for both SeaWiFS and MODIS/Aqua (NASA OB.DAAC SeaWiFS 2010; NASA OB.DAAC MODIS 2010). Only the SeaWiFS sensor was designed exclusively for biological ocean science, resulting in a relatively simple and very stable sensor, attaining the highest-quality measurements necessary for climate research, with MODIS/Aqua data considered the next best (McClain et al. 2004; McClain 2009). SeaWiFS is primarily used here with MODIS/Aqua filling occasional gaps toward the end of the record. Any remaining missing data were filled using a 3 × 3 median filter, repeated three times, although the first pass was usually sufficient. The decorrelation scales of monthly CHL in this region range from 20° for the zonal decorrelation scale to 5°–8° for the meridional. We found a temporal decorrelation scale of 2.5 months. Because log-normally transforming CHL gives too much weight to the edge of the west Pacific warm pool, we use linear calculations previously found to be more suitable for the oligotrophic waters of the tropics and subtropics (Sapiano et al. 2012). To focus on large-scale features and minimize mesoscale and finer features, the data were latitude-weighted and spatially averaged over 2° bins centered on the equator. An expanded land mask excludes pixels within 2° of the coast. We temporally smoothed over three months using a 1–2–1 weighted moving average to retain important low-frequency information while minimizing high-frequency features.

The Simple Ocean Data Assimilation (SODA) version 2.1.6 (Carton and Giese 2008; Carton et al. 2012) includes complete fields between 1958 and 2008 of three-dimensional temperature, salt, zonal and meridional currents (u, υ), zonal and meridional wind stress (τx, τy), and sea surface height (SSH) (Table 1). This version applies the Geophysical Fluid Dynamics Laboratory modular Parallel Ocean Program model version 2.1, forced by surface wind stress from ERA-40 and QuikSCAT winds. SODA is constrained by the constant assimilation of hydrographic data from World Ocean Database 2009 and the International Comprehensive Ocean–Atmosphere Dataset (Woodruff et al. 2011), namely temperatures, salinities, and sea level. Various oceanic observations used within SODA provide complimentary information that improve each other’s accuracy using an optimal data assimilation technique (Carton et al. 2012). From these variables, the depth of the 20°C isotherm (T20) was calculated as a proxy for the thermocline depth in this region (Turk et al. 2001). As with CHL, all SODA products were also latitude-weighted and spatially averaged over 2° bins centered on the equator and temporally smoothed over three months.

b. Proxy selection

CHL was cross-correlated with a variety of physical variables expected to be associated with nutrient entrainment leading to phytoplankton blooms. The strongest correlations are between CHL and sea surface temperature (SST) (Fig. 1a), significant (|r| > 0.35, p < 0.01) (Press et al. 1992) over more than 60% of the tropical Pacific and strongest along the equator (average r = −0.64 with 98% of the bins significant) due to upwelling and the lateral advection of warm, unproductive waters compared to cooler, productive waters (Barber and Chavez 1983; Chavez et al. 1999; Ryan et al. 2002; Dave and Lozier 2015). Thus, SST is an obvious choice for a predictor. Additional predictors were compared with the goal of finding one that best complements SST as an indicator of the flux of nutrients into the surface layer. After testing many other variables, the best choice for the second predictor was SSH due to its consistent correlation to CHL across most of the tropical Pacific (i.e., along the equator, average r = −0.54 with 84% of the bins significantly correlated). SSH was previously found to have a high correlation with the thermocline depth in the equatorial Pacific, except between 165°E and 165°W due to lateral advection and the pivot point of the thermocline between the warm pool and cold tongue (e.g., Turk et al. 2001). Increased SSH is associated with a thicker surface layer, deeper nutricline, and decreased CHL. Over most of the tropical Pacific, SST and SSH covary in response to dynamical processes that impact the entire surface layer such as vertical entrainment of deep water through mixing and upwelling. Additionally, SSH adds independent information about mechanisms that impact nutrient flux to the surface layer through subsurface processes. By adding SSH as the second predictor, the reconstruction increases the number of bins successfully reconstructed by 9% over just using SST alone, with the improved fidelity most notable in the equatorial cold tongue east of 160°W where, for example, subsurface Kelvin waves can lift nutrients into the photic zone without being detected in the SST signature.

Fig. 1.
Fig. 1.

Cross-correlations for the training period 9/1997–2008 between (a) chlorophyll (CHL) and sea surface temperature (SST), and (b) CHL and sea surface height (SSH). Contours show areas of significant correlation (r > 0.35, p < 0.01 for N* = 45).

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

Data pretreatment and observational error were estimated and found to contribute minimally to uncertainty in the final results (cf. section 3b). Using published uncertainty estimates for CHL, SST, and SSH (Bailey and Werdell 2006; Kennedy 2014; SSALTO/DUACS 2014, respectively), we estimate observational uncertainties in the binned, smoothed data of up to 1% of the CHL value, 4% of the SST value, and less than 3% of the SSH value over the tropical Pacific.

c. Validation data

To validate the reconstruction, independent CHL fields not used for training were pretreated in the same manner as the training data. Unfortunately in situ data for biological parameters are more sparse than for physical parameters. The National Oceanographic Data Center (NODC) World Ocean Database 2009 includes about 10 000 casts containing CHL in the tropical Pacific between 1958 and 2008 (Johnson et al. 2009) (Table 1). Limiting observations to those made at depths less than 20 m that reflect surface conditions yields 27 540 data points.

In addition to limited in situ observations, the Coastal Zone Color Scanner (CZCS) CHL observations were acquired from NASA’s Ocean Biology Processing Group from the 2010 Ocean Color reprocessing (NASA OB.DAAC CZCS 2010) (Table 1). The CZCS ocean color satellite sensor operated between October 1978 and June 1986 as part of a shared mission whose primary focus was to establish that biological variables could be observed from space in the coastal zone of the United States, with more observations near the coast than in the open ocean and minimal coverage elsewhere. In the tropical Pacific, CZCS had less than 20% of the coverage of SeaWiFS near the coast, decreasing to about 10% in the open ocean, with the least coverage in Niño-4. Furthermore, in 1982 the El Chichón volcano erupted in Mexico. Its gases and particles encircled Earth between the equator and 30°N for more than six months and then spread widely (Robock 2002), likely introducing artifacts into the satellite measurements, as it is difficult to correct for the biases in CZCS data associated with extreme aerosol events (Gordon 1997).

Spatially complete simulated CHL fields from a fully coupled three-dimensional physical–biogeochemical model are used as a consistency check (Table 1). The biogeochemical simulation of Wang et al. (2009) was developed at basin scale for the tropical Pacific between 30°S and 30°N for 1988–2007. This model has been shown to have good fidelity for simulating ecosystem dynamics and variation in biogeochemical fields (Wang et al. 2008). Its ocean general circulation model is based on a primitive equation, sigma-coordinate model coupled to an advective atmospheric mixed layer model (Gent and Cane 1989; Murtugudde et al. 1996) forced by 6-day mean surface wind stress from the National Centers for Environmental Prediction reanalysis (Kalnay et al. 1996), and climatological monthly mean solar radiation, cloudiness, and precipitation. Simulated CHL was calculated using a nonsteady carbon-to-CHL ratio, which is a function of irradiance, nitrate, iron, and temperature (Wang et al. 2008; Wang et al. 2009).

3. Reconstruction and validation

The combined covariance between the dominant modes of normalized CHL and physical variables, SST and SSH, was used to train the reconstruction algorithm. Section 3a details the method of statistical reconstruction. In section 3b we test the reconstruction through cross-validation during the dependent period and through comparison to independent datasets and model output during the independent period. We estimate uncertainty in reconstructed CHL in section 3c.

a. Canonical correlation analysis reconstruction

The CCA method is used to quantify associations between different sets of variables by finding maximum correlations between linear combinations of two datasets. We use the CCA method in EOF spectral space that was developed and described by Barnett and Preisendorfer (1987). In the present context, the leading EOF modes define a coordinate system that optimizes the correlation between CHL, SST, and SSH (Shen et al. 2001; Smith et al. 2009, 2012, 2013). A similar method, the multivariate EOF (MEOF) approach, gives equivalent results that are slightly better during the dependent period. However, the CCA performs slightly better during the independent period and is thus reported here. Qualitatively, the CCA decomposes the predictor and predictand data into a truncated subset of principal components. The cross-covariance of predictor and predictand patterns defines canonical maps that are weighted and applied to project the predictand (CHL) using the predictor (SST and SSH) data. CCA identifies patterns of maximum correlation across the entire domain that may be associated with recurring large-scale processes, and is not suited to identifying finescale patterns.

The number of EOF modes applied to the CCA reconstruction is determined by how much variance is contained in the modes. Methods to determine how many predictor-predictand patterns are significantly distinguished from random variation include Rule N (Overland and Preisendorfer 1982) and the bootstrapped eigenvalue-eigenvector tests (Jackson 1993). To apply Rule N, we generated 100 random datasets of the same space–time dimension of the actual dataset, and applied EOF separately to each to generate a null eigenvalue spectrum, with significance estimated from the 95th percentile of the randomly generated values. Similarly, for the bootstrapped eigenvalue test, each EOF was bootstrapped 100 times. Significant eigenvalues have 95% confidence intervals that do not overlap. Eigenvalues with overlapping 95% confidence intervals are considered indistinguishable from noise. A weakness of statistical reconstructions, such as the CCA, is that they filter out variations in higher modes and also cannot represent variations absent from the base period. In addition, biases in the base data may be reflected in the CCA analysis. For these reasons, we only use CCA to resolve large-scale variations associated with recurring physical processes that are represented by the predictor variables (SST and SSH).

Prior to performing EOF decomposition and CCA reconstruction, data pretreatment of CHL, SST, and SSH included removing the seasonal climatology for each bin over the training period (September 1997–December 2008). The three variables were normalized and nondimensionalized by subtracting the remaining average of each bin from its monthly anomalies and dividing by its standard deviation. The normalization was reversed after CCA reconstruction to yield dimensional CHL estimates. Monthly climatology was then recalculated and subtracted in order to center the anomalies on the reconstruction era (1958–2008).

The EOF analysis involves training spatial correlation functions of the predictand (CHL) and predictor (SST and SSH, merged) fields during the training period. EOFs are calculated on the deseasonalized, normalized pairs of the predictand and predictors. Optimal representation of CHL in terms of SST and SSH is obtained by first forming linear combinations in EOF spectral space of their canonical component vectors, uj and vk:
e1
The components of the canonical map vectors, gj and hk, are averages over time. In this case, because the data are normalized, the operator yields the correlations between SST and SSH or CHL and their respective canonical component time series (j or k) at a given location (x or x′). The number of predictor and predictand spatial points is p and q, respectively. A linear combination of the canonical component time series uj of the predictor dataset (SST&SSH) is used to reconstruct the n-dimensional predictand vector, CHL(x′, t) = [CHL(x′, 1, … t(x′, n)]T, where T denotes transpose. Projecting the CHL vectors onto the q-dimensional vector space spanned by the first q of the uj, with j = 1, …, q < p:
e2
where q″ < q is the number of canonical modes retained for the reconstruction. The number of modes retained can be determined by observing the q″ value that yields the maximum reconstruction skill. Further details, including a full mathematical description of the CCA method, are given in the appendix of Barnett and Preisendorfer (1987).

We found 15 EOF modes to be significant by Rule N and the bootstrap eigenvalue–eigenvector test, explaining 75% of the variance in the CHL alone and 80% of the combined variance in CHL, SST, and SSH. However, the leading 10 modes produced lowest validation RMSE and highest cross-validation correlations [cf. section 3b(1)] between the original SeaWiFS CHL and reconstructed CHL over the Niño areas and the entire tropical Pacific; thus, we used 10 modes in our reconstruction.

The first CCA mode (Fig. 2, left) explains 18% of the total variance in the CCA; the second mode (Fig. 2, right) explains 13%. The spatial patterns in mode 2 are similar to the canonical ENSO patterns while mode 1 has the horseshoe shape often associated with decadal variability and the central Pacific ENSO (Trenberth and Stepaniak 2001; Ashok et al. 2007; Kao and Yu 2009). Mode 2 reflects El Niños in 1997/98 and 2002/03 weakly in 2004/05 and 2006/07 and La Niñas in 1998/99–2000/01, 2005/06, and 2007/08. The next five modes suggest higher-frequency variations of the two main modes and together explain about 30% of the total variance. By modes 8, 9, and 10, the EOF amplitudes approach zero, have little basin-scale structure, and together explain less than 10% of the total variance. Excluding high-frequency patterns adds negligible uncertainty to the climate-scale patterns, consistent with other studies (Ballabrera-Poy et al. 2003; Liu et al. 2015). For the CCA method, CHL variance explained in the first 10 CCA modes is 71% using combined SST and SSH as predictors. Thus, 29% of the variability is excluded prior to calculating the reconstruction.

Fig. 2.
Fig. 2.

CCA spatial functions for (left) mode 1, which explains 18% of the total variance in CHL and (right) mode 2, which explains 13%, for (top to bottom) SST, SSH, CHL, and time-varying amplitude.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

How well the EOF modes are defined during the training period and capture variation in the reconstruction period determines the method uncertainty (Shen et al. 2014). We calculated an a priori estimate of the potential of a reconstruction that uses the dominant EOF modes to reproduce the signal in the original data by dividing the variance of the dominant 10 EOF modes by the variance of the original CHL data to yield the fraction of variance resolved (FVR) for each bin: FVR = σ2(CHL10 modes) / σ2(CHL). The spatial distribution of the FVR gives a good estimate of CHL variability that can be reconstructed by the 10 most dominant modes (Fig. 3). This plot shows that a monthly CHL reconstruction using 10 modes is expected to capture most of the variance within 10° of the equator and more than 80% of the total variance between 5°S and 5°N west of 140°W. The first 10 EOF modes do not adequately capture the high variance near coastal margins or the low variance in the oligotrophic subtropical gyres.

Fig. 3.
Fig. 3.

Fraction of variance resolved (FVR) equals the cumulative variance of the first 10 modes divided by the total variance.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

b. Validation and consistency testing

A visual inspection of the reconstructed CHL with the original SeaWiFS data during individual months indicates that the statistical method reconstructs major spatial patterns, evident in any given month and shown here for December 1997 (Fig. 4), the peak of the strongest El Niño of the training period. As expected, the reconstruction is most skillful in locations where the FVR is highest (i.e., along the equator). Elsewhere, most notably beyond 10° north and south of the cold tongue, the reconstruction misses some of the positive CHL anomalies. To quantify the skill of the CCA reconstruction, CHL reconstructions were cross-validated against a central month of CHL observations excluded from the training period [section 3b(1)]. The reconstructed CHL was also compared to available in situ observations, with in situ data mapped to EOF spectral space and CZCS ocean color estimates [section 3b(2)]. We then tested the reconstruction for consistency with CHL simulated by a fully coupled biogeochemical model [section 3b(3)].

Fig. 4.
Fig. 4.

Deseasonalized CHL during an east Pacific El Niño in December 1997: (top) SeaWiFS and (bottom) CCA reconstruction.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

1) Error estimates by cross-validation

The method of choice for assessing statistical model performance, cross-validation, is to withhold a portion of the training data and perform out-of-sample comparisons with reconstructed values (Barnett and Preisendorfer 1987; Smith et al. 1996; Mann et al. 1998; Smerdon et al. 2011). We tested the sensitivity of the cross-validation to the length of the validation interval and found that leaving five months out gives the best balance between enough months retained to train a robust reconstruction and enough months excluded to ensure the central month had minimal input to the reconstruction and was as statistically independent as possible. A map of “leave five out” cross-validations shows that 69% of bins achieve significant validation (r > 0.35 at p < 0.01), most notably away from land and generally within 10° of the equator (Fig. 5). Cross-validated CHL reconstructions were averaged over the four Niño areas and compared to SeaWiFS CHL (Table 2). The strongest correlations (r > 0.85) are found just east of the date line in western Niño-3.4 and eastern Niño-4. The main differences between the original and reconstructed CHL fields are in the magnitude of variation during ENSO events, as expected in a reconstruction that only use the first 10 EOF modes. The Niño-1&2 coastal area includes more variability than the dominant EOF modes can capture, as demonstrated by the lower fraction of variance resolved there and lower cross-validation correlation in Table 1.

Fig. 5.
Fig. 5.

Leave-five-out cross-validations for reconstructed CHL during the dependent SeaWiFS period (September 1997–2008). Contours indicate bins with significant correlation, p < 0.01 (69% of total area). Boxes delineate Niño-1&2 (0°–10°S, 90°–80°W), Niño-3 (5°S–5°N, 150°–90°W), Niño-3.4 (5°S–5°N, 170°–120°W), and Niño-4 (5°S–5°N, 160°E–150°W).

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

Table 2.

Cross-correlations of the CHL reconstructed through CCA using SST&SSH, unless noted with SeaWiFS over the training period; CCA using SST only with SeaWiFS over the training period; during the validation period with CZCS CHL; and as a consistency check with the modeled CHL (Wang et al. 2009) that shows skill west of 140°W (Niño-4). Significant correlations are r > 0.35 (p < 0.01, N* = 45). Note that Niño areas are listed west (left) to east (right). An asterisk (*) indicates that comparisons with SeaWiFS training data are leave-five-out cross-validations.

Table 2.

Additionally, we calculated cross-validations to compare the improvement gained by adding SSH as a predictor. CHL reconstructions that solely use SST have 60% of bins significantly correlated to the original SeaWiFS CHL. Over the four Niño areas, CHL reconstructed using SST only compare most closely to SeaWiFS CHL in Niño-4 and least in the cold tongue east of 160°W where using two predictors improves validation (Table 2b). The improvement with SSH as the second predictor in Niño-1&2, -3, and -3.4 suggests that subsurface nutrient entrainment processes by ocean dynamics and diabatic effects are important in this region, while mixing and advection primarily control nutrient delivery to the surface in Niño-4.

2) Validation against observations

The training data used in the reconstruction are large area averages sampled by satellites and composited over the month. In situ data are point source measurements collected over discrete intramonthly intervals too short to adequately capture monthly averages and too spatially sparse to represent the 2° bin. The paucity of in situ observations made them an inappropriate choice for validation studies. It has been shown that EOFs can be used to regress in situ data onto spatial patterns and improve their skill compared to traditional in situ only analysis (Smith et al. 1996, 1998). When there were at least three observations in a bin, the sparse data were interpolated to a regular 2° grid using the first five EOF modes, which explain 66% of the total variance of the in situ dataset. After interpolating the sparse in situ data using the EOF method, there was still too much scatter to yield statistically significant correlations with the reconstructions. Only a rough correspondence is evident between the in situ and reconstructed CHL (r ~ 0.3 in Niño-3 and -3.4). The difference in the two collection methods and amount of data being averaged makes their comparison a rough approximation. That they are within an order of magnitude of each other is encouraging.

All available CZCS ocean color data were averaged over the Niño regions and compared to the reconstructed CHL. As detailed earlier, the CZCS ocean color mission had a primary focus on the coastal zone with limited coverage elsewhere. There were 50% more observations in Niño-1&2 compared to Niño-4 and 92 months of match-ups with the reconstruction (15%). Cross-validations between CZCS and reconstructed CHL are best in Niño-1&2 and degrade away from land (Table 2c). CZCS observations in 1982 reveal positive CHL anomalies in the equatorial Niño areas corresponding to a weak La Niña in 1981/82. A strong El Niño followed in 1982/83 with a pronounced decrease in CZCS CHL in 1982/83 in Niño-1&2 (not shown). CZCS CHL matches up closer to the reconstruction than in situ CHL, but data paucity and possible atmospheric correction issues render its validation problematic.

3) Consistency testing against simulation output

As a check for consistency, complete fields of simulated CHL (cf. section 2c) were compared to SeaWiFS CHL during the dependent period (September 1997–December 2007) and to the reconstruction during the independent period (January 1988–August 1997). The simulated CHL compares best to both SeaWiFS and reconstructed CHL within 10° of the equator and west of 140°W. Over the entire tropical Pacific, 22% of simulation versus SeaWiFS bins were significantly correlated (r > 0.5, p < 0.001), and 20% for simulation versus reconstruction. The time series were averaged over each Niño area for the simulated CHL and original CHL during the dependent period and the reconstruction with the simulation for the independent period. Where the model shows skill compared to SeaWiFS west of 140°W, the simulated and reconstructed CHL correlate well (e.g., Niño-4, r = 0.87) (Table 2d).

c. Uncertainty analysis

The main uncertainty in the statistical reconstruction stems from the method of reducing the data to the dominant modes, which immediately excludes 29% of the variability. Differences between the original data and the cross-validated reconstruction were used to calculate RMSE in normalized units to assess whether we successfully reconstruct a signal that exceeds the noise of the uncertainties. For the independent period, cross-validations were used to estimate the average normalized RMSE (Smith et al. 1995, 2012, 2013). During the dependent period, uncertainty calculated as the difference between SeaWiFS observed CHL and leave-five-out reconstructed CHL cross-validations (normalized RMSE or nRMSEobs) ranged from 0.24 in the equatorial Niño areas to nRMSEobs = 0.6 for Niño-1&2. Cross-validation estimates for the independent period are within acceptable limits for open ocean, equatorial areas (nRMSECV~0.5 in Niño-3, -3.4, and -4), but unacceptably large toward the coasts (nRMSECV~0.9 in Niño-1&2). Unnormalized reconstructed CHL (Fig. 6) demonstrates high fidelity in capturing general patterns, notably ENSO, but damps their amplitudes, occasionally exceeding the margin of error (e.g., in 1998/99 in Niño-3, -3.4, and -1&2; 2006/07 in Niño-3.4 and -4).

Fig. 6.
Fig. 6.

Reconstructed CHL (blue) ± RMSECV (dotted blue) and SeaWiFS CHL (black) ± RMSEobs (dotted gray) averaged over (a) Niño-1&2, (b) Niño-3, (c) Niño-3.4, and (d) Niño-4. RMSEobs was calculated during the dependent period between SeaWiFS observations and cross-validated CHL reconstructions. RMSECV was estimated using the cross-validations described in the methods section (Section 3). Note that (a) has a larger range than (b)–(d).

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

In summary, reconstructed CHL has smallest RMSE and highest validation skill away from land and toward the equator where dynamics exert the primary control on biology. Compelling support for this statistical reconstruction of biology based upon physical proxies is its consistency with the completely independent physical–biogeochemical simulation, which relies upon atmospheric forcing and uses none of the observations in the reconstruction (i.e., SSH, SST, and CHL). The advantage of the statistical reconstruction over the fully coupled biogeochemical simulation is that it is purely empirical and incorporates dominant mechanisms, both known and unknown, and may be studied for further discovery of new mechanisms. The biogeochemical simulation only includes known, modeled mechanisms. The principal uncertainty in the reconstructed CHL fields arises from the limited number of patterns and variance resolved in the training interval. Ten patterns explain about two-thirds of the total variance in the tropical Pacific CHL field, resulting in most of the uncertainty.

4. Results and discussion

With these limitations in mind, the reconstructed CHL variations are analyzed for interannual and decadal patterns across the tropical Pacific over five decades within 5° of the equator (Niño-3, -3.4, and -4 areas) where the reconstruction has the most skill. Section 4a describes the dominant interannual patterns in the reconstruction associated with ENSO, namely La Niña and east or central Pacific El Niño. Section 4b discusses evidence of CHL variability corresponding to a weak decadal pattern, with a cool PDO regime from the beginning of the reconstruction until 1976, followed by a warm PDO regime until 1998 when the pattern switched back to cool.

a. Interannual chlorophyll variability

Basinwide reconstructed CHL averaged along the equator (5°S–5°N), smoothed over three months, is plotted between 150°E and 90°W in a Hovmöller diagram over five decades (Fig. 7a). The 0.1 mg m−3 contour demarcates the eastern edge of the oligotrophic waters of the west Pacific warm pool from the productive waters of the cold tongue (Maes et al. 2010; Radenac et al. 2012). Variations in its zonal displacement are primarily associated with ENSO amplitude and phase, warm or cold (Fig. 7b).

Fig. 7.
Fig. 7.

Reconstructed CHL across the equator as (a) longitude–time plots (5°S–5°N). The black contour, 0.1 mg m−3, demarcating the east edge of the warm pool is correlated to ONI (r = 0.85). Niño-4 (160°E–150°W) is between the dotted and solid lines; Niño-3 (150°–90°W) is eastward of the solid line. Strong La Niñas (ONI < −1) are indicated by blue arrows on the y axis, strong east Pacific El Niños (ONI > 1, EP > CP) by red arrows, and strong central Pacific El Niños (ONI > 1, CP > EP), by gold arrows, as in (b) climate indices (ONI, EP, CP). All data are smoothed over 3 months.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

The Ocean Niño Index (ONI) by the National Oceanic and Atmospheric Administration Climate Prediction Center is one index among many used to represent the interannual ENSO signal, based on a threshold of ±0.5°C warm or cold anomaly in the Niño-3.4 area (5°N–5°S, 120°–170°W) from the centered 30-yr average and lasting a minimum of five consecutive months (Trenberth and Stepaniak 2001; L’Heureux et al. 2013). While the ONI characterizes all El Niño events, the distinction between El Niños centered in the east or central Pacific and CP El Niños is captured in the east and central Pacific Niño indices (EP and CP, respectively) (Kao and Yu 2009; Yu et al. 2012). As with the datasets used in this study, all climate indices were smoothed over three months to emphasize the seasonal to multidecadal variations.

We categorize strong ENSO events as |ONI| > 1 for at least five consecutive months, and weak events as 0.5 < |ONI| < 1. We further distinguish El Niño events as east (EP > CP) or central (CP > EP). It has been noted that strong La Niña events are always strongest in the central Pacific (Yu et al. 2011), so they are not distinguished here by their EP/CP type. During the reconstruction period (Table 3), strong La Niña events began in 1973, 1975, 1988, 1998, and 2007. Strong east Pacific El Niño events started in 1972, 1982, and 1997. Strong central Pacific El Niño events started in 1957, 1963, 1965, 1986, 1991, and 2002. Our classifications are consistent with those previously reported in the literature (Ashok et al. 2007; Kao and Yu 2009; Yu et al. 2011; Yu et al. 2012), with occasional minor exceptions because we smooth indices over three months and classify an event by its most stringent qualification. For example, Ashok et al. (2007) classified the same CP (Modoki) events, plus additional events that began in 1979, 1990, and 1992 when the indices did not meet the 5-month threshold that we used. Our strong EP events are consistent with Yu et al. (2011), but they classify 1991/92 as a strong EP (this study: strong CP), include a few redundancies (e.g., 1972/73 listed as strong EP and weak CP), and only find weak CP events.

Table 3.

ENSO events: defined using the 3-month average Ocean Niño Index (ONI) when |ONI| > 0.5 for at least five consecutive months: strong if |ONI| > 1 and weak otherwise. East (central) Pacific El Niños are defined using the indices of Kao and Yu (2009).

Table 3.

In Fig. 7, the zonal distribution of CHL across the basin associated with EP and CP El Niño and La Niña events is shown over five decades. Basinwide reductions in CHL are primarily evident for EP events that began in 1972, 1982, and 1997 and also for several CP events that began in 1958, 1965, 1986, 1987, and 1991. The greatest CHL increases due to La Niña were felt in the west Pacific near 160°E during the events that began in 1973 and 1975. The strong La Niña events of 1988, 1998, and 1999 had more of a basinwide impact on biology, whereas the strong event that began in 2007 was weaker than in 1973 and 1975 but otherwise showed a similar pattern of biggest CHL increase near 160°E.

Compositing all strong events when ENSO typically peaks between November and January and meridionally averaging along the equator (2°S–2°N) reveals zonal differences between the three phases (Fig. 8). Cold phase ENSO corresponds to positive CHL anomalies and negative SST and SSH (Figs. 8a–c, respectively). The area of highest CHL variability during strong La Niñas is near 160°E in western Niño-4, likely due to lateral advection (Turk et al. 2001; Dave and Lozier 2015). Interestingly, SST and SSH anomalies are minimal there, collocated with the thermocline east–west pivot point where thermocline variability is weak. The other high CHL anomaly during strong La Niñas is near 90°W, on the west side of the Galapagos Islands, where pronounced upwelling is known to shoal the nutricline into the photic zone (Signorini et al. 1999; Wilson and Adamec 2001). Both east and central Pacific El Niños are associated with decreased CHL (Fig. 8a) and increased SST and SSH (Figs. 8b and 8c, respectively), although the longitudinal distributions of their CHL anomalies differ. EP El Niños are associated with decreased CHL eastward of 170°E with lowest values in the eastern Niño-3 corresponding to highest SST and SSH anomalies due to thermocline variability (Turk et al. 2011). During a CP El Niño, biological productivity decreased across the basin, with noticeably less CHL between 160°E and 180° compared to an EP event. East of 180°, productivity during CP events does not decline as much as for the EP events, but this difference also corresponds to weaker dynamics reflected in the SST and SSH anomalies. There is more reduction in basinwide average CHL during east Pacific events compared to central Pacific events (−0.051 vs −0.028 mg m−3, respectively), although Turk et al. (2011) hypothesized that this may be due to the strength of the events. Central Pacific events occur more frequently, but are weaker. Central Pacific events average ONI = 0.9 for the 14 central Pacific events compared to average ONI = 1.5 for the five east Pacific events. Cross-correlations with the EP and CP climate indices (Table 4) indicate that CHL has its strongest correlation with the EP in Niño-3 (r = −0.54) and with CP in Niño-4 (r = −0.75).

Fig. 8.
Fig. 8.

Strong ENSO event zonal composites along the equator (2°S–2°N) for the east Pacific El Niños (red: ONI > 1 and EP > CP), central Pacific El Niños (gold: ONI > 1 and CP > EP), and La Niñas (blue: ONI < −1) averaged between November and January for years indicated in Table 2 with deseasonalized (a) reconstructed CHL, (b) SST, and (c) SSH. Error bars are . Niño-4 (160°E–150°W) is between the dotted and dashed lines; Niño-3 (150°–90°W) is eastward of the dashed line.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

Table 4.

Monthly median reconstructed chlorophyll anomalies along the equator (5°S–5°N, longitude indicated) cross-correlated with monthly climate indices between 1958 and 2008, all smoothed over three months. All correlations are significant (|r| > 0.14, p < 0.01, N = 200). Note that Niño areas are listed west (left) to east (right).

Table 4.

The biological impact of different ENSO events was previously explored in the ocean color satellite record since 1997 (Turk et al. 2011; Radenac et al. 2012; Gierach et al. 2012; Messié and Chavez 2013). During this reconstruction era, we classified three strong and two weak EP events and seven strong and seven weak CP events, whereas previous studies were limited to one strong EP (1997/98), one strong CP (2002/03), and several weak events, with the weak 2006/07 event categorized as an EP by one study (Turk et al. 2011) and CP by another (Radenac et al. 2012). The strong CP event that began in 2002 was anomalous in that it experienced a strong CHL reduction between 160°E and 160°W but weaker basinwide CHL reductions than earlier strong CP events in the reconstructed CHL (Fig. 7). The spatial patterns of maximum variability are qualitatively consistent with distributions in Turk et al. (2011) and cluster analysis comparisons by Radenac et al. (2012) for La Niña events, strong CP El Niño, and strong EP El Niño. Contrasting spatial averages for strong and weak events clarifies differences between EP and CP (Fig. 9). A clear spatial difference between EP and CP events is evident in these maps. EP events extend basinwide and have a stronger negative impact on equatorial CHL anomalies in Niño-3 and -3.4. The CP events impact Niño-4 west of 180°, but CHL anomalies there do not relate to the index strength. These results imply that different mechanisms impact biology during EP events compared to CP events.

Fig. 9.
Fig. 9.

Warm ENSO CHL anomaly composites for (a) strong east Pacific El Niños (ONI > 1 and EP > CP), (b) weak east Pacific El Niños (1 > ONI > 0.5 and EP > CP), (c) strong central Pacific El Niños (ONI > 1 and CP > EP), and (d) weak central Pacific El Niños (1 > ONI > 0.5 and CP > EP) averaged between November and January for years indicated in Table 2.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

To determine the atmospheric or oceanic forcing linked to changes in biological productivity, we analyzed several additional variables related to advection, mixing, and the depth of the thermocline/nutricline. Similarly to Radenac et al. (2012) but over the 51 years of the reconstruction period, in Fig. 10 we show deseasonalized zonal wind stress (τx), zonal ocean current at 5 m (U), and the depth of the 20°C isotherm (Z20) as a proxy for the thermocline depth (cf. section 2b). During strong EP El Niño events, positive (eastward) anomalies in surface current and wind stress extend east of 150°W (albeit weakly in 1972) along with a basinwide deepening of the thermocline and nutricline that corresponds to decreased CHL at the surface (Fig. 7). By contrast, anomalies associated with CP El Niño events are weaker and do not extend as far to the east, although there is localized deepening of the thermocline and lower CHL that sometimes extends across the basin (e.g., 1987). The subsurface view (Fig. 10) supports the idea that biological productivity has different forcing mechanisms for EP and CP events. For the EP events, there is a basinwide deepening of the thermocline and subsequent reduction in equatorial upwelling. The CP events experience localized westerly wind forcing, weaker eastward surface current perturbation, and thermocline deepening that is generally confined west of 180°. Strong and weak CP events have similar westerly wind stress anomalies in Niño-4 while surface current and thermocline depth changes are variable and inconsistent. This implies that winds cause localized decreased nutrient flux and suppression of biology in the vicinity of 160°E–180° during CP events rather than basin-scale oceanographic forcing.

Fig. 10.
Fig. 10.

Longitude–time plots along the equator (2°S–2°N) of deseasonalized (a) zonal wind stress (positive = westerly), (b) zonal current at 5 m (positive = eastward), and (c) 20°C isotherm depth (positive = deep). Contour, lines, and arrows as in Fig. 7. All data are smoothed over 3 months.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

b. Decadal chlorophyll variability

A unique benefit of the current analysis is that multidecadal variability in basin-scale biological patterns can be examined for the first time. Decadally averaged CHL shows coherent patterns that shifted between positive and negative anomalies over the five decades (Fig. 11). The first decade, 1958–67 (Fig. 11e), was closest to the 51-yr climatology everywhere except for high CHL near the Costa Rica Dome and low CHL off Peru in Niño-1&2. The patterns intensified during 1968–77 (Fig. 11d) with high CHL over most of the region from the Costa Rica Dome westward to 160°E in a wishbone shape north and south of the equator, except for low CHL in Niño-1&2 and Niño-3. The patterns effectively switched during the following two decades, 1978–87 (Fig. 11c) and 1988–97 (Fig. 11b), with most of the region experiencing low CHL while Niño-1&2 and Niño-3 had high values. During 1998–2007 (Fig. 11a) the CHL anomalies switched again but differently: high CHL prevailed east of 180°, especially in the Costa Rica Dome and in Niño-1&2, while western Niño-4 had low CHL.

Fig. 11.
Fig. 11.

Decadal average CHL anomalies for (a) 1998–2007, (b) 1988–97, (c) 1978–87, (d) 1968–77, and (e) 1958–67.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

The switching of the decadal CHL patterns coincided with phase changes of the Pacific decadal oscillation (PDO) around 1976 and 1998 (Zhang et al. 1997; Mantua et al. 1997; Hare and Mantua 2000; Chavez et al. 2003). We composited reconstructed CHL, SST, and SSH between November and January, when the PDO is typically strongest, for cool phase (1958–75) and warm phase (1977–97) conditions to compare interdecadal changes along the equator (2°S–2°N) (Fig. 12). West of 180°, SST warmed, SSH became shallower, and CHL decreased, implying that increased stratification of the surface waters above the salinity barrier layer inhibited nutrient flux into the photic zone and lowered biological productivity there. In the east between 110° and 140°W, however, significant differences in CHL and SST are both higher, contrary to the usual inverse relationship. The spatial distribution of differences between PDO phases (warm minus cool) show the CHL decrease west of 180° along the equator and more broadly away from the equator as well as the increase between 110°–140°W along the equator (Fig. 13a); also, SST warmed over most of the region (Fig. 13b) and SSH increased away from the equator, consistent with lower CHL there, but became thinner toward the equator west of 160°W (Fig. 13c). Whether the multidecadal changes in CHL correspond to local wind forcing or subsurface circulation changes is explored next.

Fig. 12.
Fig. 12.

PDO cool phase (blue: 1958–75) and warm phase (red: 1977–97) deseasonalized zonal composites along the equator (2°S–2°N) averaged between November and January for (a) CHL, (b) SST, and (c) SSH with error bars: . Niño-4 (160°E–150°W) is between the dotted and dashed lines; Niño-3 (150°–90°W) is eastward of the dashed line.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

Fig. 13.
Fig. 13.

Average differences between the warm (1977–97) and cool (1958–75) PDO regimes for (a) CHL (mg m−3), (b) SST (°C), (c) SSH (cm), (d) τx (N m−2), and (e) τ [N m−2 (104 km)−1]. Significant differences (red and blue) exceed the combined error of both regimes, + . The change in τy was insignificant and is not shown here.

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

Nutrient flux from depth toward the sunlit surface is often wind driven, and thus we consider wind variables first. Tropical Pacific winds are predominately zonal and easterly, causing equatorial divergence and upwelling. Just off the equator, wind stress curl causes vertical motion through the divergence of Ekman flow and may drive nutrients toward the surface. Note that τx and both experienced significant changes between the cool (1958–75) and warm eras (1977–97). Basin-averaged differences between PDO phases (warm minus cool) show limited areas of significant change in zonal wind stress (Fig. 13d): more positive (westerly) between 110°–140°W along the equator and farther east to the south, but more negative (easterly) off the equator in the west. Positive τX between 110° and 140°W would support equatorial downwelling, inconsistent with the increase in CHL there (Fig. 13a). Wind stress curl changed significantly over a broader area between PDO eras and became more cyclonic just north of the equator (Fig. 13e), which implies more westerly winds at the equator causing reduced upwelling. Reduced upwelling would also be consistent with lower CHL around 160°E, although stratification above the salinity barrier layer is more effective for trapping nutrients at depth. While the wind patterns inhibited equatorial upwelling and may have had a secondary effect upon the decreased CHL near 160°E, these comparisons indicate that the CHL trend along the equator around 110°–140°W was not primarily wind driven.

To determine whether the CHL differences at 110°–140°W between the PDO phases relate to subsurface oceanographic changes, we compared subsurface temperatures and currents that relate to nutrient flux toward the surface. In particular, the Equatorial Undercurrent (EUC) transports critical nutrients from a depth of about 200 m in the west Pacific, rising as it transits eastward across the basin to support biological production through the vertical flux of iron in the central and east Pacific (Martin 1990; Chavez et al. 1991; Chavez et al. 1999; Wilson and Adamec 2001; Ryan et al. 2006). We calculated the average depth of the EUC as the depth of maximum zonal current between 2°N and 2°S. The subsurface temperature change between the PDO phases (warm minus cool) and the average depth of the EUC for each phase are shown in Fig. 14. Subsurface temperature changes in the upper ocean (<100 m) were consistent with the SST change of warmer temperatures during the warm phase east of 160°E. Between 110° and 140°W, the average EUC depth shoaled by about 25 m during the warm phase compared to the earlier cool phase. Shoaling of the EUC over the twentieth century was previously reported (Drenkard and Karnauskas 2014). We infer that such a long-term change to the EUC could lift critical nutrients from depth into the photic zone sooner in the current’s eastward transit to promote biological production farther west along the equator. This subtle change represents a significant multidecadal anomaly with implications for the structure of local marine ecosystems. More analysis of critical nutrients is needed to ascertain whether this is the case.

Fig. 14.
Fig. 14.

Multidecadal subsurface temperature change along the equator (2°S–2°N): PDO warm (1977–97) minus cool (1958–75), with the equatorial under current (EUC) depth-averaged over each phase and superimposed for cool (solid) and warm (dashed).

Citation: Journal of Climate 30, 18; 10.1175/JCLI-D-16-0202.1

5. Summary and conclusions

Taking advantage of the fact that physical oceanography controls biological variability over seasonal to interannual time scales in the tropical Pacific, we have created the first multidecadal, basinwide view of CHL between 1958 and 2008 through statistical reconstruction using its strong correlation with SST and SSH. The CCA reconstruction reproduces the majority of the variability in the 2°, monthly resolution CHL time series, over 69% of the tropical Pacific and nearly all of the equatorial Niño areas (e.g., Fig. 4). Independent observations are limited, but ocean color observations by CZCS match up best to the reconstructed CHL where there is greater data density toward the coast (Table 2c). Reconstructed CHL is consistent with independently simulated output from a realistically forced, fully coupled ocean–biogeochemistry model (Table 2d), suggesting that large-scale biological variability can be reproduced using SST and SSH in the equatorial Pacific where dynamics exerts the primary control on biology. Cross-validation confirms that using the combination of SST and SSH for the reconstruction reproduces CHL patterns with reasonable signal above noise (Table 2a; Figs. 5 and 6) and gives better results than using SST as the sole predictor (Table 2b). The fraction of variance resolved by the reconstruction depends upon the strength of the correlations between CHL and the predictors during the training period (Fig. 3).

We have analyzed reconstructed CHL to understand patterns of basin-scale, low-frequency biological response to interannual and multidecadal climate variation. While the impact of ENSO on biology has been observed in the existing ocean color record, this five-decade reconstruction adds more events. Five east Pacific El Niño events (three strong, two weak) and 14 central Pacific El Niño events (seven strong, seven weak) caused large-scale reductions in CHL with increased SST and SSH characteristically different across event types. The greatest basinwide impact on biology occurred during east Pacific El Niño events (Figs. 9a,b), with the biological impact proportional to the physical strength. In contrast, the biological signature during the central Pacific events was not proportional to their strength (Figs. 9c,d). Averaging all strong El Niño events confirms that CP events depressed biological production about 20° farther west and with less basinwide reduction toward the east than EP events, corresponding to weaker changes across the basin in SST, SSH (Fig. 8) as well as τx, U, and Z20 (Fig. 10). Decadal CHL variation had a smaller but discernable magnitude (Fig. 11) and appears to correspond to variations in the PDO. Although the PDO warm phase is generally associated with lower CHL along the equator, especially west of 180°, consistent primarily with increased stratification and secondarily with decreased upwelling, there is a region between 110° and 140°W for which CHL appears to increase as a consequence of subsurface oceanographic changes (Fig. 12). Specifically, shoaling by the EUC during the PDO warm phase likely lifted nutrients into the photic zone. The shoaling EUC seems to be associated with a long-term trend in CHL in the east that is different from the trend farther west where the nutricline is well below the photic zone.

In the tropical Pacific, we find patterns of basinwide multidecadal CHL variation corresponding to slow oceanic changes over this 51-yr reconstruction. The extent to which ENSO is changing as the climate warms remains unclear (Fedorov and Philander 2000; Newman et al. 2003; Newman et al. 2011; IPCC 2014), yet interdecadal oscillations appear to impact ENSO (e.g., Gu and Philander 1997; Deser et al. 2004). Previous studies predict long-term changes in the extent of low versus high CHL regions and that the inferred influence on productivity will impact phytoplankton biomass and ocean ecosystems (e.g., Behrenfeld et al. 2006; Martinez et al. 2009; Boyce et al. 2010). Our study suggests that multidecadal changes are impacting the biology of the equatorial Pacific.

Acknowledgments

Discussions with Jim Carton and Kayo Ide at the University of Maryland improved the robustness of these results. Three anonymous reviewers and the editor provided beneficial suggestions that clarified and shaped this paper into its current form and we are very grateful for their collective advice. We gratefully acknowledge the Ocean Biology Processing Group at NASA Goddard Space Flight Center for the production and distribution of the ocean color data used here: SeaWiFS, MODIS/Aqua, CZCS, and the NODC for collection and distribution of in situ observations. We thank Semyon Grodsky and Gennady Chepurin at the University of Maryland for help obtaining the SODA reanalysis fields as part of Jim Carton’s group, who contribute to their development and continued refinement and improvement. Wendy Wang provided the output of her fully coupled biogeochemical model that was used as a consistency check of the reconstruction. Part of this research was supported by the National Oceanic and Atmospheric Administration, U.S. Department of Commerce, Grant NA09NES4400006. Additional support was provided by the National Aeronautics and Space Administration Goddard Space Flight Center Earth Sciences Division. Page charges are supported by Global Science and Technology, Inc.

REFERENCES

  • Alvera-Azcárate, A., A. Barth, J.-M. Beckers, and R. H. Weisberg, 2007: Multivariate reconstruction of missing data in sea surface temperature, chlorophyll and wind satellite fields. J. Geophys. Res., 112, C03008, doi:10.1029/2006JC003660.

    • Search Google Scholar
    • Export Citation
  • Ashok, K., S. K. Behera, S. A. Rao, H. Weng, and T. Yamagata, 2007: El Niño Modoki and its possible teleconnection. J. Geophys. Res., 112, C11007, doi:10.1029/2006JC003798.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bailey, S. W., and P. J. Werdell, 2006: A multi-sensor approach for the on-orbit validation of ocean color satellite data products. Remote Sens. Environ., 102, 1223, doi:10.1016/j.rse.2006.01.015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ballabrera-Poy, J., R. G. Murtugudde, J. R. Christian, and A. J. Busalacchi, 2003: Signal-to-noise ratios of observed monthly tropical ocean color. Geophys. Res. Lett., 30, 1645, doi:10.1029/2003GL016995.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barber, R. T., and F. P. Chavez, 1983: Biological consequences of El Niño. Science, 222, 12031210, doi:10.1126/science.222.4629.1203.

  • Barnett, T. P., and R. Preisendorfer, 1987: Origins and levels of monthly and seasonal forecast skill for United States surface air temperatures determined by canonical correlation analysis. Mon. Wea. Rev., 115, 18251850, doi:10.1175/1520-0493(1987)115<1825:OALOMA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Behrenfeld, M. J., and et al. , 2006: Climate-driven trends in contemporary ocean productivity. Nature, 444, 752755, doi:10.1038/nature05317.

  • Boyce, D. G., M. R. Lewis, and B. Worm, 2010: Global phytoplankton decline over the past century. Nature, 466, 591596, doi:10.1038/nature09268.

  • Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate, 5, 541560, doi:10.1175/1520-0442(1992)005<0541:AIOMFF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Carton, J. A., and B. S. Giese, 2008: A reanalysis of ocean climate using Simple Ocean Data Assimilation (SODA). Mon. Wea. Rev., 136, 29993017, doi:10.1175/2007MWR1978.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Carton, J. A., H. F. Seidel, and B. S. Giese, 2012: Detecting historical ocean climate variability. J. Geophys. Res., 117, C02023, doi:10.1029/2011JC007401.

    • Search Google Scholar
    • Export Citation
  • Chavez, F. P., K. R. Buck, K. H. Coale, J. H. Martin, G. R. DiTullio, N. A. Welschmeyer, A. C. Jacobson, and R. T. Barber, 1991: Growth rates, grazing, sinking, and iron limitation of equatorial Pacific phytoplankton. Limnol. Oceanogr., 36, 18161833, doi:10.4319/lo.1991.36.8.1816.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chavez, F. P., P. G. Strutton, G. E. Friederich, R. A. Feely, G. A. Feldman, D. Foley, and M. J. McPhaden, 1999: Biological and chemical response of the equatorial Pacific Ocean to the 1997 and 1998 El Niño. Science, 286, 21262131, doi:10.1126/science.286.5447.2126.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chavez, F. P., J. Ryan, S. E. Lluch-Cota and M. C. Niquen, 2003: From anchovies to sardines and back: Multidecadal change in the Pacific Ocean. Science, 299, 217221, doi:10.1126/science.1075880.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Church, J. A., N. J. White, R. Coleman, K. Lambeck, and J. X. Mitrovica, 2004: Estimates of the regional distribution of sea level rise over the 1950–2000 period. J. Climate, 17, 26092625, doi:10.1175/1520-0442(2004)017<2609:EOTRDO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dandonneau, Y., 1992: Surface chlorophyll concentrations in the tropical Pacific: An analysis of data collected by merchant ships from 1978 to 1989. J. Geophys. Res., 97, 35813591, doi:10.1029/91JC02848.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dave, A. C., and M. S. Lozier, 2015: The impact of advection on stratification and chlorophyll variability in the equatorial Pacific. Geophys. Res. Lett., 42, 45234531, doi:10.1002/2015GL063290.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Deser, C., A. S. Phillips, and J. W. Hurrell, 2004: Pacific interdecadal climate variability: Linkages between the tropics and the North Pacific during boreal winter since 1900. J. Climate, 17, 31093124, doi:10.1175/1520-0442(2004)017<3109:PICVLB>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Drenkard, E. J., and K. B. Karnauskas, 2014: Strengthening of the Pacific Equatorial Undercurrent in the SODA reanalysis: Mechanisms, ocean dynamics, and implications. J. Climate, 27, 24052416, doi:10.1175/JCLI-D-13-00359.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fedorov, A. V., and S. G. Philander, 2000: Is El Niño changing? Science, 288, 19972002, doi:10.1126/science.288.5473.1997.

  • Feldman, G. C., D. Clark, and D. Halpern, 1984: Satellite color observations of the phytoplankton distribution in the eastern equatorial Pacific during the 1982-1983 El Niño. Science, 226, 10691071, doi:10.1126/science.226.4678.1069.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Follows, M. J., S. Dutkiewicz, S. Grant, and S. W. Chisholm, 2007: Emergent biogeography of microbial communities in a model ocean. Science, 315, 18431846, doi:10.1126/science.1138544.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gent, P. R., and M. A. Cane, 1989: A reduced gravity, primitive equation model of the upper equatorial ocean. Comput. Phys., 81, 444480, doi:10.1016/0021-9991(89)90216-7.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gierach, M. M., T. Lee, D. Turk, and M. J. McPhaden, 2012: Biological response to the 1997-98 and 2009-10 El Niño events in the equatorial Pacific Ocean. Geophys. Res. Lett., 39, L10602, doi:10.1029/2012GL051103.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gordon, H. R., 1997: Atmospheric correction of ocean color imagery in the Earth Observing System era. J. Geophys. Res., 102, 17 08117 106, doi:10.1029/96JD02443.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gu, D., and S. G. H. Philander, 1997: Interdecadal climate fluctuations that depend upon exchanges between the tropics and extratropics. Science, 275, 805807, doi:10.1126/science.275.5301.805.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hare, S. R., and N. J. Mantua, 2000: Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr., 47, 103145, doi:10.1016/S0079-6611(00)00033-1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • IPCC, 2014: Climate Change 2014: Synthesis Report. R. K. Pachauri and L. A. Meyer, Eds., IPCC, 151 pp.

  • Jackson, D., 1993: Stopping rules in principal components analysis: A comparison of heuristical and statistical approaches. Ecology, 74, 22042214, doi:10.2307/1939574.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Johnson, D. R., T. P. Boyer, H. E. Garcia, R. A. Locarnini, O. K. Baranova, and M. M. Zweng, 2009: World Ocean Database 2009 documentation. S. Levitus, Ed., NODC Internal Report 20, 175 pp.

  • Kahru, M., S. T. Gille, R. Murtugudde, P. G. Strutton, M. Manzano-Sarabia, H. Wang, and B. G. Mitchell, 2010: Global correlations between winds and ocean chlorophyll. J. Geophys. Res., 115, C12040, doi:10.1029/2010JC006500.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and et al. , 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kao, H.-Y., and J.-Y. Yu, 2009: Contrasting eastern-Pacific and central-Pacific types of ENSO. J. Climate, 22, 615632, doi:10.1175/2008JCLI2309.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kaplan, A., M. A. Cane, Y. Kushnir, A. C. Clement, M. B. Blumenthal, and B. Rajagopalan, 1998: Analyses of global sea surface temperature 1856–1991. J. Geophys. Res., 103, 18 56718 589, doi:10.1029/97JC01736.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kaplan, A., Y. Kushnir, and M. A. Cane, 2000: Reduced space optimal interpolation of historical marine sea level pressure: 1854–1992. J. Climate, 13, 29873002, doi:10.1175/1520-0442(2000)013<2987:RSOIOH>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Karnauskas, K., 2013: Can we distinguish canonical El Niño from Modoki? Geophys. Res. Lett., 40, 52465251, doi:10.1002/grl.51007.

  • Kennedy, J. J., 2014: A review of uncertainty in in situ measurements and data sets of sea surface temperature. Rev. Geophys., 52, 132, doi:10.1002/2013RG000434.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kessler, W. S., M. J. McPhaden, and K. M. Weickmann, 1995: Forcing of intraseasonal Kelvin waves in the equatorial Pacific. J. Geophys. Res. Oceans, 100, 10 61310 631, doi:10.1029/95JC00382.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Le Borgne, R., R. T. Barber, T. Delcroix, H. Y. Inoue, D. J. Mackey, and M. Rodier, 2002: Pacific warm pool and divergence: Temporal and zonal variations on the equator and their effects on the biological pump. Deep-Sea Res. II, 49, 24712512, doi:10.1016/S0967-0645(02)00045-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lee, T., and M. J. McPhaden, 2010: Increasing intensity of El Niño in the central-equatorial Pacific. Geophys. Res. Lett., 37, L14603, doi:10.1029/2010GL044007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lehodey, P., M. Bertignac, J. Hampton, A. Lewis, and J. Picaut, 1997: El Niño Southern Oscillation and tuna in the western Pacific. Nature, 389, 715718, doi:10.1038/39575.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lehodey, P., and et al. , 2006: Climate variability, fish, and fisheries. J. Climate, 19, 50095030, doi:10.1175/JCLI3898.1.

  • Leonard, C. L., and C. R. McClain, 1996: Assessment of interannual variation (1979–1986) in pigment concentrations in the tropical Pacific using the CZCS. Int. J. Remote Sens., 17, 721732, doi:10.1080/01431169608949040.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • L’Heureux, M. L., D. C. Collins, and Z.-Z. Hu, 2013: Linear trends in sea surface temperature of the tropical Pacific Ocean and implications for the El Niño–Southern Oscillation. Climate Dyn., 40, 12231236, doi:10.1007/s00382-012-1331-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, W., and et al. , 2015: Extended reconstructed sea surface temperature version 4 (ERSST.v4): Part II. Parametric and structural uncertainty estimations. J. Climate, 28, 931951, doi:10.1175/JCLI-D-14-00007.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Maes, C., J. Sudre, and V. Garçon, 2010: Detection of the eastern edge of the equatorial Pacific warm pool using satellite-based ocean color observations. SOLA, 6, 129132, doi:10.2151/sola.2010-033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mann, M. E., R. S. Bradley, and M. K. Hughes, 1998: Global-scale temperature patterns and climate forcing over the past six centuries. Nature, 392, 779787, doi:10.1038/33859.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mantua, N. J., and S. R. Hare, 2002: The Pacific decadal oscillation. J. Oceanogr., 58, 3544, doi:10.1023/A:1015820616384.

  • Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78, 10691079, doi:10.1175/1520-0477(1997)078<1069:APICOW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Martin, J. H., 1990: Glacial-interglacial CO2 change: The iron hypothesis. Paleoceanography, 5, 113, doi:10.1029/PA005i001p00001.

  • Martinez, E., D. Antoine, F. D’Ortenzio, and B. Gentili, 2009: Climate-driven basin-scale decadal oscillations of oceanic phytoplankton. Science, 326, 12531256, doi:10.1126/science.1177012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McClain, C. R., 2009: A decade of satellite ocean color observations. Annu. Rev. Mar. Sci., 1, 1942, doi:10.1146/annurev.marine.010908.163650.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McClain, C. R., G. C. Feldman, and S. B. Hooker, 2004: An overview of the SeaWiFS project and strategies for producing a climate research quality global ocean bio-optical time series. Deep-Sea Res. II, 51, 542, doi:10.1016/j.dsr2.2003.11.001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and D. Zhang, 2002: Slowdown of the meridional overturning circulation in the upper Pacific Ocean. Nature, 415, 603608, doi:10.1038/415603a.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., T. Lee, and D. McClurg, 2011: El Niño and its relationship to changing background conditions in the tropical Pacific Ocean. Geophys. Res. Lett., 38, L15709, doi:10.1029/2011GL048275.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Messié, M., and M.-H. Radenac, 2006: Seasonal variability of the surface chlorophyll in the western tropical Pacific from SeaWiFS data. Deep-Sea Res. I, 53, 15811600, doi:10.1016/j.dsr.2006.06.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Messié, M., and F. P. Chavez, 2013: Physical-biological synchrony in the global ocean associated with recent variability in the central and western equatorial Pacific. J. Geophys. Res. Oceans, 118, 37823794, doi:10.1002/jgrc.20278.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Messié, M., M.-H. Radenac, J. Lefèvre, and P. Marchesiello, 2006: Chlorophyll bloom in the western Pacific at the end of the 1997–1998 El Niño: The role of the Kiribati Islands. Geophys. Res. Lett., 33, L14601, doi:10.1029/2006GL026033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Murtugudde, R., R. Seager, and A. Busalacchi, 1996: Simulation of the tropical oceans with an ocean GCM coupled to an atmospheric mixed layer model. J. Climate, 9, 17951815, doi:10.1175/1520-0442(1996)009<1795:SOTTOW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Murtugudde, R., S. Signorini, J. Christian, A. Busalacchi, C. McClain, and J. Picaut, 1999: Ocean color variability of the tropical Indo-Pacific basin observed by SeaWiFS during 1997–1998. J. Geophys. Res., 104, 18 35118 366, doi:10.1029/1999JC900135.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • NASA OB.DAAC CZCS, 2010: Coastal Zone Color Scanner (CZCS) ocean color data, 2010.0 reprocessing. NASA Ocean Biology Distributed Active Archive Center, Goddard Space Flight Center. Accessed 26 July 2011, doi:10.5067/NIMBUS-7/CZCS_OC.2010.0.

    • Crossref
    • Export Citation
  • NASA OB.DAAC MODIS, 2010: Moderate Resolution Imaging Spectroradiometer (MODIS) Ocean Color Data, 2010.0 Reprocessing. NASA Ocean Biology Distributed Active Archive Center, Goddard Space Flight Center. Accessed 9 April 2010, doi:10.5067/AQUA/MODIS_OC.2010.0.

    • Crossref
    • Export Citation
  • NASA OB.DAAC SeaWiFS, 2010: Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Ocean Color Data, 2010.0 Reprocessing. NASA Ocean Biology Distributed Active Archive Center, Goddard Space Flight Center, Greenbelt MD. Accessed 29 September 2010, doi:10.5067/ORBVIEW-2/SEAWIFS_OC.2010.0.

    • Crossref
    • Export Citation
  • Newman, M., G. P. Compo, and M. A. Alexander, 2003: ENSO-forced variability of the Pacific decadal oscillation. J. Climate, 16, 38533857, doi:10.1175/1520-0442(2003)016<3853:EVOTPD>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Newman, M., S.-I. Shin, and M. A. Alexander, 2011: Natural variation in ENSO flavors. Geophys. Res. Lett., 38, L14705, doi:10.1029/2011GL047658.

  • Overland, J. E., and R. W. Preisendorfer, 1982: A significance test for principal components applied to a cyclone climatology. Mon. Wea. Rev., 110, 14, doi:10.1175/1520-0493(1982)110<0001:ASTFPC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Picaut, J., M. Ioualalen, C. Menkes, T. Delcroix, and M. J. McPhaden, 1996: Mechanism of the zonal displacements of the Pacific warm pool: Implications for ENSO. Science, 274, 14861489, doi:10.1126/science.274.5292.1486.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, 1992: Statistical description of data. Numerical Recipes in C: The Art of Scientific Computing. 2nd ed. Cambridge University Press, 609–655.

  • Radenac, M.-H., and M. Rodier, 1996: Nitrate and chlorophyll distributions in relation to thermohaline and current structures in the western tropical Pacific during 1985–1989. Deep-Sea Res. II, 43, 725752, doi:10.1016/0967-0645(96)00025-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Radenac, M.-H., F. Leger, A. Singh, and T. Delacroix, 2012: Sea surface chlorophyll signature in the tropical Pacific during eastern and central Pacific ENSO events. J. Geophys. Res., 117, C04007, doi:10.1029/2011JC007841.

    • Search Google Scholar
    • Export Citation
  • Robock, A., 2002: Volcanic eruption, El Chichón. The Earth System: Physical and Chemical Dimensions of Global Environmental Change, Vol. 1, M. C. MacCracken and J. S. Perry, Eds., John Wiley & Sons, 736–736.

  • Rousseaux, C. S., and W. W. Gregg, 2012: Climate variability and phytoplankton composition in the Pacific Ocean. J. Geophys. Res., 117, C10006, doi:10.1029/2012JC008083.

    • Search Google Scholar
    • Export Citation
  • Ryan, J. P., P. S. Polito, P. G. Strutton, and F. P. Chavez, 2002: Unusual large-scale phytoplankton blooms in the equatorial Pacific. Prog. Oceanogr., 55, 263285, doi:10.1016/S0079-6611(02)00137-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryan, J. P., I. Ueki, Y. Chao, H. Zhang, P. S. Polito, and F. P. Chavez, 2006: Western Pacific modulation of large phytoplankton blooms in the central and eastern equatorial Pacific. J. Geophys. Res., 111, G02013, doi:10.1029/2005JG000084.

    • Search Google Scholar
    • Export Citation
  • Rykaczewski, R., and J. P. Dunne, 2011: A measured look at ocean chlorophyll trends. Nature, 472, E5E6, doi:10.1038/nature09952.

  • Sapiano, M. R., C. W. Brown, S. Schollaert Uz, and M. Vargas, 2012: Establishing a global climatology of marine phytoplankton phenological characteristics. J. Geophys. Res., 117, C08026, doi:10.1029/2012JC007958.

    • Search Google Scholar
    • Export Citation
  • Shen, S. S. P., W. K. M. Lau, K.-M. Kim, and G. Li, 2001: A canonical ensemble correlation prediction model for seasonal precipitation anomaly. NASA/TM-2001-209989, 53 pp.

  • Shen, S. S. P., N. Tafolla, T. M. Smith, and P. A. Arkin, 2014: Multivariate regression reconstruction and its sampling error for the quasi-global annual precipitation from 1900 to 2011. J. Atmos. Sci., 71, 32503268, doi:10.1175/JAS-D-13-0301.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Signorini, S. R., C. R. McClain, and Y. Dandonneau, 1999: Mixing and phytoplankton bloom in the wake of the Marquesas Island. Geophys. Res. Lett., 26, 31213124, doi:10.1029/1999GL010470.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smerdon, J. E., A. Kaplan, D. Chang, and M. N. Evans, 2011: Corrigendum: A pseudoproxy evaluation of the CCA and RegEM methods for reconstructing climate fields of the last millennium. J. Climate, 24, 12841309, doi:10.1175/2010JCLI4110.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., and R. W. Reynolds, 2005: A global merged land–air–sea surface temperature reconstruction based on historical observations (1880–1997). J. Climate, 18, 20212036, doi:10.1175/JCLI3362.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., A. G. Barnston, M. Ji, and M. Chelliah, 1995: The impact of Pacific Ocean subsurface data on operational prediction of tropical Pacific SST at the NCEP. Wea. Forecasting, 10, 708714, doi:10.1175/1520-0434(1995)010<0708:TIOPOS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., R. W. Reynolds, R. E. Livezey, and D. C. Stokes, 1996: Reconstruction of historical sea surface temperatures using empirical orthogonal functions. J. Climate, 9, 14031420, doi:10.1175/1520-0442(1996)009<1403:ROHSST>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., R. E. Livezey, and S. S. P. Shen, 1998: An improved method for analyzing sparse and irregularly distributed SST data on a regular grid: The tropical Pacific Ocean. J. Climate, 11, 17171729, doi:10.1175/1520-0442(1998)011<1717:AIMFAS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., P. A. Arkin, and M. R. P. Sapiano, 2009: Reconstruction of near-global annual precipitation using correlations with sea surface temperature and sea level pressure. J. Geophys. Res., 114, D12107, doi:10.1029/2008JD011580.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., P. A. Arkin, L. Ren, and S. S. P. Shen, 2012: Improved reconstruction of global precipitation since 1900. J. Atmos. Oceanic Technol., 29, 15051517, doi:10.1175/JTECH-D-12-00001.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, T. M., S. S. P. Shen, L. Ren, and P. A. Arkin, 2013: Estimating monthly precipitation uncertainty beginning 1900. J. Atmos. Oceanic Technol., 30, 11071122, doi:10.1175/JTECH-D-12-00197.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sprintall, J., and M. Tomczak, 1992: Evidence of the barrier layer in the surface layer of the tropics. J. Geophys. Res., 97, 73057316, doi:10.1029/92JC00407.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • SSALTO/DUACS, 2014: SSALTO/DUACS User Handbook: MSLA and (M)ADT Near-Real Time and Delayed Time Products (reference CLS-DOS-NT-06-034). CNES, 29 pp. Accessed 15 November 2014. [Available online at http://www.aviso.altimetry.fr/fileadmin/documents/data/tools/hdbk_duacs.pdf.]

  • Trenberth, K. E., and D. P. Stepaniak, 2001: Indices of El Niño evolution. J. Climate, 14, 16971701, doi:10.1175/1520-0442(2001)014<1697:LIOENO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Turk, D., M. J. McPhaden, A. J. Busalacchi, and M. R. Lewis, 2001: Remotely sensed biological production in the equatorial Pacific. Science, 293, 471, doi:10.1126/science.1056449.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Turk, D., C. S. Meinen, D. Antoine, M. J. McPhaden, and M. R. Lewis, 2011: Implications of changing El Niño patterns for biological dynamics in the equatorial Pacific Ocean. Geophys. Res. Lett., 38, L23603, doi:10.1029/2011GL049674.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Uitz, J., H. Claustre, B. Gentili, and D. Stramski, 2010: Phytoplankton class-specific primary production in the world’s oceans: Seasonal and interannual variability from satellite observations. Global Biogeochem. Cycles, 24, GB3016, doi:10.1029/2009GB003680.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, X. J., R. Le Borgne, R. Murtugudde, A. J. Busalacchi, and M. Behrenfeld, 2008: Spatial and temporal variations in dissolved and particulate organic nitrogen in the equatorial Pacific: Biological and physical influences. Biogeosciences, 5, 17051721, doi:10.5194/bg-5-1705-2008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, X. J., M. Behrenfeld, R. Le Borgne, R. Murtugudde, and E. Boss, 2009: Regulation of phytoplankton carbon to chlorophyll ratio by light, nutrients and temperature in the equatorial Pacific Ocean: A basin-scale model. Biogeosciences, 6, 391404, doi:10.5194/bg-6-391-2009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wilson, C. and D. Adamec, 2001: Correlations between surface chlorophyll and sea surface height in the tropical Pacific during the 1997–1999 El Niño–Southern Oscillation event. J. Geophys. Res., 106, 3117531188, doi:10.1029/2000JC000724.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Woodruff, S. D., and et al. , 2011: ICOADS release 2.5: Extensions and enhancements to the surface marine meteorological archive. Int. J. Climatol., 31, 951967, doi:10.1002/joc.2103.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, J.-Y., H.-Y. Kao, T. Lee, and S. T. Kim, 2011: Subsurface ocean temperature indices for central-Pacific and eastern-Pacific types of El Niño and La Niña events. Theor. Appl. Climatol., 103, 337344, doi:10.1007/s00704-010-0307-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, J.-Y., Y. Zou, S. T. Kim, and T. Lee, 2012: The changing impact of El Niño on US winter temperatures. Geophys. Res. Lett., 39, L15702, doi:10.1029/2012GL052483.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal variability: 1900–93. J. Climate, 10, 10041020, doi:10.1175/1520-0442(1997)010<1004:ELIV>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save