Ocean Biological Modeling and Assimilation of Ocean Color Data Watson Gregg NASA/GSFC/Global Modeling and Assimilation Office Assimilation Objectives: Improved State and Flux Estimation (Chlorophyll and Primary Production) Modeling Objectives: New Derived Variables Linkages Between Ocean and Atmosphere Improved Climate Predictions
Radiative Model (OASIM) Circulation Model (Poseidon) Biogeochemical Processes Model Winds SST Layer DepthsIOP E d (λ) E s (λ) Sea Ice NASA Ocean Biogeochemical Model (NOBM) Winds, ozone, relative humidity, pressure, precip. water, clouds (cover, τ c ), aerosols ( τ a, ω a, asym) Dust (Fe) Advection-diffusion Temperature, Layer Depths E d (λ) E s (λ) Chlorophyll, Phytoplankton Groups Primary Production Nutrients DOC, DIC, pCO 2 Spectral Irradiance/Radiance Outputs: Global model grid: domain: 84 S to 72 N 1.25 lon., 2/3 lat. 14 layers
Diatoms Biogeochemical Processes Model Ecosystem Component Chloro- phytes Cyano- bacteria Cocco- lithophores Si NO 3 NH 4 Herbivores N/C Detritus Fe Silica Detritus Phytoplankton Nutrients Iron Detritus
N/C Detritus Herbivores Phyto- plankton Dissolved Organic Carbon Dissolved Inorganic Carbon pCO2 (water) pCO2 (air) Winds, Surface pressure Biogeochemical Processes Model Carbon Component
Blue = NOBM; Green = Data
Ocean Color Assimilation: The SEIK filter (Lars Nerger, GMAO) Generally an ensemble Kalman filter Simplification Keep state covariance matrix constant (store ensemble perturbations, integrate ensemble mean state) essentially an ensemble OI scheme Application to Ocean Color Daily assimilation of gridded data into surface layer Chlorophyll distribution log-normal assimilate logarithmic quantities Satellite errors can affect results explicitly define regional satellite errors estimated from global analysis of in situ data
Comparison with In-Situ Data Spatially and temporally coincident data (daily) Strong improvement compared to free-run model Several regions: Assimilation with smaller error than SeaWiFS
Assimilation: Conclusions Major improvement of state estimates (chlorophyll) occasionally superior to SeaWiFS estimates Substantial improvement of flux estimates (primary production) but model still controlling Predictive capability on order of days New work on assimilation methods needed and ongoing
Blue = NOBM; Green = Data New Derived Variables: Phytoplankton Groups
Balch et al. (2005) Calcite Sep-Dec Balch et al. (2005) Calcite Apr-Jun Balch et al. (2005) Calcite Jul-Aug
Likely locations of coccolithophore blooms from Iglesias-Rodriguez et al., 2002)
Kamykowski et al. (2002) Diatoms Annual
Alvain et al. (2005) Jan Alvain et al. (2005) Apr Alvain et al. (2005) Jun Red = diatoms Green/yellow = cyanobacteriaBlue = cocco (NOBM) haptophytes (Alvain) NOBM
New Derived Variables: Functional Groups Conclusions Model compares favorably with in situ data, but there are major discrepancies Comparison with satellite observations is sometimes encouraging: coccolithophores in North Atlantic with Balch and Brown diatoms with Kamykowski And sometimes disappointing: coccolithophores in North Pacific with Balch general patterns with Alvain Emerging field and convergence is fleeting, even with definition of groups (except diatoms)