A Forecasting system for the Southern California Current Emanuele Di Lorenzo Arthur Miller Bruce Cornuelle Scripps Institution of Oceanography, UCSD
Forecast the mesoscale eddies Understand the physics that control their generation and evolution Assess the biological response
Observational Dataset Southern California Coast and Baja Temperature, Salinity and Zooplankton 1949 – 2003 seasonal data 20 m vertical resolution, from 0– 500 m km horizontal grid CalCOFI historical sampling grid California Cooperative Oceanic Fisheries Investigation Hydrography
The Strategy Initialize the model by assimilating the slowly evolving component of the eddy field which can potentially lead to forecast skill over a period of 2 months
The Strategy The Method The Green’s Function Method Initialize the model by assimilating the slowly evolving component of the eddy field which can potentially lead to forecast skill over a period of 2 months
A typical sampling of a mesoscale eddy E1 “SSH”
AVVISO TOPEX/ERS [m] E1 E2 SSH E1 E2 E1 “SSH” SSH CalCOFI Coastal observation Data Assimilation Ocean Circulation Model Satellite Data Independent verification Assimilation of CalCOFI T,S 65% reduction in error variance relative to the model initial guess!
SeaWIFS [ M N/m 3 ] E2 E1 E3 Chl-a [ M N/m 3 ] E1 E2 E3 Chl-a E3 Chl-a Ecosystem Model Forecasting and Hindcasting Ocean Productivity Independent verification CalCOFI in Situ REFERENCE: Di Lorenzo, E., A. J. Miller, D. J. Neilson, B. D. Cornuelle, and J. R. Moisan, 2003: Modeling observed California Current mesoscale eddies and the ecosystem response. International Journal of Remote Sensing, in press.
You need to know the physics that goes into the assimilation scheme
E1 “SSH”
1010 JAN APR JUL OCT SKILL associated with Persistence of Initial Condition SSHSurface TT 150 m days
E1 “SSH” Data is collected over a 20 day period
JAN APR JUL OCT SSHSurface TT 150 m days SKILL evolution when the true initial condition is replaced with a 20 day average in a dynamical forecast
JAN APR JUL OCT SSHSurface TT 150 m days
How about the uncertainties in Forcing Functions?
1010 JAN APR JUL OCT SKILL evolution with errors in Forcing Surface TT 150 m days
JAN APR JUL OCT SKILL evolution with errors in Forcing (and Open BC) Surface TT 150 m days SKILL evolution with errors in initial condition JUNE
Wavenumber Spectra
Forecast the mesoscale eddies Real time forecast of CalCOFI in April 2003 SCCOOS nowcast-forecast with UCLA and JPL Understand the physics that control their generation and evolution Error Covariances Seasonal dependence Assess the biological response In progress…. Concluding remarks:
ΔρΔρ Ekman Pumping Δρ increase Δh westward propagation off the Bight Wind P. Conception Shelf (a) Instability processes on continental slope currents (b) (c) April July
Ocean Temperature Zooplankton Log e Tot. Vol Anomalies C 1 0 Observations along the California Coast
Assimilation Method
E1 “SSH” CalCOFI Coastal observation Assimilation of CalCOFI T,S
E1 E2 E1 “SSH” SSH CalCOFI Coastal observation Data Assimilation Ocean Circulation Model Assimilation of CalCOFI T,S
What have we learned about the mesoscale dynamics?
What have we learned about the mesoscale dynamics?
The Green’s Function Method Green’s Function
JAN APR JUL OCT Changes in SKILL associated with errors in Forcing Surface TT 150 m days