1. Regional and Coastal Circulation Modeling: California Current System Art Miller Scripps Institution of Oceanography ECOFOR Workshop Friday Harbor, WA September 7-10, 2012 ECOFOR Workshop Friday Harbor, WA September 7-10, 2012

2. Regional and Coastal Circulation Modeling: California Current System The physical-biological observational datasets motivate many modeling studies with a Unifying Scientific Motivation: How do changes in surface forcing (heat fluxes, wind stresses) alter stratification, upwelling cells and mesoscale eddy statistics and the consequent upward nutrient fluxes and subsequent biological response?

3. Regional and Coastal Circulation Modeling: California Current System Brief Review of Two Classes of Modeling: 1) Long-term climate hindcasts - Deterministic: Explain observed changes in forced physical structures - Stochastic: Identify relations among variables and input forcing 2) Data assimilation runs - Enhance observations in space and time for process diagnostics - Initialize predictions of eddies and forced components Brief Review of Two Classes of Modeling: 1) Long-term climate hindcasts - Deterministic: Explain observed changes in forced physical structures - Stochastic: Identify relations among variables and input forcing 2) Data assimilation runs - Enhance observations in space and time for process diagnostics - Initialize predictions of eddies and forced components

4. Regional and Coastal Circulation Modeling: California Current System Brief Review of Two Classes of Modeling: 1) Long-term climate hindcasts - Deterministic: Explain observed changes in forced physical structures - Stochastic: Identify relations among variables and input forcing Brief Review of Two Classes of Modeling: 1) Long-term climate hindcasts - Deterministic: Explain observed changes in forced physical structures - Stochastic: Identify relations among variables and input forcing

5. Examples of deterministic CCS hindcasting - Curchitser et al. 2005 used a basin-scale hindcast to show that The 2002 cold/fresh anomaly in the northeast Pacific was due to enhanced mixing during the preceding winter in the center of the Alaska gyre - Hermann et al. 2009 used a regional and basin-scale model to isolate remotely from locally driven sea level and current ENSO variability in the CCS and ACS regions - Rykaczewski and Dunne 2010 used a global greenhouse-gas forced run to predict that nitrate and primary production will increase in the CCS due to differences in subduction and age of upwelled waters

6. Coastal upwelling regions controlled by PDO and NPGO Di Lorenzo et al., GRL, 2008

7. Adjoint runs of passive tracer in upwelling zon: Surface warming causes shallower coastal upwelling cell Less nutrient flux to surface Positive PDO Phase Model Adjoint backward runs of passive tracer in upwelling zone: Reveal how weaker upwelling winds cause shallower coastal upwelling cell (Chhak and Di Lorenzo, 2007) Negative PDO Phase Surface layer transport into coastal upwelling zone Mid-depth (150m) transport into coastal upwelling zone More nutrient flux to surface

10. Regional and Coastal Circulation Modeling: California Current System Brief Review of Two Classes of Modeling: 2) Data assimilation runs - Enhance observations in space and time for process diagnostics - Initialize predictions of eddies and predictable forced components Brief Review of Two Classes of Modeling: 2) Data assimilation runs - Enhance observations in space and time for process diagnostics - Initialize predictions of eddies and predictable forced components

11. Near-Real-Time CCS Data Assimilation by UC, Santa Cruz May 2, 2012 7-day fits using mostly surface data with ROMS @ 10km Broquet et al. (2009)

12. SCCOOS 3DVar ROMS model (JPL-UCLA) Surface CODAR is a key variable Daily updates with 1km resolution every 6 hrs 72-hour forecasts executed daily Yi Chao et al.

13. Data Assimilation Fits for April 2002 and 2003 - Strong constraints over 30-day periods allows diagnosis of 4D physical processes that help explain the large disparity in sardine spawning Nearshore spawning, many eggs: El Nino Song et al., 2012, JGR Offshore spawning, fewer eggs: La Nina Data includes: T-S (CalCOFI, Argo, CUFES), SLH (AVISO), SST (AVHRR)

14. Data Assimilation Model Fits: (1) Quantifying Transport Stronger offshore transport and upwelling in 2002 Weaker offshore transport and stronger convergence in 2003 Song et al., 2012, JGR Red: Egg density Grey Scale Arrows: Surface Currents

15. Data Assimilation Model Fits: (2) Quantifying Upwelling Sources Adjoint tracer model (run backwards) for source waters (boxes) of surface ocean 2002 source waters in offshore spawning area transported from more productive upwelled surface water near the coast Song et al., 2012, JGR Orange indicates location of water 30 days before arriving in BOX

16. Data Assimilation Model Fits: (2) Quantifying Upwelling Sources Adjoint tracer model (run backwards) for source waters (boxes) of surface ocean 2003 source waters in nearshore spawning area transported from more productive deep water in the central California Current Song et al., 2012 Orange indicates location of water 30 days before arriving in BOX

17. Regional and Coastal Circulation Modeling: California Current System Brief Review of Two Classes of Modeling: 1) Long-term climate hindcasts - Deterministic: Explain observed changes in forced physical structures - Stochastic: Identify relations among variables and input forcing 2) Data assimilation runs - Enhance observations in space and time for process diagnostics - Initialize predictions of eddies and forced components Brief Review of Two Classes of Modeling: 1) Long-term climate hindcasts - Deterministic: Explain observed changes in forced physical structures - Stochastic: Identify relations among variables and input forcing 2) Data assimilation runs - Enhance observations in space and time for process diagnostics - Initialize predictions of eddies and forced components

18. Thanks ! ECOFOR Workshop