Quantifying Predicting and Exploiting Uncertainty SIO Modeling Component

Quantifying, Predicting, and Exploiting Uncertainty SIO Modeling Component Bruce Cornuelle, Ganesh Gopalakrishnan, Ibrahim Hoteit, Julie Mc. Clean, Yoo Yin Kim SIO Lihue, Jan 12, 2009

Goals: Large-scale modeling • Understand the influences on the Kuroshio and the experimental region • Predictability of the environment – Kuroshio intrusions – Cold events • One word: “Sensitivity”: propagate uncertainties – IC, Bathymetry, forcing, resolution, …

2008 Tasks • Gather the pertinent observations for the region (w/Niiler) • Obtain HYCOM output, compare to observations • Set up grid with forcing and topography to match MIT (Lermusiaux) • Convert and apply HYCOM as boundary and initial conditions • Compare to MIT results and to observations from pilot experiment

Domain: to 140 E; hard to keep it small

Northern Boundary Condition: HYCOM T V (m/s) Assimilated HYCOM!

HYCOM: transport at N boundary (27 N) Temperature Transport (NOT heat!)

MITgcm runs: 1/12 degree • ECCO IC and BC, NCEP forcing – Bathymetry from ETOPO 2 – Dec 2001 -- Dec 2004 • ECCO IC, HYCOM BC, NCEP forcing – Dec 2003– Dec 2004 • HYCOM bathy, IC, BC, NCEP forcing – Jan 2004– Nov 2004

MITgcm runs: 1/24 degree • ECCO IC and BC, NCEP forcing – Bathymetry from ETOPO 2 – Dec 2001 -- Dec 2004 • ECCO IC, HYCOM BC, NCEP forcing – Dec 2003– Dec 2004 • Just to see sensitivity to resolution, BC

MITgcm; 1/12, ECCO IC, HYCOM OBC

MITgcm; 1/12, ECCO IC, HYCOM OBC

MITgcm; 1/12, ECCO IC, HYCOM OBC

MITgcm; 1/12, HYCOM IC, HYCOM OBC

MITgcm; 1/12, ECCO IC, HYCOM OBC

MITgcm; 1/12, HYCOM IC, HYCOM OBC

MITgcm; 1/24, ECCO IC, HYCOM OBC July-September 15 m Velocity Mean (m) RMS (m)

MITgcm; 1/24, ECCO IC, HYCOM OBC July-September 50 m Temp. Mean (m) RMS (m)

To Do • More reality checks and sensitivity tests – Dynamical diagnostics • 2008 runs with NCOR topography, COAMPS or NOGAPS forcing, HYCOM IC/OBC • Adjoint run for experimental region (and cold dome region) • Predictability studies

Thank you!

2008 -2009 Tasks • Extend the model runs and compare to observations and MIT modeling results. • Evaluate the effects of the boundary conditions including adjoint calculations • Start to fit the regional model to the observations to have a better test bed • Do adjoint sensitivity calculations for the experimental region and environmental keys. • Coordinate with Heaney on optimal sampling calculations.

2009 -2011 Tasks • Make hindcast estimates for the experimental region before, during, and after the experiment. • Include ensemble calculations for uncertainty estimates, including predicted probability of detection (PPD).

Drifter trajectories calculated by Yoo Yin Kim 1/12 th degree run, then 1/24 th degree run

1/12 degree

1/12 degree

MITgcm; 1/12, HYCOM Bathy, IC, BC

MITgcm; 1/12, HYCOM Bathy, IC, BC

HYCOM; 1/12

HYCOM; 1/12

MITgcm; 1/24, ECCO IC, HYCOM OBC

MITgcm; 1/24, ECCO IC, HYCOM OBC

MITgcm; 1/12, ECCO IC, HYCOM OBC July-September 15 m Velocity Mean (m) RMS (m)

MITgcm; 1/12, ECCO IC, HYCOM OBC July-September 50 m Temp Mean (m) RMS (m)

MITgcm; 1/12, ECCO IC, HYCOM OBC July-September SSH Mean (m) RMS (m)

MITgcm; 1/12, ECCO IC, HYCOM OBC July-September 15 m Velocity Mean (m) RMS (m)

MITgcm; 1/12, ECCO IC, HYCOM OBC July-September 50 m Temp. Mean (m) RMS (m)

MITgcm; 1/12, ECCO IC, HYCOM OBC July-September 50 m Salinity Mean (m) RMS (m)

MITgcm; 1/12, ECCO IC, HYCOM OBC July-September 50 m Salinity Mean (m) RMS (m)

MITgcm; 1/12, ECCO IC, HYCOM OBC July-September SSH Mean (m) RMS (m)

MITgcm; 1/24, ECCO IC, HYCOM OBC July-September 50 m Salinity Mean (m) RMS (m)

MITgcm; 1/24, ECCO IC, HYCOM OBC July-September SSH Mean (m) RMS (m)

1/24 degree

1/24 degree