Coastal Ocean Modeling at CIOSS Alexander Kurapov, in collaboration with R. Samelson, G. Egbert, J. S. Allen, R. Miller, S. Erofeeva, A. Koch, S. Springer, J. Osborne - Pilot real-time forecast model of coastal ocean circulation off Oregon - Modeling flows on the shelf and in the coastal transition zone off Oregon for the period of summer 2002 GLOBEC field program (CIOSS/GLOBEC) Use satellite data for model evaluation & data assimilation Use data assimilation for dynamically-consistent interpolation of data sets Use models to interpret observed variability
Coastal Ocean Dynamics off Oregon: Wind-driven alongshore currents Upwelling Vertical shear in horizontal currents Model: lines - v, color - σ θ Model SSH (1-10 Aug 2002) and Topex tracks GOES SST, Aug 2002 Surface observations present information about 3D processes originated over shelf + Undercurrent + Frontal instabilities + Internal tides
Motivation: Satellite observations are sparse compared to spatial and temporal scales of coastal processes Models can provide information on the sources of origin and evolution of dynamical structures observed from space Data assimilation (DA) can be used as a dynamically-based interpolator of sparse data sets - extend utility of the satellite products to the shelf - establish connection between surface observed fields and subsurface, unobserved, fields Difficulties: Nonlinear, Ageostrophic Dynamics
Good work plan would involve: - Model development - Model verification against observations - Understanding processes in the coastal ocean - Development of a DA capability (appropriate to research or operation goals and adequate with respect to the processes) - Transition to the real-time model - QC of the real-time product Human resources is an issue!!
Real-time forecast model of coastal flows off Oregon: ftp://ftp.coas.oregonstate.edu/pub/lana/nctz/sscforecast.html (model support and development: S. Erofeeva / CIOSS) Daily updates of three day forecasts, w/ information posted online (daily maps on pink background: for today and next 2 days) System launched in 06/2005 Major modifications in April 2007, May 2008 (better boundary conditions)
Present model configuration: ROMS, 3-km resolution, 30 vertical layers (terrainfollowing coordinates) Domain: see on the right (shown are daily averaged surface currents and temperature) Atmospheric forcing: mesoscale NOAA-NAM Boundary conditions (BC): 9-km NCOM-CCS (Shulman et al., NRL) When NCOM forecast not available: backup model w/ climatologic BC (monthly mean fields from 4 years of NCOM outputs) The model reproduces dynamics both over the shelf and in the CTZ, esp. energetic current separation off Cape Blanco Future developments: DA, river influences, tides
Comparison with MODIS SST weekly composites (Venegas): The model SST variability is qualitatively correct Need further model modification & DA to improve timing and location of separation zones
Model verification using OrCOOS time-series observations: Mid-shelf mooring (NH10, at 44.65N) and model, south-north velocity, May-June 2008 Mean surface currents (May-June 2008) HF radar (Kosro) Model 12 m 25 m 42 m 56 m 70 m
Additional model verification: Summer 2002 GLOBEC study (Koch) Monthly aver. SST: 200 m ROMS: GOES:
Comparisons with alongtrack SSH (Aviso): The level of horizontal dissipation in the model influences coastal current separation Track 206 July 30, 2002
Statistical comparison of 9-km NCOM and 3-km ROMS (nested in NCOM) using alongtrack SSH altimetry: alongtrack anomalies correlations Corr(NCOM, AVISO)=0.53 Corr(ROMS, AVISO)=0.75
Depth-averaged alongshore currents: model vs. mid-shelf moorings (mooring data: Kosro, Hickey, Ramp)
Surface drifter trajectories (Summer 2002 cruise, J. Barth, vs. ROMS, A. Koch) Star release point markers drifter locations on days 1, 2,, 5 Bathym. contours: 100, 200, and 1000 m (bold) Can satellite DA help improve predicted trajectories?
Nontrivial vertical structure of the westward jet separated from Cape Blanco: wind ROMS SSH 1-10 Aug 2002 Potential density u TKE, (1 Aug 2002) Asymmetry in the jet structure is due to nonlinear effect on surface Ekman transport: τ y /ρ(f+ζ) ζ=v x -u y South North 1/f
Passive tracer (labels) analysis helps to find zones of upwelling and subduction within the jet: Initialize the field of z- labels on July 30, 2002 Advect it as a passive tracer Aug 1 Aug 3 (deepening of the jet) Labeling technics (xyz): see Kuebel - Cervantes, Allen, & Samelson, 2004
Nested model simulations at higher, 1-km, resolution (Osborne, w/ NSF support) Resolving submesoscale eddy processes affects the frontal structure Ave SST, Aug, 2002: 3-km resolution ROMS (A. Koch) 1-km resolution ROMS (J. Osborne)
Development of the data assimilation component (w/ Egbert, Allen, Miller, supported by ONR): Variational approach = Optimal interpolation, in which time-space interpolation (covariance) functions depend on the model dynamics (background ocean conditions) (the DA estimate fits the model dynamics and available data over a given time interval in the least-squares sense) Representer cycling algorithm (Xu & Daley, 2000): assimilation forecast
Kurapov et al., 2008, Dyn. Atm. Oceans, accepted: NL model: ROMS TL & ADJ codes: OSU AVRORA - tests in an idealized upwelling setting [twin experiments, alongshore uniform conditions]: Sections of alongshore velocity (line contours, every 0.1 m/s) and density (color), day 10: Assimilation of surf. currents improves prediction of subsurf. fields Prior (w/ spatially uniform wind stress) Bold contour: 0.5 m/s True (w/ wind stress, reduced inshore of upwelling front) Nonlinear model w/ stress improved by assimilation of surface currents
SUMMARY: Satellite data provide information about processes that originate near coast Radial returns from P. M. Kosro s array of long range HF radars (June 2004) Model-data comparisons using satellite data have guided model improvement 50N Assimilation of these observations, esp. in combination with in-situ measurements, should allow us to improve model estimates (in 3D) 48N 46N Comprehensive high-resolution models provide information about nonlinear and ageostrophic dynamics near surface 44N 42N ERS-1/2 TOPEX TOPEX-2 GFO We live in a time of new discoveries in the coastal ocean 40N 128W 127W 126W 125W 124W 123W 122W Sat. tracks 2002-05 (courtesy of P.T. Strub)