Presentation is loading. Please wait.

Presentation is loading. Please wait.

Going Against the Flow: Ocean Circulation Data for F^3 Carter Ohlmann and Satoshi Mitarai Computing help from: Kirk Ireson, Jeff Lee, and Brian Emery.

Similar presentations


Presentation on theme: "Going Against the Flow: Ocean Circulation Data for F^3 Carter Ohlmann and Satoshi Mitarai Computing help from: Kirk Ireson, Jeff Lee, and Brian Emery."— Presentation transcript:

1 Going Against the Flow: Ocean Circulation Data for F^3 Carter Ohlmann and Satoshi Mitarai Computing help from: Kirk Ireson, Jeff Lee, and Brian Emery

2 Lagrangian Ocean Circulation Drifting Buoys: - measure paths of ocean circulation - measure eddy dispersion quantities - resolve advective terms in N-S equations fate and transport of Montecito effluent Southern California Bight connectivity small scale (10 m) coastal mixing understanding the California Current System Carter Ohlmann; ICESS/UCSB

3

4 Outline  models - necessary, each typically gives a unique solution  drifter observations - relevance to connectivity studies  existing drifter data in SCB - range of time/space scales  using data with models to enhance understanding of connectivity  work with Satoshi …. very much in the preliminary stages Goals  demonstrate need to validate model results with data  show existing drifter data from Southern California Bight  motivate use of these data with models in SCB MPA studies

5 Mean SSH for California Current from four models. Offsets were applied to make plots with the same color scale. Solutions are averages from 1990 to Figure from Centurioni, Ohlmann, and Niiler (JPO, 2008). Which model of the mean SSH is most representative? Why?

6 Square root of GMEKE from same four models. Figure from Centurioni, Ohlmann, and Niiler (JPO, 2008). Which model of the EKE is most representative? Why?

7 Very important point demonstrated by previous slides:  data are necessary for interpreting model results My conjecture:  thorough model assessment is non-trivial, requiring a quantitative comparison of a number of parameters specific to model configuration, its planned use, the available data, and the circulation

8 ocean current observations are available from many instruments Eulerian observations give point measurements or time/space means  satellite altimetry  ship/glider/AUV surveys  moorings (ADCP)  high frequency radar Lagrangian observations follow the path of a water parcel  drifters (surface currents)  floats (subsurface currents) Eulerian means show velocities near zero corresponding Lagrangian trajectories show significant displacements Figures from Ohlmann and Niiler, 2005

9 Lagrangian drifter data are particularly relevant to connectivity 1)connectivity is a Lagrangian process 2)measure flow evolution in both time and space 3)provide independent means of model validation (not easily assimilated) 4)directly measure Lagrangian Stochastic Model parameters (σ 2, T l, D x,y ) 5)can resolve a large range of scales (minutes -> years; meters -> 100’s of km) 6)~13,500 drifter days of data exist in the SCB region; more coming

10 SCB drifter data on scales not resolved in regional models More than 300 velocity observations within a 2 x 2 km grid cell during 5 days to measure sub-grid-scale energy. Data in Ohlmann et al. (2007; JAOT). 2 km Is sub-grid-scale energy (not resolved in SCB models) significant? data show EKE of ~9 cm 2 s- 2 within 2 km box

11 SCB drifter data on scales connecting shelf to intertidal Drifter triplets deployed weekly at a specific location. Drifters sample for ~6 hours as part of state funded interdisciplinary project. Data collection planned for 52 weeks. 7 km data show distribution of absolute dispersion (i.e. from a point) how much data are necessary for a useful comparison with models?

12 drifter data on the sub-meso-scale Drifter sets deployed for 3-4 days just north of San Francisco as part of a validation experiment for oil spill response. 50 km is retention time in SCB models realistic? data from Garfield, Largier, Ohlmann, and Paduan

13 SCB drifter data on the regional scale Drifters deployed ~quarterly from 1993 – drifters sampling for an average of ~24 days give ~13,500 drifter days of data. 250 km Dever et al., 1998

14 data show sensitivity to start location  descriptive patterns “look similar” to ROMS results shown by Satoshi  must go beyond “look similar” with quantitative assessments 30-day drifter tracks deployed at 3 start locations ( ) separated by ~40 km few tracks in SBCtracks exit SBC to east tracks confined to SBC

15 data show seasonal variations RED: winter data (Dec, Jan, Feb) BLUE: spring data (Mar, Apr, May) northward (relaxation) motion in winter southward (wind- driven) flow in spring 30-day tracks in different seasons

16 data show inter-annual (ENSO) variations RED: data from Jan ’97 -> June ’98 BLUE: data from July ’98 -> Dec ’99 red ENSO tracks go farther north, and move south-east along coast 30-day tracks in different years

17 Correct connectivity solutions to a non-trivial problem require:  collaborative (models-data) paradigm with thorough assessment  specific metrics for assessing model skill quantitatively  “applications-assessment” (?) funding (not “research” or “observations”)  need patience and persistence; “the devil is in the details” Summary  connectivity requires models; models require data to assess skill  drifter data are key (Lagrangian, time-space, scales, not assimilated)  data exist; more forthcoming; how much are necessary?  SCB data suggest sub-grid-scale, seasonal, and inter-annual signals

18 Work with Satoshi since MLPA presentation 1)examine SCB Lagrangian circulation with drifter observations and ROMS trajectories 2)examine SBC Lagrangian circulation with trajectories computed from HF radar and ROMS trajectories

19 observationscorresponding ROMS trajectories individual track comparisons are qualitatively different 45 day trajectories, n = 69, quarterly releases during 1996

20 observationscorresponding ROMS trajectories Lagrangian autocorrelation curves are qualitatively similar 14 day trajectories, n = 44/41, quarterly releases during 1996

21 Hypothesis: EKE on “sub-grid-scales” does not significantly alter “larger-scale” flow statistics, but prohibits individual track agreement. EKE as a function of spatial resolution for a 1 x 1 km grid Ohlmann et al. 2007

22 cartoon showing that two particles initially separated by some delta-x will further separate in time observations of time separation in SBC give power law growth ~ t 2.4. Ohlmann et al. in prep. background information on “two particle statistics”

23 mean square dispersion between observed and ROMS trajectories grows as t 1.7 vs. t 3 from theory this shows ROMS has utility compared with simply considering a turbulent ocean; but, if you fall overboard in the SCB………. 3 km 300 km

24

25 trajectories from HF radarcorresponding ROMS trajectories daily releases during 2002, n=360, track length while in domain HF radar observations allow for computation of large numbers of trajectories from data

26 trajectories from HF radarcorresponding ROMS trajectories qualitatively good agreement when considering an entire year statistical distributions of where trajectories exit the HF radar spatial domain

27 trajectories from HF radarcorresponding ROMS trajectories qualitatively good agreement when considering an entire year; statistical distributions of how long trajectories stay within the HF radar spatial domain

28 Where we’re headed: Converge statistical comparisons shown for ROMS and HF radar, with specific trajectory comparisons shown for ROMS and data. What we want to accomplish: Quantify error bounds on trajectories as a function of time and space scales, and ????. Why are we doing this? To greatly enhance meaning of Satoshi’s modeled connectivity matrices.

29 Data (overlaid vectors) are necessary to determine model skill. Mean SSH from ROMS with the observed unbiased geostrophic velocity field superimposed. Figure from Centurioni, Ohlmann, and Niiler (JPO, 2008).


Download ppt "Going Against the Flow: Ocean Circulation Data for F^3 Carter Ohlmann and Satoshi Mitarai Computing help from: Kirk Ireson, Jeff Lee, and Brian Emery."

Similar presentations


Ads by Google