Transcript SCUD

SCUD
Diagnostic Surface Currents SCUD
and application to marine debris
Jan Hafner and Nikolai Maximenko
[email protected], [email protected]
IPRC/SOEST University of Hawaii
Hydrodynamics of Marine Debris
workshop
5IMDC – 20 March 2011 Honolulu, Hawaii
Outline

Motivation

Methodology

Data

Model formulation

Application to Marine Debris

Future
Motivation

Ocean Surface Currents – important factor in marine debris
problem

Direct measurements difficult – few in situ observations

Utilize satellite data to arrive with surface ocean currents

supported by the following agencies:

NASA Physical Oceanography Program (Ocean Surface Topography Science Team)

US National Fish and Wildlife Foundation

JAMSTEC

NOAA sponsoring IPRC

Our direct motivation is from applications on marine debris
Methodology



Task: to develop a simple diagnostic model of
surface ocean currents to fit drifters' trajectories
Input parameters: AVISO sea level anomaly
(geostrophic current component)
Ocean surface wind data: daily QSCAT – wind
driven current component ( Ekman)
DATA



Drifter data: AOML - Atlantic Oceanographic and
Meteorological Laboratory www.aoml.noaa.gov
8058 drifters, drogued at 15m
from 1979 till 2008, interpolated on 6 hourly
intervals
DATA



AVISO mean sea level anomaly maps: 1/3
degree maps, merged product (up 4 satellites),
weekly time frequency, starting Oct. 1992
(www.aviso.oceanobs.com)
MDOT - Mean Dynamic Ocean Topography,
developed by Maximenko et al. (2009), ½ degree
map produced using combined drifters, sea
altimetry, GRACE and surface wind data, 19922002.
QSCAT 3-day moving averages of surface winds
(10 m), ¼ degree daily maps July 1999 –
November 2009, (www.ssmi.com)
Formulation of the diagnostic model
USCUD(x,y,t) =
U0 + uhx⋅∇xh(x,y,t) + uhy⋅ ∇yh(x,y,t) + uwx⋅wx(x,y,t) + uwy⋅wy(x,y,t)
And similarly
VSCUD(x,y,t) =
V0 + vhx⋅∇xh(x,y,t) + vhy⋅ ∇yh(x,y,t) + vwx⋅wx(x,y,t) + vwy⋅wy(x,y,t)
Where: USCUD , VSCUD - modeled ocean current components
U0 , V0 - constant coefficient (mean)
h
- sea level anomaly
wx, wy U and V component of surface wind (QSCAT)
uhx, uhy , uwx , uwy - U component coefficients corresponding to
sea level gradient and surface wind (function of x and y only)
vhx, vhy , vwx , vwy - similarly corresponding V component
coefficients
Formulation of the diagnostic model
The coefficients are solved by minimizing the cost function:
Fcost=Σ[(Udrifter ‐ USCUD)2 + (Vdrifter ‐ VSCUD)2]
where the summation is over all drifters' data in a given lat/lon box
(total 5,700,000 6-hourly data points).
RESULTS
RESULTS
Local Scale
SCUD application on marine debris
transport and convergence
Where the marine debris goes?
How it gets there ?
Numerical Experiment: SCUD currents
applied on ocean tracers released daily
from coast and weighted by coastal population
count
Animation of tracer transport by
SCUD currents
Structure of SCUD tracers “patches”
SCUD model application on marine debris
What model can do:
zones of convergence
structure of the patches
trajectories = pathways
What model cannot do:
prediction
vertical structure of marine debris
coastal processes – emission and
deposition of marine debris
Future – what is needed
1. operational SCUD product requires QSCAT to be replaced
with ASCAT winds
2. global inventory of marine debris sources and sinks
in the ocean and onshore is needed
3. effect of vertical mixing on floating debris needs to be
included in the model
4. coastal dynamical processes, esp. high frequency and debris
deposition processes, need to be considered in the model
5. validation of SCUD model results by in situ data needed
Thank you
Data preprocessing


AVISO, MDOT and QSCAT wind data were
interpolated on times and locations of 6-hourly
drifters' data
Filtering out high frequency signal by Hanning
cosine filter with halfwidth = inertial frequency,
minimum frq. ~ 3 days (9°37' lat.)
Fit to the data
Absolute misfit to drifters' data
R.M.S. of cost function (m/s)
Global average misfit : 0.162 m/s
(0.118 m/s for USCUD
and 0.107 m/s for VSCUD)
Relative misfit to the drifters' data
Ratio of cost function and drifters'
R.M.Ss.
Global average : 0.566
(0.541 and 0.653, for U and V
components respectively)
Data and Access



¼ degree surface currents maps: daily from
01Aug1999 till 19Nov2009 (span of QSCAT data)
SCUD dataset is open for free unrestricted use and
distribution
Disseminated by APDRC servers :
http://apdrc.soest.hawaii.edu/projects/SCUD/

LAS, LAS7, OpeNDAP, DChart

SCUD manual :
http://apdrc.soest.hawaii.edu/projects/SCUD/SCUD_manual_02_17.p
df

SCUD users listserver:
http://apdrc.soest.hawaii.edu/projects/SCUD/registration.html
From S.Pacific ST gyre
From N.Pacific ST gyre
To S.Pacific ST gyre
To N.Pacific ST gyre
From Hawaii
To Hawaii
Trajectories of real drifter starting from (left column) and
ending in (right column) the South Pacific (top row), North
Pacific (middle row), and Hawaii (bottom row).
Statistics
R.M.S. of modeled velocities related to
sea level (m/s)
R.M.S. of modeled velocities
related to surface winds (m/s)