940 likes | 1.31k Views
What Have We Learned after 5 Years of LEO? Oscar Schofield & Scott Glenn Coastal Ocean Observation Lab (COOL) Rutgers University. Science web site http://marine.rutgers.edu/cool. Operational web site http://www.thecoolroom.org.
E N D
What Have We Learned after 5 Years of LEO?Oscar Schofield & Scott Glenn Coastal Ocean Observation Lab(COOL)Rutgers University Science web site http://marine.rutgers.edu/cool Operational web site http://www.thecoolroom.org
Our Partners: Robert Arnone, Trisha Bergmann, Don Barrick, Buzz Bernstein, Paul Bissett, Emmanuel Boss, Mike Crowley, Gary Fahnenstiel, Clayton Jones, Gary & Barbara Kirkpatrick, John Kerfoot, Dave Kohler, Steve Lohrenz, Alex Kahl, Josh Kohut, Mark Moline, Dave Millie, Chaya Mugdal, Matthew Oliver, Hugh Roharty, Emily Romanna, Doug Webb, Alan Weidemann, AND John Wilkins Paul Bissett Mark Moline Dale Haidvogel Scott Glenn Fred Grassle Bob Chant Oscar Schofield The Jersey Observatory Family
What is one of oceanography’s Holy Grails? We need to collect data over decadal scales and over large spatial (10-1000 km) scales so we can define the mean behavior variability and the trends occurring due to natural and anthropogenic changes. Superimposed on these changes are episodic events, which impact the physics, chemistry, and biology. Quantifying the relative importance of these fluctuating properties requires an adaptive sampling capacity so we can collect statistics on the importance of the episodic events. This requires data to be collected for over a decade to assure that several episodic events are encountered. This will be the key To define if there is any trended change in response to human Activity. Until then we will reconciled to debate on AM radio with Politcal extremists from both sides of the spectrum. So we need a new to way to observe the ocean. Ocean observatories have (hopefully??) evolved to fill this function.
Gary Kirkpatrick David Millie Oscar Schofield WHY OCEAN OBSERVATORIES?
NJSOS Our Observatory Experience SATELLITES CODAR LEO-15 Sustained LEO-CPSE Integrated
LEO-15: A sustained observatory 3km x 3km 1996-Present
LEO-15: A sustained observatory Nor’Easter Upwellng 25 0 Temperature Depth (m) 15 15 175 210 Julian Day >6 0 Chlorophyll a (mg L-1) Depth (m) <2 15 175 210 Julian Day Succeses High resolution time series during summer upwelling High resolution data sets ideal for validating 1-D models of sediment transport Variability in coastal optical properties Goal is to move into preoperational status, where it does not require a science team for operation, operations being transferred to staff
30 X 30 km LEO CPSE An Integrated Observatory
New Jersey Coastal Upwelling July 6, ’98 - AVHRR July 11, ‘98 - SeaWiFS Chlor-a (mg/m3) Temperature oC 19 20 21 22 24 .1 .3 .5 1 2 4 40N 40N Historical Hypoxia/Anoxia Field Station Field Station LEO LEO 39N 39N 75W 74W 75W 74W Barnegat Cape May
SST Seasonal temperature variation is the primary signal. Summer upwelling is 2nd
Causes of Hypoxia/Anoxia Surface bloom wind SW upwelling Stratification favorable wind Decay Bacteria Depletes Oxygen
12 m 15 m 20 m 25 m 30 m 35 m 40 m 50 m 100 m 500 m 1000m 2500m Hypoxia/Anoxia & Bottom Bathymetry Warsh – NOAA 1989
Modeled Effect of Bathymetric Variability on Upwelling 1 m/s current velocity Along shore subsurface deltas cause upwelling to be 3d, not 2d. North wind Barnegat delta LEO delta Cape May delta
Courtesy of Hans Graber, Rich Garvine, Bob Chant, Andreas Munchow, Scott Glenn and Mike Crowley
15m 6 North South
South, offshore flow North Fluorometer
1.0 1 Tidal cycle Upwelling Absorption at 440 nm (m-1) Depth (m) r2 = .95 r2 = .74 6 0 12 30 60 0 Time (hr) POC represents potentially 182μmol oxygen/kg Depleted during an average upwelling
A Clear Box Collaboratory Experiment Dr. Lisa Covi, Social Psycologist The COOLroom Operational Collaboratory COOLroom Skunk Works Model COOLroom War Room Model Evaluate Radical Collocation in the COOLroom to improve Virtual Collocation Systems. Provide guidance for the Regional Collaboratory
# of participating scientists 250 40 40 NOPP & ONR CPSE 200 30 30 Joint Sediment Study 150 # of Research Institutions 20 20 Traditional NSF Ocean Study 100 10 10 50 0 0 0 0 1991 1991 1993 1993 1995 1995 1997 1997 1999 1999 Year
Atmosphere/Ocean Forecast Models 3-D visualization Forecast Briefing Operational Low-Res COAMPS Atmospheric Model Experimental High-Res COAMPS Atmospheric Model Air-Sea Interaction Model ROMS Ocean Model (KPP and MY 2.5 Turbulent Closure) Bottom Boundary Layer Model
Real-Time Ensemble Validation HR COAMPS / ROMS 2 4 6 8 10 12 KPP Depth (m) 2 4 6 8 10 12 26 24 22 20 18 16 14 12 10 8 Depth (m) 2 4 6 8 10 12 Depth (m) MY2.5 18 18 18 19 19 19 20 20 20 21 21 21 July, 2001 July, 2001 July, 2001 Thermistor • In an observationally rich • environment, ensemble forecasts • can be compared to real-time data • to assess which model is closer to reality • and try to understand why.
EcoSim Bio-Optical Model Physical/Biological Models Operational Low-Res COAMPS Atmospheric Model Experimental High-Res COAMPS Atmospheric Model Air-Sea Interaction Model ROMS Ocean Model (KPP and MY 2.5 Turbulent Closure) Bottom Boundary Layer Model
EcoSim 2.0 Model Formulation Air/Sea CO2 Dust Physical Mixing and Advection Light N2 Iron CO2 NH4 NO3 PO4 SiO4 Relict DOM Cocco-litho-phores Benthic Flora Pelagic Diatoms Dino- flagellate Tricho-desmium Synecho- coccus G. breve Excreted DOM Lysed DOM Hetero- Flagellet Viruses Copepod Ciliates Bacteria Sediment Detritus Predator Closure
+ ESSE Flow Diagram ESSE Smoothing via Statistical Approximation ^ DY0/N Field Initialization Central Forecast ^ ^ Y0 Ycf(-) Ymp(-) Shooting Sample Probability Density Measurement Model OA via ESSE Measurement Model Select Best Forecast Options/ Assumptions Mean SVDp Performance/ Analysis Modules Perturbations Minimum Error Variance Within Error Subspace (Sequential processing of Observations) Adaptive Error Subspace Learning + Scalable Parallel Ensemble Forecast Error Subspace Initialization Normalization Peripherals Analysis Modules Convergence Criterion Continue/Stop Iteration Breeding DE0/N + DP0/N - - + Most Probable Forecast + Synoptic Obs A Posteriori Residules dr (+) Historical, Synoptic, Future in Situ/Remote Field/Error Observations d0R0 + - - Data Residuals Measurement Error Covariance ^ d-CY(-) Ensemble Mean + + ^ eq{Yj(-)} Gridded Residules ^ Y(-) + - ^ ^ j=1 Y(+) Y(+) Y1 Yj Yq ^ - Y1 Yj Yq + 0 + - E(-) P(-) ^ - + 0 + + - +/- ^ E0 P0 j=q 0 uj(o,Ip) with physical constraints Continuous Time Model Errors Q(t) Key Ea(+) Pa(+) E(+) P(+) Field Operation Assumption
Hindcast sensitivity studies Large diatoms July 31 SeaWiFS Chlor-a 3 (mg/m ) .5 2 39:30N 3 Node A 4 UCSB 5 Small diatoms 39:00N Measured Total Chlorophyll Measured 3-5 mg Chl a m-3 Diatom Chlorophyll Modeled 2-3 mg Chl a m-3
Red Tide Observed at 790 nm on 22 July 2000 With the PHILLS Sensor 100 meters
Ceratium fusus c a b Bioluminescence Potential 1e6 4e10 Photons/sec/ml 0 6 12 Depth (m) 18 24 a 0 1.0 2.0 Distance (km)
Ship Grid Patterns BL Isosurfaces 1E10 ph/s/35L 0 3E11 ph/s/.35L Depth (m) 15 Latitude (~5km) Longitude (~2km)
BL Isosurfaces 5E10ph/s/.35L 1E11ph/s/.35L Depth (m) Latitude (~300m) Longitude (~500m)
Scientists want real-time observational nowcasts and model forecasts …. DO REAL PEOPLE CARE?
Rutgers Web Site Statistics Data Type Other 14% NODES 13% CODAR 17% Gulf Stream 10% MET 17% LEO 6% Non-profit (7%) Rutgers Web Site Statistics By Hour SST 53% Education (6%) East Coast 19% 8000 June General Public 69% 7000 Military & Government(4%) July 6000 August NYB 37% 5000 Average Hourly Hits September 4000 October 3000 November 2000 MAB 28% December 1000 January 0 Region 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00
Where we do go from here? “The very few existing time-series stations paint a compelling picture of important oceanic changes in physics, chemistry and biology. Yet these stations capture the time domain at only a single point. New strategies for observing the appropriate spatial correlation are required.” -- Ocean Sciences at the New Millennium Ocean Sciences Decadal Committee 2001
Why do we want the spatial correlation? Nowcasting & Forecasting FOR A SINGLE PARTICLE FOR AN OCEAN OF PARTICLES Observations Models Motivation • To get these initial conditions: • Modelers like to assimilate maps of coherent array data • Modelers do not like to assimilate incoherent time series data
New Jersey Shelf Observing System (NJ-SOS) 300 X 300 km NJSOS An Integrated & Sustained Observatory Satellites, RADAR, Gliders
International Constellation of Ocean Color Satellites X-Band Earth Observing Satellites EOS (MODIS) USA 2001 NEMO (COIS) USA 2004 Orbview-2 (SeaWiFS) USA Op. HY-1 (COCTS/CZI) China 2002 FY1-C (MVISR) China Op. FY1-D (MVISR) China 2002 IRS-P3 (MOS) India Op. IRS-P4 (OCM) India Op. ADEOS-2 Japan 2002 (GLI/POLDER) ENVISAT(MERIS) Europe 2002
oC 14 18 22 23 24 25 26 27 FY-1D Temperature 40N 40N 38N 38N FY-1D Chlorophyll-a .1 .3 1 2 3 4 5 6 7 mg/m3 36N 36N 66W 66W 74W 74W 70W 70W FY-1D Sept. 12, 2002 13:38 GMT
Combining the optics and physical features for water mass tagging Here using a mask of temp and optics marking water masses From Oliver (former Mote intern)
Phills _bbb555_Arnone Smoothed July 31, 2001