Objectively mapped salinity based on dive profiles
Bird GPS fixes, 4 June 2014 (19:20 – 19:40 UTC)
Comparing a High-Resolution Coupled Wave-Current Ocean Model and Biologging Derived Oceanographic Data
Mouth of the Columbia River (MCR) Observations and Modeling
Project Description and Objectives
- Biologging of marine animals offers a distributed, autonomous, and scalable platform for acquisition of oceanographic measurements in remote coastal regions around the globe.
- Our long term goal is to gain a quantitative understanding of the long-term bathymetry, temperature/salinity profiles, wave/current interactions at complex coastal sites (e.g. tidal inlets) using a combination of numerical models and tagged marine bird observations.
- Our current objective is to compare the results from the numerical model and the remote sensing observations. After the validation of the model physics for our study periods, obtain an estimate of the underlying bathymetry and ocean state via data assimilation and inversion methods.
Dorukhan Ardağ1, James Lerczak1, Greg Wilson1, Tuba Özkan-Haller1, & Dylan Winters1
1College of Earth, Ocean, and Atmospheric Sciences, Oregon State University, ardagd@oregonstate.edu
Adam Peck-Richardson2, Don Lyons2, & Rachael A. Orben3
2Department of Fisheries and Wildlife, Oregon State University
3Department of Fisheries and Wildlife, Oregon State University, Hatfield Marine Science Center
Comparisons Between the Model Results and the Biologging Data
Preliminary tests at MCR for May 23rd to June 20th 2014 and June 5th through 27th 2019:
- ROMS Time Step: 15 s, Courant ~0.8
- Mellor/Yamada 25 mixing.
- Vtransform=2, Vstretching=4, θs=5, θb=0.4
- SWAN time step: 15 min.
For scatter-plot and vertical transect comparisons, MCR is divided into an outside and inside region to investigate spatial influence on model skill. Observations were also separated into different time regimes to examine the impact of monthly tidal cycles (i.e. spring & neap tide). The model results are then interpolated horizontally and vertically to match the bird data spatiotemporally.
ROMS-SWAN Coupled Modeling for MCR
- Coupled Ocean–Atmosphere–Wave–Sediment Transport (COAWST) Modeling System (Warner et al., 2010):
- Ocean Model: Regional Ocean Modeling System (ROMS)
- Wave Model: Simulating WAves Nearshore (SWAN)
- 426 x 312 (Curvilinear) x 40 Vertical levels
- 200m Resolution (max depth ~ 530 meters)
Initial & Boundary Conditions (IC & BC):
- (2014) West Coast Operational Forecast System (WCOFS), Hybrid Coordinate Ocean Model (HYCOM) Region 7
- (2019) LiveOcean
- Tides: OTPS (OSU Tidal Prediction System)
- Climate forcing: NAM-Regional forecast model - 12 km resolution
- River discharge: USGS 14246900
- 2D Spectra: NDBC Buoy 46029
In-Situ Model Comparisons (2014)
- Biologging provides valuable bathymetry, temperature/salinity profiles, and air/sea temperature contrast, across a wide area and over long time periods (months to years).
- Comparisons between the biologging temperature data and model results are mostly in agreement. Model shortcomings are mostly attributed to the misrepresentation of Initial Conditions.
- The spatially and temporally diverse dataset gathered by cormorants is thus valuable to assimilate and improve operational model results.
Three cormorant species were tagged:
2013 & 2014
- 18 Brandt’s Cormorants (coastal/marine)
- 24 Double-crested Cormorants (estuarine/freshwater)
2019
- 2 Pelagic Cormorants (coastal/marine)
- 22 Brandt’s Cormorants (coastal/marine)
GPS Fixes for all recorded dives
for 2013 and 2014 by tagged birds
Brandt’s Cormorant with backpack sensor attached (Photo taken by Adam Peck-Richardson)
Bird GPS fixes, 27 May 2014
Mouth of Columbia River. Highlighted GPS fixes indicate dive profiles shown to the right.
Easting distance along transect (km)
For more information about our project and/or to see the location of the tagged birds please go to our website below or simply scan the QR code with your phone
osudashcams.com
COAWST Model was setup and calibrated by Akan et. al (2016 & 2017) for 2005 and 2013 (shown here), as part of DARLA (Data Assimilation and Remote Sensing for Littoral Applications), on OSU servers:
- Saturn 1 is located closely to where the salt wedge stops propagating after each flood cycle hence there is a lot of vertical mixing. During neap tide salt-wedge moves further upstream and the difference between the observed and the modeled water temperature gets more pronounced.
- In 2014, there was no river temperature data to utilize in the model, thus, a long term average was used. WCOFS daily averages were used for IC and BC. (Courtesy of Dr. Kurapov)
- SWAN boundary conditions were created by using the 2D Spectra from NDBC Buoy 46029 for Western, Southern and Northern boundaries
- SWAN significantly slowed down computations
Acknowledgements: The authors wish to acknowledge following groups, organizations and individuals that were instrumental in the 2013 & 2014 data collection campaign (left column) and those provided valuable insight, data access or computational resources throughout the modeling process:
Easting distance along transect (km)
Biologging Tag Specifications
Tags Used in 2013 & 2014
- Earth and Ocean Technology (Kiel, Germany)
- Integrated GPS, pressure, temperature tag
- Archival GPS-Tdlog (37g)
- “Smart” GPS shuts off when pressure indicates dive. 0.5-1 Hz when at surface.
- Star-Oddi (Garðabær, Iceland)
Tags Used in 2019
- TechnoSmArt (Rome, Italy)
- Smart GPS, regulated by the pressure sensor (27g)
- Solar cell battery recharging
- Cell phone data transmission
- 9-axis inertial motion unit (IMU; 3-axis acceleration; 3-axis gyro; 3-axis magnetometer)
- Ornitella (Vilnius, Lithuania)
- Solar cell battery recharging (30g)
- Cell phone data transmission
- Smart GPS with Geofencing capabilities
- Dan Roby - USGS Oregon Cooperative Fish & Wildlife Research Unit
- Ken Collis - Real Time Research
- Bird Research NW - Astoria Field Crews
- USACE - Portland District
- Dr. Çiğdem Akan – University of North Florida
- Dr. Alexander Kurapov – NOAA
- Dr. Parker MacCready – University of Washington
- Dr. John Warner – USGS
- This work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number: TG-OCE190003
- Center for Coastal Margin Observation & Prediction (CMOP) Network
These results indicate a moderate change in model skill spatially and, as expected, higher temperature range outside the MCR. Just like the in-situ comparisons, during neap tide model skill decreases. These results point out an apparent difference in observed and modeled ocean water temperature. Hence the two scatter plots on the right show model comparisons initiated with different IC.
Same time period and area with different initial conditions
Modified WCOFS IC: Temperature -2oC
Vertical temperature shear observed by the birds (blue line) is captured only after modifying the IC in the model (red and yellow lines)
Vertical Transect Comparisons
This 20 minutes long transect was just outside the mouth, (Low Tide was at 18:34 UTC that day). The cormorant was able to capture both vertical and horizontal temperature shear. The model on the other hand shows a well mixed temperature transect.
To address the low model skill during neap tide, it was initiated with different IC.
- Data assimilation (Wilson et al., 2012, 2014; Moghimi et al., 2016) is being used to combine model predictions with tagging observations:
- Bathymetry can be retrieved at MCR from surface currents while accounting for effects of stratification
- Temperature profile data (above figures) suggests a sensitivity to IC errors, which is being explored for assimilation
- In Summer & Fall of 2019, more birds were tagged at the MCR, (some are still active). The current effort will be extended to include that dataset and SAR imagery. Other sensors (e.g., inertial measurement unit, IMU) are now being added to the tags to extend their capabilities
- Tests with higher resolution (40m) will be completed to investigate if differences between the observations and the model results occurred due to the lower resolution.
- Other littoral sites are being considered for future tagging efforts in this project, including Alaska, UAE, and Southwest African Coast
VAR RMSE SKILL
--------------------------------------
S_JettyA 4.47 0.95
S_Saturn1 9.89 0.79 T_JettyA 1.25 0.89
T_Saturn1 12.56 0.64 Z_Hammond 0.23 0.98
- In 2014, birds were captured during mid-nesting season (Peck-Richardson, 2018). In 2019 they were captured from May-September near the end of nesting to collect data during post-breeding dispersal.
- In 2014 they were caught at night by hand from blinds. In 2019 cormorants were captured on the colony, using spotlights and landing nets at night, or with nets during the day.
- Tags were attached via a backpack harness. Archival tags were removed 4-5 days after deployment. GSM tags transmit data and are expected to remain attached for up to a year.
In-Situ Model Comparisons (2019)
2019 preliminary Modeling results are shown here in order to:
- Show the model skill using LiveOcean
- Emphasize the observed and modeled ocean temperature difference at Saturn02 station (deep water at 35m).
Comparisons for 2013 is shown on the left WHOI Pod-1 location is marked above (Akan et.al 2017).