Category Archives: Modeling

Delft3D model simulations posted

From Ad Reneirs & Patrick Rynne:

We’ve posted the latest model predictions for NRI for the 28th-29th at the link below.
Please click on the .zip file link in the center of the page.

we are driving up today, predictions will be more timely in the days ahead once we get

-Patrick & Ad

Below is an image on flooding tide from the model simulation showing current vectors on top of the bathymetry from Apr 16/17.    Although the comparison is qualitative, note the difference in how far offshore the ebb tide jets relative to the NEARCOM simulation.

New NRI NEARCOM model results on the Apr 16/17 Bathy (Tom, Julie, & Fengyan)

The UDEL folks have posted new NEARCOM model results on their website.   The basic conditions are the same.  One M2 tidal cycle, with normally incident Hsig=1 m, and then varying the drag coefficient.     It is hard to tell from these simulations how far offshore the tidal jet goes.    We will be simulating drifter trajectories on these model currents and posting that soon.

Delft3D simulation on the new FRF bathymetry – first results (w. Rynne & Reniers) – updated

Today we have near gale force winds at New River Inlet.     But there are model simulations to work up.     As readers will recall, the FRF did a recent (Apr 16/17) bathymetry survey after the dredging of the main channel.   The upshot is that the main channel is still very shallow and narrow.     We have done two test dye releases (see earlier posts).   In contrast to model predictions based on bathymetry ~ 1 year old (see earlier posts w/ NEARCOM tide / drifter  and Delft3D tide / dye  results) where there is a strong tidal jet that strongly advects stuff many km offshore,  the dye both time just ooozed over the southern ebb shoal or out the new channel and northern part of the shoal.    Clearly there was no strong tidal jet.     Also,  Britt & Steve also have found that tidal velocities are basically in phase with tidal elevations, ie, progressive, whereas the models were predicting a 1.5 hr lag (half progressive/standing).   So both NEARCOM and Delft3D are being re-run with the new bathymetry.   Here are some preliminary results from Delft3D run with just tidal forcing and a dye release.

First analysis of the tides.   See the image below.   Contrast this with the analysis on older bathymetry

As can be seen the tidal velocities are very strong inside the inlet but decrease rapidly at and offshore of the shoal (not here log10 velocities is colored).   Bathymetry contours (4, 6, 8 m depth) are shown as dashed.

More interesting is the dye simulation.   Now ~42 hours (previously only 20 hours) of the model has been finished, however this dye simulation is very different from previous simulations on the old bathy. See the youtube video below. It is clear that the dye behaves very differently oozing evenly out over the entire shoal rather than in a jet.   The dye does not make it as far offshore.    This is qualitatively consistent with our preliminary dye deployments.   Note that there is no wind or stratification in this run.  All the other model parameters (eddy diffusivities, etc.) were the same as the previous run.

For ease of comparison, take a look at the dye simulation with the old bathymetry but the same conditions (below). Clearly quite different.


Delft3D dye simulations of New River Inlet (w. Ad Reniers & Patrick Rynne)

The youtube movie below shows Log10 dye concentration (no units) at New River Inlet released just up the inlet over t=2-4 hr.    This simulation is forced by the M2 tide and has no stratification, no wind, and no waves.  Top: time series of tidal elevation at (x,y)=(0,0) m.   The red dashed line represents the current time location.    Bottom:  plan view of Log10 surface dye concentration.   Only a subset of the model domain is shown.   Land values are white and dashed lines represent bathymetry contours (2, 4, 6, 8, 10, 12 m depth).    The colorbar gives log10 concentration: Check the video below:

I also looked at the bottom dye concentration.   Although there are subtle differences the movies look pretty much the same.   For these purposes this unstratified simulation is pretty well-mixed.

At first release, the dye jets about 2 km offshore of the inlet mouth.    There is a significant amount of dye flux back into the inlet.    However, this may be a result of no wind, no waves, no  alongshore tidal propagation, and no other shelf-scale forcing mechanisms that drive shelf-scale alongshore currents.

If folks are interested in the .mp4 video file (better resolution) it can be downloaded via NRI_dye_surface_delft3d

Please feel free to leave comments and suggestions.   -falk

NEARCOM tidal analysis – delta t from high tide to water ebb (w. Matt Spydell)

In the previous post on Delft3d model output,  we showed in one of the panels, the time it takes from high tide until the water starts to ebb.    Notationally this could be written as t_{high tide} – t_{Uebb},  where t_{Uebb} is the time when the alongchannel tidal velocity switches to ebb (out of the inlet).     This time lag is related to the M2 eta/V phase lag, but is slightly different due to skewness in the tidal velocities related to the water level going up and down.    To better compare the two models, we made the same calculation with the 3 NEARCOM model runs – see previous posts.  They are  (a-top) waves+tide, (b-mid) tide only, and (c-bottom) 10X drag coefficient, waves+tide.   The results are shown below.  The left panels are the time lag from high tide to velocity, the R panel is the std of alongchannel velocity (the same as in previous posts).

The results of the tide-only run are broadly consistent with the analysis of the Delft3D model output.   Although the coordinate system that NEARCOM and Delft3D are using are different, the pattern of time-lag is similar for the tide only case.   Note also that for 10X higher drag coefficient, the tidal velocities and phase radically change.    Both Delft3D and NEARCOM model data are output at 1/2 hr intervals (99% sure).   This limits the resolution of estimating this time lag.    However, based on the various simulations, it appears that an average of 2 hr lag inside the inlet is appropriate.   -Falk

Delft3D model tidal analysis (w. Ad Reniers & Patrick Rynne)

In order to understand how to time drifter and dye releases, I’ve become preoccupied with the New River Inlet tidal currents and the phase lag relative to the tide level (call this eta).   As is well known a standing wave has a 90 deg (pi/2) phase lag between eta and velocity and a progressive wave has a 0 deg lag.    In a tidal inlet, the tide is somewhere in between the two.    Another way of asking the question is how soon after high tide will the water velocity ebb (go out the inlet) – that is.  time of max (eta) – time when U switches to flood.   Here, numerical models can help us, although one must be careful as the tide-velocity phase relationship depends on the friction inside the inlet.

In a previous posts, NEARCOM model results were shown.    Ad Reniers and Patrick Rynne at Univ. Miami have kindly provided me with some Delft3D M2 tidal model simulations of the New River Inlet.  I believe this is the same bathymetry as for the NEARCOM model runs.     Patrick sent me the data over 4 M2 tidal cycles output every 1/2 hour.    Results are shown below.     The panels are (top,L ) bathymetry, (top,R) M2 tide magnitude [i.e., eta = eta_0 cos( omega t) ] ,  (mid, L)  major axis velocity magnitude, (mid,R) minor axis velocity magnitude, (bottom, L) phase lag between eta and major axis velocity and (bottom, R) time in hrs from high tide to when water starts ebbing.

Note that there are similarities and differences between the Delft3D and NEARCOM simulations.    From these plots it is hard to tell what the tide magnitude, currents, and phase is within the inlet.    Below is the same quantities at the deepest portion of the inlet at any cross-shore position.  The bathymetry panel (top, L) marks the deep channel locations with white asterisks (squint cause they are hard to see).

As expected, the tide magnitude decreases as one goes up the channel (top, R).  This is fairly linear.   The major axis velocities (~ 1 m/s) also sorta decrease but oscillates, because the channel depth is not regular.    The minor axis velocities are small.    The phase lag is interesting.   Near the mouth (x ~ 0 m), the lag is ~21 deg – far closer to a progressive wave than a standing wave.   This corresponds to a 2.5 hour lag between high tide and water velocity beginning to ebb (the phase difference and time lag are not excactly related because of nonlinearities and skewness in tidal velocities).   The phase slowly increases up the inlet (more negative x).  And the time to go from high tide to ebb velocity also decreases.

We will compare these results with the other models (NEARCOM and ADCIRC) coming soon.  -Falk


NEARCOM model analysis of Tides by Tom Hsu et al.

by Tom Hsu, Fengyan Shi, and Julie Chen (UDEL):  I just like to follow up with Falk’s comments on time series and phase lag between surface elevation and flow velocity magnitude. More information from the model run is attached. In this document, you will find time series at four different locations. Please also notice that in these figures, we show tidal velocity magnitude (not individual components), which is probably more representative. Basically, max ebb occurs 0.5~1 hour before low tide.

As Falk mentioned, the results seems to be sensitive to bottom friction used. We are evaluating the sensitivity and will give you an update soon. We also plan to run realistic tide, which can make a difference in phase. Once we have update, we will post on the blog (and we may provide a link where you can download more specific and detailed data from our own website).

Bathymetry:  from Summer 2011

Offshore Conditions:   Normally incident waves Hs=1 m.  Drag coef Cd=0.01, Tide Level  =0.623 m


Max Flood

Time series of tidal elevation and tidal velocity magnitude:

Point 1:

Point 2:

Point 3:

Point 4:

drifters on NEARCOM model currents

In order to get an idea of the offshore extent of the tidal jet, exchange, and overall dispersion, Matt Spydell has taken the NEARCOM (tide+waves) model output provided by Tom Hsu, Fengyan Shi, and Jia-Lin Chen @ UDEL and released numerical drifters inside the inlet at the start of the ebb tide.  These drifters are advected by the currents and also randomly dispaced in a random walk (for details see Spydell et al. JGR 2012 in press) with a diffusivity of 0.1 m^2/s.   Arguably this is small for the inlet but ….

The top panel shows the tidal elevation.  time=0 is 1.5 hr after high tide.  The panels below show the numerical drifters as red dots.  At t=1min after release, they are tightly clustered near the release location.   The drifters then get ejected out the inlet after 1 hour.   They spread and eventually some start to get pulled back to the inlet on a flood tide (t =480 min, 8 hrs after release).

It is interesting to compare this to releasing drifters right at high tide (1.5 hours before the release above).   Now, the drifter rush all the way up the inlet before turning around on ebb tide and being ejected roughly 6 hours after release.

Note, in this simulations after 8 hrs, some drifters are more than 3 km from shore!

NEARCOM tidal simulation results

In order to plan drifter and dye deployments as part of RIVET at new river inlet, we’ve been looking at model simulations from UDEL (Hsu et al. – NEARCOM),  U Miami (Reniers et al. – Delft3D) as part of the DRI, and also UNC (Leuttich et al. – ADCIRC) who have graciously given us some model simulation data.    We’ll be posting results from each of the simulations here.

First up is NEARCOM (U Del).    They ran 3 scenarios.   They are:

  1. M2 Tides + normally incident waves (offshore Hsig = 1 m)
  2. M2 Tide only (no waves)
  3. M2 Tide + waves but with drag coefficient 10X larger

The (U,V,eta) over the inlet domain was decomposed into principal components to get major axis and minor axis tidal ellipses.     Then the phase lag between the tide level (eta) and the major axis tidal velocity was computed (zero is a progressive wave, 90 deg is a standing wave).   Below is shown (left panels) the phase difference and (right panel) std of major axis velocity (multiply by sqrt(2) to get peak amplitude)  for the 3 cases above.

Things to note:

  1. The waves do not seem to have a strong effect on the tidal currents
  2. Friction makes a big difference
  3. It is hard to tell how many hrs the velocity lags the tide level in this plots.   Soon we will have ADCIRC and Delft3D runs analyzed to compare to.   This will make it easier to see.

For the tide+waves NEARCOM (#1 above) run a time series is a little clearer.  Below is a eta (blue), U,V (red,green) timeseries over 12 hours of at one location inside the inlet.

Note how the maximum in tide occurs at t=4 hrs, but the velocity does not start to ebb (switch sign, see red) until t=5.75 hrs or so.