open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi Blue Kenue users,

I'm a marine geology graduate student at the University of Victoria in British Columbia, Canada. I'm having some difficulties with PSed, the Lagrangian sediment transport model included in Blue Kenue. The PSed manual can be found under "help" in the Blue Kenue GUI. Some PSed example files can be found in this post:
www.opentelemac.org/index.php/assistance...tasets-now-available

My PSed simulations are apparently being aborted due to several initialization errors. Unfortunately, I don't understand why any of these occur:

1) errors reading both the current file (UV2.t3v) and the depth file (Depth2.t3s);

2) an error opening the depth file (Depth2.t3s); and

3) a positional error (the selected sediment parcel source is outside the grid).


With the hope that someone can spot any places where I may have gone wrong, I've assembled step-by-step annotated screenshots of my input file inspections, simulation parameter set-up, the error notification and so on. Download these screenshots in either PowerPoint (PPTX) or PDF format:


PowerPoint:​

~$Mullan_PSed Test_July 25 2015.pptx
<drive.google.com/file/d/0B6EWmovNx2s2OUN...g/view?usp=drive_web>
​

PDF:​

Mullan_PSed Test_July 25 2015.pdf
<drive.google.com/file/d/0B6EWmovNx2s2anc...0/view?usp=drive_web>
​

My input data (2D tidal velocities in the Salish Sea) is sourced from a Canadian Department of Fisheries and oceans tidal model. The tidal velocities and finite element mesh were converted into a SELAFIN file (*.slf). Within Blue Kenue, this SLF file was used to create the following PSed input files which you are welcome to download:


1) Bathymetry (*.t3s) file​

Bathy3.t3s
<drive.google.com/file/d/0B6EWmovNx2s2LUN...8/view?usp=drive_web>

​I added the negative signs to the bathymetry values in this file because all the nodes in the model have bathymetry (elevation) values below sea-level. I've also tried a PSed simulation using the positive bathymetry values originally included in the SLF file (I had the same unsuccessful outcome).


2) Flow-depth (*.t3s) file​

Depth2.t3s
<drive.google.com/file/d/0B6EWmovNx2s2RmN...8/view?usp=drive_web>

​The only major difference between this file and the bathymetry T3S file above is that this depth file contains positive values. Since tidal elevation (vertical) changes are small relative to the height of the water column, I figured it was safe to assume that the flow depth at any given node is equal to the absolute value of the bathymetric value (seabed elevation) there. I modified the header of this T3S file slightly to make it match the style of the sample file posted to the forum link I provided above.


3) Current file (*.t3v)​

UV2.t3v
<drive.google.com/file/d/0B6EWmovNx2s2Mkp...U/view?usp=drive_web>

​This file contains time-varying tidal 2D (depth-averaged) velocity components (U & V) for all nodes in a binary format.


The sediment transport simulation attempt I have detailed in the screenshots was simply intended as a test-run. I eventually wish to use near-seabed velocities from the bottom layer of a 3D run of the tidal model to simulate the particle transport pathways at numerous seabed locations for several different grain size parcels.


I welcome your advice, questions or other remarks!

Thank you,
Sean Mullan

Hi

The first point i noticed is taht you generate an binary current file while an ascii current file is needed!

Try to change this first and check the consequences

Hope this helps

Hi c.coulet,

Thanks for your quick response. Are you certain that a binary T3V file is the wrong choice for a time varying UV velocity data-set? My rationale for this choice results from the following manual quotes:

"All PSed simulations require a background flow field. This can be either a steady state flow field or a temporally varying one (such as a tidal flow field). Two input files are required for the hydrodynamics; a depth file, and a file containing the velocity vectors [...] The velocity file is a .t3v file that
contains the velocity components. The input data can be either steady-state conditions or can contain temporally varying data (Page 4, section 4.2, PSed4.3 – Technical Documentation)."

"Files having data that does vary over time must be in binary format (Page 160, Blue Kenue Reference Manual, September 2011)."

"Both t3s and t3v files may contain time-varying data, which is always stored in binary format. Non-time-varying data may be stored in ASCII or binary format (Page 166, Blue Kenue Reference Manual, September 2011)."


Any suggestions are greatly appreciated,
Sean

Hi

I doesn't read the PSED manual so you're probably right.
My remark just comes from the screen copy you've made where we could read : Error while reading the Ascii current file ...
But this could also be a mistake in the code...

I've no other idea about your problem.
Probably only the Blue Kenue / Psed developers could help you on this particular point

Regards

Thanks c.coulet

Sean, I'll look at your files tonight to see what's going on. I'll post the solution(if possible) here on the forum and by email. Cheers, Julien

Thanks Julien! You've been so much help. I'll probably be obtaining a slightly different version of the DFO tidal model today. I'll let you know how it looks once I run it through your data conversion code.

Sean: I would suggest to have a time-varying depth file as well. Let me know if that works.

Using a new and improved modeled tidal velocity data-set, I've attempted to apply Julien's advice about using a time-varying depth (flow-depth) file in my sediment transport simulations. Unfortunately, the simulation still ends up being aborted. Here are the initialization errors causing the failure:

1) errors opening and reading the binary current file;

2) errors opening and reading the binary (flow-)depth file; and

3) the sediment source is outside the grid (positional error).

I've once again annotated step-by-step screen shots of my attempt to set-up the PSed simulation:
​
Mullan PSED test AUG 14 2015.pdf
drive.google.com/file/d/0B6EWmovNx2s2cGx...U/view?usp=drive_web
​
Mullan PSED test AUG 14 2015.pptx
drive.google.com/file/d/0B6EWmovNx2s2X0F...M/view?usp=drive_web
​
Does anyone spot places where I've gone wrong in the set-up?

Here are the files I used:

1) Bathymetry file (T3S). These "elevation" values don't change with
time. They are all zero or negative (at or below sea-level).
​
Bathy_negative.t3s
drive.google.com/file/d/0B6EWmovNx2s2bmV...k/view?usp=drive_web
​

2) Flow-depth file (T3S). Time-varying format (binary/multiframe). Each of the 1415 hourly frames contains the same depth values. These numbers are
the absolute value of the bathymetry values at each node.
​
FLOW DEPTH_binary_multiframe.t3s
drive.google.com/file/d/0B6EWmovNx2s2ajI...k/view?usp=drive_web
​

3) Current file (T3V). Time-varying U and V tidal velocities (binary
format).
​
VELOCITY UV_binary_multiframe.t3v
drive.google.com/file/d/0B6EWmovNx2s2dGY...M/view?usp=drive_web
​
Each of these files seem to work fine within Blue Kenue. Here are some animations illustrating the 2D tidal flow-patterns in my Salish Sea (British Columbia and Washington State) study area. If your connection is fast enough, these velocity velocity videos are best viewed in 1080p resolution:

1) 2D UV - Boundary Passage:


2) 2D UV - Boundary Passage Close-Up:


3) 2D UV - Connecting Straits of the Central Salish Sea



The physical oceanographer who ran this tidal model will soon be providing me with 3D results (10 vertically stacked layers of variable thickness) over the entire finite element model domain. I'd like to use the bottom-layer results as inputs in PSed simulations. These runs will accompany PSed simulations done with the 2D (depth-averaged) results I've provided and discussed here. I'm seeking to explain the distribution of seabed sediment types, landforms and observed morphodynamics (from repeat multibeam sonar surveys). As a geologist, I'm new to the realm of hydrodynamic and sediment transport modelling. I welcome any thoughts forum members may have about the design of sediment transport modelling studies (especially with PSed). Here are several things concerning me:

1) Is it reasonable to use bottom layer (3D model) velocity results in a PSed simulation, since the software expects 2D hydrodynamic model inputs? I'm uncertain how well the logarithmic decay of velocity towards the seabed will be modeled if the 3D model's depth-averaged bottom layer input velocity values are "placed" in the middle of the entire water column by PSed. My guess is that the near-seabed velocities will be underestimated. The water column in my study area ranges from ~0 to 290 m deep and dramatic bathymetric changes exist along transects of only a couple hundred meters (i.e. there are steep parts). At the deepest node locations, the thickness of the bottom layer (of 10) in the 3D tidal model is therefore ~29 m.

2) Related to #1, how reasonable is it to assume (at least for fine sand and coarser) that sediment is confined to the bottom layer (vertical diffusion isn't an issue). This seems to be an assumption necessitated by treating the bottom layer results from a 3D tidal model as 2D results (for the entire water column) in PSed.

3) Is the 1-hour time-step I selected so large that it could jeopardize the PSed simulation's "numerical stability?" (for my set-up parameters see the document mentioned earlier). The PSed technical documentation indicates that a time-step of ~1 second or less may be appropriate for a coastal setting. I selected 1-hour because it matches the input hydrodynamic tidal model's time-step and because I was concerned about generating huge file sizes (at least until I figure PSed out). Could this be where some of my woes are arising?

I appreciate any thoughts or advice offered,
Sean

Hi Sean:
1.PSED does not work in a geographic coordinate system but in a metric projected coordinate system. You can change this in BlueKenue:Right-Click data->Properties->Spatial Tab->Assign…->LatLong&NAD83.Press OK. Select Combo-box and change LATLONG to UTM.
2.BlueKenue uses OpenGL libraries for display and works only in single-precision (7-digits). If you work in UTM, the display will look weird since of round-off errors. Hence, I would shift everything near 0,0. In BlueKenue:Right-click->Edit tab->Shift x/y (-40000,-5300000)
3.Resave your bathymetry, depths,velocities.

Test case:https://www.dropbox.com/s/t82j2d2p1zlo6bd/Sean.zip?dl=0

Cheers, Julien