[alister.hood]

I did also notice that there was something strange about the geometry file my colleague has been having problems working with, but I don’t know if it is related.
Generally inspecting the geometry .hdf files with HDFview shows that they claim their “File Type” is “HEC-RAS Geometry”. But this one said it was “HEC-RAS Results”, and it also had the “Projection” saved in it like a results file does.
I don’t know what caused this, but fiddling around with a bunch of stuff has actually fixed it somehow.

[alister.hood]

OK, I think I mostly understand the problem.
HEC-RAS cannot write to the geometry .hdf if some other process is also using it, but it fails silently.
I tested by opening it in hdfview, but the problem could potentially be the geometry editor, another instance of RAS Mapper, (I know I sometimes open the same project twice by mistake), or maybe an antivirus scanner or something.
So I think most likely my colleague is mistaken when he thinks he closes and reopens and the terrain association is correct. Maybe he has had the project open in two instances of HEC-RAS at the same time and is somehow getting confused.

[alister.hood]

4.5 years later and here we’re finding this very frustrating. Check the terrain association and start a model running overnight, come back the next day and find it has run with the wrong terrain. In my case it always seems that if I set the terrain association, save, close and reopen HEC-RAS and it remembers the terrain association I’ll be able to run it successfully.

But I have a colleague with one particular project that is causing him trouble. He can change the terrain association, close and reopen, and it shows the correct terrain still, but when he runs a simulation the geometry changes back to the previous terrain association. Or if he has deleted that, it changes back to “Terrain.hdf”, which is the first terrain that was created in the project.
This happens reliably on his computer, but if we copy the project to my computer I can change the terrain association and run it fine.

Trying to investigate further, it seems the .prj and .rasmap files don’t store any information about terrain associations, it is simply written into the geometry .hdf file. But when my colleague changes the terrain association it isn’t written to the file, so when he closes and reopens I don’t know how RAS Mapper is remembering that he’s changed it…

[alister.hood]

FWIW I generally find that the search bar does work, at least if I search from here https://therassolution.kleinschmidtgroup.com/the-ras-solution-forum/
I always thought that it didn’t work inside a specific forum, but even that seems to work now.
One thing that seems to produce an error is including a colon : in a search term – which would be quite common when searching for an error message that HEC-RAS has produced.

[alister.hood]

Thanks.

In the end I did it with GRASS – the workflow might not be obvious or particularly user friendly, but it seems to work well:
– create a series of rasters with r.mapcalc
– combine them into a spatio temporal raster dataset
– convert that to a 3d raster
– set correct vertical region settings with raster_3d=
– export to netcdf
I then compress the output using the tools from nco (my examples compress by a factor of 500 to 1000!). I tried compressing with cdo but HEC-RAS complained that all the data had a “duplicate timestep”.

Yes, I did look at HEC-MetVue and I think it would have also been *possible* with either it or GageInterp:
– Create times series input file with HEC-DSSVue
– Create suitable .ctl file after carefully reading GageInterp documentation (there don’t seem to be any examples) https://www.hec.usace.army.mil/software/hec-rts/documentation/GageInterp_UsersManual.pdf
– Use either HEC-MetVue “Gage2Tin” utility or GageInterp to create the temporal raster. Might need to use the same time series in more than one gage location (I’m not sure if it can use a single gage).
But this didn’t seem like a great workflow and these tools don’t support normal coordinate systems, so I would have to work in a UTM zone.

[alister.hood]

What is happening when you try to load them? RAS Mapper rejects them? Nothing is visible? They show in the wrong place?
What happens if you try to load them in a GIS e.g. QGIS? Do they show in the right location? Does the GIS need to ask you what coordinate system to use for them?
I haven’t tested whether RAS supports ecw, but geotiff should normally load fine if both it and the project are correctly georeferenced.

[alister.hood]

What version of HEC-RAS?
Do you think this could be the “Soils Layer not Recognized” known issue listed here? https://www.hec.usace.army.mil/confluence/rasdocs/raski/latest/

[alister.hood]

Did you solve the problem?
There are no links and photos.

[alister.hood]

Not currently – it has been mentioned as a future feature.

[alister.hood]

OK, it doesn’t make any sense, but I see it depends what “additional shapefile data” is imported when creating the layer (i.e. with the “Add Field” button).
– If I don’t add any fields, the CN numbers are mapped as “Final value”.
– If I add a field for CN value, the CN numbers are mapped as “Final value”.
– If I add a field for CN value and another for abstraction ratio, the abstraction ratios are mapped as “Final value”.

[alister.hood]

Just posting for the potential benefit of anyone who encounters this in future:
I had a project generating this error with a 2D model in 6.4.1
Restarting not just HEC-RAS but the computer did not help. Opening the geometry editor it turned out to be just an error about two computation points being too close together; I’m not sure if I somehow managed to save the geometry in RAS Mapper without being prompted about the error, or just overlooked it.

[alister.hood]

Yes, sub-grid bathymetry is what makes HEC-RAS great, and I don’t like to run models with such small cell sizes, but sometimes it is necessary.

[alister.hood]

For the record, on another project I’m getting this error with 500mm cells, but not with 600mm.

[alister.hood]

I use HEC-RAS for detailed modelling of urban overland flow paths – normally I wouldn’t use cells that small, but sometimes local government insist on it because they are more familiar with other modelling software where you need to use smaller cells to get good results.

[alister.hood]

OK, I found an old project where I used 200mm cell size, and 6.3.1 can still run it OK, so the problem obviously isn’t as simple as just the cell size. But my problematic geometry is very simple.

[Author]

Posts