Anyone go down the HDF5 rabbit hole?

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

Anyone go down the HDF5 rabbit hole?

Toby
Greetings,

I am toying with the idea of using the hdf plan output files and taking them directly into program R for post processing.  

I can bring in a depth map and create a raster from the HDF5 layers, but have not thought through how to create a velocity map.  I know of many interpolation methods to better create a raster, but have only used a simple quick and dirty method for now.  

Two questions for all of you....

1.) does anyone know what type of interpolation RAS uses to create raster data from irregular cell spacing?

2.) does anyone know how RAS renders the static velocity map one sees in the RAS Mapper?

example code below...
https://www.dropbox.com/s/7mpp9gh6cr72g2f/hdf5_HEC_RAS.R?dl=0

Thanks!
Reply | Threaded
Open this post in threaded view
|

Re: Anyone go down the HDF5 rabbit hole?

osterwald
Hi,

do you have any progress on this?
I am trying to create the depth map myself and am wondering the same questions like yours.

Until now I could export the computational points with the depth in hdf5 but the depth is based on the minimum elevation of the cell (cell 25 m based on 5 m terrain) so I assume the better way should be we average the elevations for 1 cell, then use the WSE to substract the averaged heights?

Again, if anybody knows how HEC-RAS interpolates the depths for sloping rendering option I would very thankful to know.
Reply | Threaded
Open this post in threaded view
|

Re: Anyone go down the HDF5 rabbit hole?

cameron
I will have to double check, but I believe it calculates the WSE for each cell and back calculates the depth from that and the terrain.
Reply | Threaded
Open this post in threaded view
|

Re: Anyone go down the HDF5 rabbit hole?

osterwald
cameron wrote
I will have to double check, but I believe it calculates the WSE for each cell and back calculates the depth from that and the terrain.
Yes it is as far as I could figure it out!

But my problem is that i have a 25 m calculation mesh over 5 m terrains so if I use the horizontal render mode it will show many fragmented inundations. The sloping render mode will just overestimated the depth at the borders to higher elements in the terrain. The hybrid mode's export function does'nt work so I would like to know which algorithmus is behind the sloping mode so I could implement it myself to get the desired output.
Reply | Threaded
Open this post in threaded view
|

Re: Anyone go down the HDF5 rabbit hole?

cameron
export out the cell center as a point shapefile, assign a WSE to each cell center, then create a wse raster from the points. You can then subtract your terrain from the WSE to get a sloping surface.

To help, set the depths at which it displays from 0.001 (default) to -9999.
Reply | Threaded
Open this post in threaded view
|

Re: Anyone go down the HDF5 rabbit hole?

osterwald
cameron wrote
export out the cell center as a point shapefile, assign a WSE to each cell center, then create a wse raster from the points. You can then subtract your terrain from the WSE to get a sloping surface.

To help, set the depths at which it displays from 0.001 (default) to -9999.
Thank you very much. I did it but it doesn't change the fact that fragmented inundation areas will still exist.

Anyone has any idea about the interpolation algorithmus behind the sloping render method?
Or did anybody has some success in exporting the depth using "hybrid render method"? For me it doesn't work and just give me the old sloping values. :(
Reply | Threaded
Open this post in threaded view
|

Re: Anyone go down the HDF5 rabbit hole?

cameron
You can not export out the results with the hybrid method that I am aware of.

The description below is from HEC and what the hybrid method does.  

"The hybrid rendering method takes the best parts of the sloping method and
attempts to make it more accurate when it detects the potential for failure.
We do this with a couple of tricks to raise and lower the water surface as
needed. When shallow flow is detected, we fall back to the horizontal
rendering mode for that particular 2D cell since the horizontal method renders
low-volumes much more accurately. This will show the water stair-stepping,
but won't incorrectly show large areas of dry land. We also add extra weight
to the water-surfaces computed at the center of a face, so the downhill ravine
case won't be so dramatically affected by the high-ground face points. Finally,
we make an adjustment to the computed face-point values to drag them up
or down so their average matches the computed cell water surface. This
allows us to create a sloping water-surface, but ensures that we don't
dramatically over-estimate or under-estimate the rendered volume."
Reply | Threaded
Open this post in threaded view
|

Re: Anyone go down the HDF5 rabbit hole?

osterwald
Was it not supposed to be exportable with the hybrid-option?

As I could read it here (http://hecrasmodel.blogspot.com/2016/10/creating-static-results-maps-shapefiles.html), someone reported the bug to the development team and got this reponse in 2016:

"Thank you for reporting this problem. I have been able to reproduce the issue here with a separate dataset and have submitted the bug to our development team. The bug should be corrected in the next HEC-RAS release."

But unfortunately it was not be fixed.

I reported it again at the middle of april but no response until now.
Reply | Threaded
Open this post in threaded view
|

Re: Anyone go down the HDF5 rabbit hole?

cameron
I have never been able to do it. It does not represent reality as they are tweaking things to just make it look nicer.

The only issue with render mode and exporting out rasters was part of the 5.0.5 fix. The link below shows all of the fixes since 5.0.0

https://www.hec.usace.army.mil/software/hec-ras/documentation/HEC-RAS_5.0.7_Release_Notes.pdf