*** Error in `./wrf_hydro.exe': free(): invalid next size (fast): 0x000000001e1a1e20 ***

[Yu-Fen Huang]

Hi,

Does anyone know what this error means and what can I do to resolve the issue?

`*** Error in `./wrf_hydro.exe': free(): invalid next size (fast): 0x000000001e1a1e20 ***`

With the same domain and forcing, I have no problem running Diffusive-wave gridded routing. However, this error occurred when I changed the channel routing to either reach-based routing. 

Note: I attached the log files and namelists.

Thank you,

Yufen

[Yu-Fen Huang]

Hi,

Just add a note: the error occured after reading the Route_Link.nc.  Does anyone have the same experience or know how to solve the issue? Much appreciated.

Best regards,

Yufen

--

Yu-Fen Huang

Department of Natural Resources & Environmental Management

University of Hawai‘i at Mānoa

Personal website | Tsang Stream Lab

[Kevin]

Yu-Fen,

You are using the same domain files for gridded diffusive-wave and reach-based routing? That is not usually supported, because reach-based routing requires the presence of a LINKID variable in Fulldom_hires.nc. Can you check your inputs to make sure that variable has reach ID values that are present in your RouteLink file?

Kevin

[Yu-Fen Huang]

Hi Kevin,

Thank you for the reply. I have both `LINKID` in `Fulldom_hires.nc` and `link(feature_id)` in Route_Link. I checked the box of "Crate reach-based routing (RouteLink) files? (optional) in the GIS preprocessing. I attached the metadata of those two netCDF files.

Side note: If I borrow the `spatialweights.nc` from NWM v2.1, turn on the UMAP, and change the dimension name of `GWBUCKPARM.nc` to make it consistent, then I can run the Musk-Cunge reach routing with my domain without that error, though I'm not sure if the the linkages between stream and the watershed will be correct or not.

Thank you,

Yufen

[Parth]

Hi Yu-Fen and Kevin,

I have run into similar issues. It turns out a similar set of DOMAIN files works perfectly alright with the earlier WRF-hydro version (in fact, the 5.2.0 beta2). Have you come across any solution? 

The issue here persists with the Route_link.nc with both gcc and intel compilers on the latest version. I am using all the tools from the current 5.2 version (soil_properties, GIS preprocessing toolkit etc.). 

Below attached is the slurm error output. I would be happy to share any other data if needed. Any help would be greatly appreciated.

Thank you so much,

Parth

[Yu-Fen Huang]

Hi Parth,

How many cpu did you use for your run? I found out that the error may be gone by reducing the cpu for parallel computing. For me, I'll need to reduce it to 4.

Best regards,

Yufen

[Parth]

Hi Yufen,

It turns out that the HPC (Summit, in my case) had some compiler issues with specially configured ifort Summit macros in V5.2. Instead, I used the regular ifort macros (not the specially configured one; primarily, it changed my COMPILER90 from mpiifort to mpif90), and it is working perfectly now. I appreciate your help!

Parth