ArcGIS 10.7 and getting FRXST_OUT points

[Arik Idrisy]

Hello,

Previously for the pre-processing step we had been using ArcGIS 10.3.1 but recently upgraded to 10.7. After that switch our wrf_hydro simulations have not been producing the frxst_out csv. We're using the same wrf_hydro settings as before. The simulation still runs fine, correct number of diag files and what not, there is just no frxst_outs file. Bellow is the exerpt from the run.log.

 before call output_chrt
 before call output_chrt
 WARNING: No forecast or gage points found for frxstPtsOut. No file will be created.
 end calling output functions
 end calling output functions

For more context we use the docker container provided for all of our simulations. Any ideas or help would be greatly appreciated.

Best Regards,

Arik

[Arik Idrisy]

Is there anything else I can provide that might make it easier to get some feedback or assistance?

[Molly McAllister]

Hi Arik, 

That does seem strange. I'll ask our associate, Kevin Sampson for some input regarding your situation. 

Cheers,

Molly

--
You received this message because you are subscribed to the Google Groups "wrf-hydro_users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wrf-hydro_use...@ucar.edu.
To view this discussion on the web visit https://groups.google.com/a/ucar.edu/d/msgid/wrf-hydro_users/647542cd-c427-4ca8-8f99-c6dc5b0bccef%40ucar.edu.

--

Molly J. McAllister

Associate Scientist III

National Center for Atmospheric Research

moll...@ucar.edu
office: (303) 497-8253

My working day may not be your working day. Please do not feel obliged to reply to this email outside of your normal working hours.

[Kevin]

Arik,

I have a few questions to help diagnose the issue, the first set relating to your WRF-Hydro configuration, and the second set relating to the GIS Pre-processor configuration. When you upgraded to ArcGIS 10.7, did you also upgrade WRF-Hydro around the same time? Have you checked your hydro.namelist to ensure that "frxst_pts_out = 1"? If that is not producing the forecast point time-series you want, have you looked at the hydro.namelist option "CHANOBS_DOMAIN = 1" for a netCDF time-series output?

When you built your domain files with the ArcGIS Pre-processing tools, did you use the exact same settings as you did before, including providing gage points as a CSV file to the GIS pre-processor? You may want to check that the frxst_pts variable in fulldom_hires.nc has values other than -9999, and that those values coincide with valid channel cells on the CHANNELGRID variable.

Kevin

[Arik Idrisy]

Hey Kevin,

Thanks for getting back to me. 

When we switched to ArcGIS 10.7 we did not upgrade WRF-Hydro. I checked my hydro.namelist options and tried a few different settings while I was trying to fix the problem. I've had `frxst_pts_out = 1` the whole time and tried switching the CHANOBS_DOMAIN variable as well. 

My colleague does the pre-processing so I don't know the details but he said he used the same steps as he did before. We follow this video https://www.youtube.com/watch?v=yaaKYNhhCbA when we do pre-processing. When I check the fulldom_hires.nc file I do see the -9999 values. Does that mean we're having an error in our pre-processing step?

Best,

-Arik

[Kevin]

Arik,

Would you be able to share a link to your GIS Pre-processing outputs? This would be the .zip file and .log file that came out of the GIS Pre-processor.

Thanks,

Kevin

[Arik Idrisy]

Kevin,

I can't find the .log file but I've attached the zip file. 

EDIT: We weren't saving .log files. If need be my colleague has said he's can re-do the step and save the log file this time.

-Arik

[Kevin]

Arik,

Thank you for posting the files; they were very helpful in diagnosing the issue. Here are my notes: 

1) The reason you are not seeing any forecast points output in the frxst_out csv file is that none of your forecast points are on a channel pixel (value = 0 in CHANNELGRID variable). You will need to move your forecast point by altering the latitude and longitude of the forecast points in your forecast point CSV file that is provided to the GIS Pre-processor. This process is described in the GIS Pre-processing documentation, Section 9.1 "Manual Specification of Station Points". 

2), It looks as though you may have used a high-resolution DEM that had vertical units of cm. See the WRF-Hydro GIS Pre-processing documentation section 5.4.6 "Input Elevation Raster" for more information. You will need to convert this to units of meters. The GIS Pre-processing tools assume the vertical units of your input raster are meters, and because you provided a raster with units of cm the "Fill" process reached the maximum fill limit (default of 1000m) in many locations. The resulting channel network is then highly discontinuous. If you use a DEM with vertical units of meters, you will see a continuous channel network without breaks in the network.

I think following this guidance will help to resolve the original issue you inquired about and provide a more realistic simulation.

Kevin

[Arik Idrisy]

Thanks for the help, we're taking your suggestions into action right now. Ill update the post on its outcome.

[Arik Idrisy]

Hey Kevin,

So we tried the modifications you suggested however we're still getting the same errors. I'm attaching the gis_output.zip bellow in case that's helpful. I don't know if we just failed at making the modifications or if something else happened.

-Arik

On Thursday, May 28, 2020 at 5:29:44 PM UTC-4, Kevin wrote:

[Kevin]

Arik,

Thank you for sending the output files from the GIS Pre-processor. For future reference, please also send the log file that is generated with it. It will have the same name as the .zip file, but with .log appended to the end of it.

I took a look at your files and I see that the forecast point has not moved. You may need to revisit the GIS Pre-processor documentation Section 9.1 "Manual Specification of Station Points" and alter the latitude and longitude position of your desired forecast point in your CSV file so that it is located on a channel cell. Here is what it looks like in the most recent set of files:

You can see that the red cell, which is your forecast point, is not on a channel (the blue and green colored cells). Perhaps when you re-ran the tool, you did not specify the newly edited forecast point CSV file?

Kevin