WRF v4.6.1 Failing at wrf.exe with SRTM 30 m Terrain for Bologna (333 m Resolution)

G

gaddamnagaraju_112

New member
Dear All,

Greetings of the day.

I am attempting to simulate the WRF (version 4.6.1) model at 0.333 km resolution for Bologna to study urban heatwaves. For this purpose, I am using high-resolution SRTM (30 m) data combined with SRTM (90 m) and cglc_modis_lcz datasets.

I downloaded and properly converted the datasets as described in the provided here. The model successfully runs up to real.exe, but when I submit wrf.exe, the model fails.

I would kindly appreciate any suggestions on how to run the simulation successfully at this high resolution.
Additionally, I have already implemented several recommended configurations:
  • Set smooth_cg_topo = .true.
  • Adjusted epssm = 0.2 (tested up to 0.9)
  • Set w_damping = 1.
  • Applied adaptive time-step settings
For your reference, I have attached the folder, it contains:
  • WPS namelist
  • WRF namelist
  • GEOGRID.TBL
  • rsl error/output files
I would greatly appreciate your assistance in successfully running the model at high resolution for Bologna.

Note: The model runs perfectly without SRTM data, but the topography representation is not accurate. Therefore, we are attempting to include SRTM data to improve the terrain representation.

Thank you for your time and support.

Sincerely,
Nagaraju Gaddam
 

Attachments

gaddamnagaraju_112 said:
Dear All,

Greetings of the day.

I am attempting to simulate the WRF (version 4.6.1) model at 0.333 km resolution for Bologna to study urban heatwaves. For this purpose, I am using high-resolution SRTM (30 m) data combined with SRTM (90 m) and cglc_modis_lcz datasets.

I downloaded and properly converted the datasets as described in the provided here. The model successfully runs up to real.exe, but when I submit wrf.exe, the model fails.

I would kindly appreciate any suggestions on how to run the simulation successfully at this high resolution.
Additionally, I have already implemented several recommended configurations:
  • Set smooth_cg_topo = .true.
  • Adjusted epssm = 0.2 (tested up to 0.9)
  • Set w_damping = 1.
  • Applied adaptive time-step settings
For your reference, I have attached the folder, it contains:
  • WPS namelist
  • WRF namelist
  • GEOGRID.TBL
  • rsl error/output files
I would greatly appreciate your assistance in successfully running the model at high resolution for Bologna.

Note: The model runs perfectly without SRTM data, but the topography representation is not accurate. Therefore, we are attempting to include SRTM data to improve the terrain representation.

Thank you for your time and support.

Sincerely,
Nagaraju Gaddam
Click to expand...
sf_urban_physics = 0, 0, 0, 3,

Looking at your namelist.input it looks like urban is turned off on 3 of the 4 domains. According to the user guide it needs to be turned on all of them.
 
Hi Nags,

Your namelist.input and namelist.wps both look fine to me.

Your rsl.error.0029 shows a few points with CFL violation. rsl.error.0027 then gives segmentation fault and the model dies. There is no other error message in all other rsl files. Based on this information, let's first try to overcome the CFL issue.

Please reduce your time step to 27, then rerun this case. if it still crashes, check all the rsl files and see whether there exists CFL violation.

Can you also clarify how you implement SRTM data in WPS? I am suspicious that this data may be problematic.
 
Ming Chen said:
Hi Nags,

Your namelist.input and namelist.wps both look fine to me.

Your rsl.error.0029 shows a few points with CFL violation. rsl.error.0027 then gives segmentation fault and the model dies. There is no other error message in all other rsl files. Based on this information, let's first try to overcome the CFL issue.

Please reduce your time step to 27, then rerun this case. if it still crashes, check all the rsl files and see whether there exists CFL violation.

Can you also clarify how you implement SRTM data in WPS? I am suspicious that this data may be problematic.
Click to expand...
Dear MingChen (@Ming Chen)

Thank you for your kind response.

However, I am still facing the same issue. Even after using the value 27, and later reducing the timestep to 12, the same error persists.

Below I describe my procedure step-by-step for implementing the Bologna-covered single-file SRTM 1'' topography in WRF/WPS:

Downloading SRTM data

I downloaded the Bologna-covered single-file SRTM 1'' dataset from the source website. The file extension was .hgt. After downloading, I converted the file into GeoTIFF (.tiff) format

Converting GeoTIFF to binary format
I then converted the GeoTIFF file into WPS binary format using the following command:

./convert_geotiff -w 4 -t 1200 -u "meters MSL" -d "Topography height" sourceSRTM.tiff

This process generated an index file along with several binary data files. I followed the methodology described in the documentation webiste

Modifying GEOGRID.TBL: I added the following lines in GEOGRID.TBL:

interp_option= topo_1s: average_gcell(4.0)+four_pt+average_4pt
rel_path= topo_1s: SRTM_topo_1s

Changes in namelist.wps
geog_data_res = 'srtm_1s+cglc_modis_lcz',
'srtm_1s+cglc_modis_lcz',
'srtm_1s+cglc_modis_lcz',
'srtm_1s+cglc_modis_lcz'

geog_data_path = '/scratch/atmos/nagaraju.gaddam/WRF/WPS_GEOG'

Verification: I also cross-checked the station observation elevations with the WRF topography variable (HGT_M).

Despite following these steps, the same error still occurs. I would greatly appreciate any suggestions or corrections you might recommend.

Sincerely,
Nagaraju Gaddam
 
Thank you for the detailed description of SRTM data processing. It looks fine.

Would you please upload figures of terrain height for all the domains with and without SRTM? Thanks.
 
Ming Chen said:
Thank you for the detailed description of SRTM data processing. It looks fine.

Would you please upload figures of terrain height for all the domains with and without SRTM? Thanks.
Click to expand...
Dear @Ming Chen,

Greetings of the day!,

Please find attached folder for youre kind review!

In this you will find the default (cglc_modis_lcz) and srtm 3om and 90m terrain height plots

Sincearly,
Nagaraju Gaddam
 

Attachments

It looks like the topography shown in srtm30m/d03_topography.png is different to that in srtm90m/d03_topography.png. While it is understandable that different resolution may yield different terrain details, the difference between the above two plots indicate that the terrain pattern is largely different.

Please confirm that the terrain height shown in srtm30m/d03_topography.png is correct.
 
You must log in or register to reply here.
Back
Top