ERROR initInterp set_mindist: Cannot find any input points matching output point:

[liliyao]

Hi,

I am running two simulation periods with offline CLM from ctsm5.1.dev118. The simulation area is within the continental United States (CONUS). The component sets are: HIST_DATM%GSWP3v1_CLM50%BGC-CROP_SICE_SOCN_SROF_SGLC_SWAV. The resolution used is 0.125nldas2.

Period 1:

1. time period: 1980-2004
2. Restart file utilized is from the end of the final spin up.
3. Surface dataset and land use timeseries file utilized are generated using the ctsm5.2.mksurfdata (to take advantage of the dynamic urban capability in this version) with default raw data.
4. Climate forcing is my own forcing, i.e., TGW-WRF from 1980-2004
5. This run was completed successfully

Period 2:

1. time period: 2005-2019
2. Restart file utilized is from the end of the Period1.
3. The surface dataset and land use timeseries are incorporated using my own Land Use and Land Cover (LULC) area fraction values (I generate the surface dataset and land use timeseries using ctsm5.2.mksurfdata, and subsequently modify the outputs with my own Land Use and Land Cover (LULC) area fraction values)
4. Climate forcing is TGW-WRF from 2005-2019.
5. This run is failing.

I am getting the following error message:

215: ERROR initInterp set_mindist: Cannot find any input points matching output point:
215: subgrid level, index = pft 40
215: lat, lon = 0.57268616081063939 , 4.2160609743488022
215: ltype: 6
215: ctype: 6
215: ptype: 0
215:
215: Consider rerunning with the following in user_nl_clm:
215: init_interp_fill_missing_with_natveg = .true.
215: However, note that this will fill all missing types in the output
215: with the closest natural veg column in the input
215: (using bare soil for patch-level variables).
215: So, you should consider whether that is what you want.
215: ENDRUN:

However, the mentioned lat/lon values are suspicious, as this point is not within the CONUS. I am uncertain about how the model got these lat/lon values, and I'm unsure how to resolve this error. I would greatly appreciate any insights you might have on this matter. I have also attached the log files from the period2 simulation .

Thank you very much!

 

[oleson]

Thanks for all of that information. In your lnd log file I see:

Surface Grid Characteristics
longitude points = 464
latitude points = 224
total number of land gridcells = 319

I would expect the number of land gridcells to be much larger than 319 for this grid. So I wonder if there is something wrong with the grid descriptions in either the mesh file that is used to generate the land mask/land fraction, or in the surface dataset.

 

[liliyao]

Hi Dr. Oleson, I applied a mask mesh to the 464x224 grid, resulting in only 319 active grids for both the period 1 and period 2 simulations.

 

[oleson]

Sounds like you are running a sparse grid intentionally then?
The error seems to be associated with wetland (ltype=ctype=6). Maybe you have a wetland point in period 1 but not in period 2? As suggested, you could try setting init_interp_fill_missing_with_natveg = .true.. Although I'm not sure why the lat/lon is outside your domain.

 

[liliyao]

Hi Dr. Oleson, Yes, I am intentionally using a sparse grid. I compared the two surface datasets for the two periods. It's not the case that a wetland point is present in period 1 but not in period 2. Period 2 ran successfully with the parameter init_interp_fill_missing_with_natveg set to .true.. However, I am still interested in figuring out why I encountered the error. Do you have any suggestions for further investigation into the error?

Thank you very much!

 

[slevis]

If I followed correctly what happened, then you encountered the error because you used a restart file from period 1 to start period 2, but you used a different fsurdat file in period 2 than you used in period 1.

 

[liliyao]

Hi Sam, your understanding is correct.