Clarity needed on radiation inputs; errors with Kessler

[ewquon]

Posting here on behalf of @Robson-Passos who originally raised an issue on the erftools issues page (erf-model/erftools#40). He is running a real-data case driven by wrfinput/bdy.

1. I do not fully understand the appropriate value to use for the surface temperature rad_t_sfc.
2. With Kessler MP, I get the warning:

Warning: LSM model is being used, but no mapping is defined to fill its variables from WRFinput!

And the error:

0::Assertion 'dtt > -90' failed, file "../../Source/Utils/ERF_MicrophysicsUtils.H", line 38 !!!

When switching to erf.moisture_model = "Kessler" I instead get:

0::Assertion 'i < (this->std::vector<T, Allocator>::size())' failed, file "../../Submodules/AMReX/Src/Base/AMReX_Vector.H", line 36, Msg: Out of bound error, index: 29 size: 29 !!!

[ewquon]

@Robson-Passos could you clarify which moisture model you used that resulted in the first error, and also share the rest of the backtraces

[Robson-Passos]

Hello, @ewquon.

I don’t remember which option caused the first error, and I wasn’t able to reproduce it again using other options. However, the error:

0::Assertion `i < (this->std::vector<T, Allocator>::size())' failed, file "../../Submodules/AMReX/Src/Base/AMReX_Vector.H", line 36, Msg: Out of bound error, index: 21901 size: 2 !!!

continues to occur, regardless of the moisture_model option. So it may not be related to that.

I am attaching my input file and the backtrace.

inputs.txt

Backtrace.0.txt

The WRF input files are available at the link: https://drive.google.com/drive/folders/1sf-7RYVLe27OP4Zcumi8Y0TR46g1mU72

Best regards,
Robson

[AMLattanzi]

I can take a look at this case.

In response to the first post:

1. I do not fully understand the appropriate value to use for the surface temperature rad_t_sfc.

A. The rad_t_sfc is a constant value corresponding to the surface temperature seen by the radiation module. If you are running with an LSM, which it appears you are, this value is only used for the first time step because RRTMGP is called pre-step and the LSM is called post-step. So with an LSM, as long as this value isn't erroneous, it only affects the heating rates for the first step.

2. With Kessler MP, I get the warning:

Warning: LSM model is being used, but no mapping is defined to fill its variables from WRFinput!

And the error:

0::Assertion 'dtt > -90' failed, file "../../Source/Utils/ERF_MicrophysicsUtils.H", line 38 !!!

When switching to erf.moisture_model = "Kessler" I instead get:

0::Assertion 'i < (this->std::vector<T, Allocator>::size())' failed, file "../../Submodules/AMReX/Src/Base/AMReX_Vector.H", line 36, Msg: Out of bound error, index: 29 size: 29 !!!

A. The LSM warning can be ignored with the NOAH-MP LSM. That warning was put in place for the simple land surface model that is native to ERF and must parse data from the wrfinput file. We should add another logic condition so that doesn't get printed to the log file (on that now). The assertion is saying that the temperature is out of the valid range for the saturation computation. And finally, the vector error is usually attributed to something not allocated over all the levels.

I'll grab your files and try to replicate. Any PRs will be linked here.