The purpose of the TWC_VALIDATION code is to provide an objective evaluation of the sameness of the output from different MPAS simulations. With this validation test, we expect the results to be nearly identical. This scheme only works with MPAS global simulations.
- Comparing the output from the same code run with different compilers
- Comparing the output from the same code run with different optimzation levels of the same compiler
- Comparing the output from the same code run with different versions of the same compiler
- Comparing the output from the same code run with different hardware
- Comparing the output from the same code run with different floating point precision
- Comparing the output from different physics schemes
- Comparing the output from different dynamics options
- Comparing the output from different meshes
- Comparing the output from different horizontal or vertical resolution
- Fill in the run-time configuration file,
input.json
. There is aREADME_JSON.md
file with specific details. - The scripting system uses python 3. The following python modules are required:
- numpy
- netcdf4-python
- scipy
- matplotlib
- To try out this system on the NCAR cheyenne environment, there is an
EXAMPLES
directory that has several JSON files that are set up to work specifically on cheyenne. The following is sufficient to run this validation script on cheyenne.
> ncar_pylib
- Run the executable driver script
./validate.csh
, no arguments are required as everything comes in from the JSON file. - This script has three parts:
- Read the NETCDF MPAS model data, select the correct variables and geophysical locations, compute differences, output temporary text files for use by the ANOVA program
- Run a three-factor ANOVA test on the following variables:
u
(horizontal momentum),theta
(potential temperature), andqv
(water vapor mixing ratio) - Interpret the ANOVA output to produce a probability.
- This test takes approximately 60 seconds to conduct (nCells = 163842). Most of the time is spent finding which grid cells are within the correct latitude / longitude box.
For ease of use, the results are broken into three pieces:
- Very confident that the results are the same. For this variable, this is considered a POSITIVE indicator that this field is the same. To assume that response, we need all three variables to display this symbol:
▕▔▔▔╲
▏ ▕
▏ ▕
▏ ▕
▏ ▕▂▂▂▂
▂▂▂▂▂▂╱┈▕ ▏
▉▉▉▉▉┈┈┈▕▂▂▂▂▂▂▏
▉▉▉▉▉┈┈┈▕ ▏
▉▉▉▉▉┈┈┈▕▂▂▂▂▂▂▏
▉▉▉▉▉┈┈┈▕ ▏
▉▉▉▉▉┈┈┈▕▂▂▂▂▂▂▏
▉▉▉▉▉╲┈┈▕ ▏
▔▔▔▔▔▔╲▂▕▂▂▂▂▂▂▏
- Very confident that the results are different. For this variable, this is considered a NEGATIVE indiactor that this field is the same. To assume this response, we need any single variable to display to display this symbol:
▉▉▉▉▉▉▉▉▉▉▉
▉▉▉ ▉▉
▉▉ ▉▉
▉▉ ▉▉ ▉▉ ▉
▉▉ ▉▉▉ ▉▉▉ ▉▉
▉▉ ▉▉
▉▉ ▉▉
▉▉ ▉▉▉▉▉▉▉▉▉ ▉▉
▉▉ ▉▉ ▉▉ ▉▉
▉▉ ▉ ▉ ▉▉
▉▉▉ ▉▉
▉▉▉ ▉▉
▉▉▉ ▉▉
▉▉ ▉▉▉
▉▉▉▉▉▉▉▉▉▉▉
- Gray area, just not sure. This is an indeterminate region. This is the state when there are no NEGATIVE indicators, but there are also not all POSITIVE indicators.
▉▉▉▉▉▉▉▉
▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉
▉▉▉▉▉▉▉▉ ▉▉▉▉▉▉▉▉
▉▉▉▉▉ ▉▉▉▉▉▉
▉▉▉▉ ▉▉▉▉▉
▉▉▉▉▉
▉▉▉▉▉▉▉
▉▉▉▉▉▉▉
▉▉▉▉▉▉▉
▉▉▉▉▉
▉▉▉▉▉
▉▉▉▉▉
▉▉▉▉▉
▉▉▉▉▉
▉▉▉▉
▉▉▉▉