Add check for consistency in land cover and soil

[GreyREvenson]

Purpose

This PR fixes issue #97.

Additions

Added land cover and soil consistency checks to initialize_from_file in RunModule.f90. These checks throw an error and stop the program if any cell is initialized with the following conditions:

• land use type != water and soil type = water
• land use type = water and soil type != water
• land use type != ice and soil type = ice
• land use type = ice and soil type != ice

Testing

I manually/hardcoded problematic values for vegtyp and isltyp to produce errors:

where (domaingrid%mask.eq.1) domaingrid%vegtyp = parametersgrid%ISWATER
where (domaingrid%mask.eq.1) domaingrid%isltyp = 1
returns => ERROR: one or more grid cells have a water land cover (vegtyp = 16) and a non-water soil type (isltyp != 14)

where (domaingrid%mask.eq.1) domaingrid%vegtyp = 1
where (domaingrid%mask.eq.1) domaingrid%isltyp = 14
returns => ERROR: one or more grid cells have a non-water land cover (vegtyp != 16) and a water soil type (isltyp = 14)

where (domaingrid%mask.eq.1) domaingrid%vegtyp = parametersgrid%ISICE
where (domaingrid%mask.eq.1) domaingrid%isltyp = 1
returns => ERROR: one or more grid cells have an ice land cover (vegtyp = 24) and a non-ice soil type (isltyp != 16)

where (domaingrid%mask.eq.1) domaingrid%vegtyp = 1
where (domaingrid%mask.eq.1) domaingrid%isltyp = 16
returns => ERROR: one or more grid cells have a non-ice land cover (vegtyp != 24) and an ice soil type (isltyp = 16)

Notes

• Implemented checks using where statements to avoid iterating over the grid
• Checks were implemented with hardcoded water and ice values for soil type. To avoid using hardcoded soil type values, we could require that the user provide ISWATER and ISICE values for the utilized soil classification scheme (as the user already must provide for the utilized land cover classification scheme).

[drakest123]

This PR has sanity checks to ensure grid cells that define water or ice do not also have a vegetation type. The sanity checks touched only the initialize_from_file subroutine in RunModule.f90.

• Compiled and executed: run/noah_owp_modular.exe
• test/noahowp_test.exe ../run/namelist.input > output_test
• No errors were detected.
• Sanity tests were coded.
• Checks were optimized by using the “where” keyword.
• The algorithm makes use of an array, err_grid(:,:). This may be a target for optimization if memory usage becomes an issue.
• It might be convenient for the user to report problematic grid cell locations rather than the fact that there is/are problematic grid cells.

[drakest123]

If possible, report problematic grid cell locations rather than the fact that there is/are problematic grid cells.

[GreyREvenson]

Thanks for the review @drakest123. I went ahead and add locational information to the error messages.

[GreyREvenson]

@SnowHydrology Do you think this PR's checks are too restrictive? Here, we're requiring grid cells that have ice land cover to have ice soils as well.

[SnowHydrology]

This looks good with one exception: using stop prevents the framework from returning useful error information. Instead, I'd recommend using return and then having the initialize routine return BMI_FAILURE. See examples here and here.

[GreyREvenson]

Note: I'm leaving the stop statements as-is with the assumption that the issue will be corrected by PR #108.

[GreyREvenson]

Note: it may be preferable to move these error-related objects to the ErrorCheckModule if PR #108 elects to make greater use of that module for error checking and messaging purposes.