# ESMValTool
# recipe_preprocessor_test.yml
---
documentation:
  description: |
    Various example preprocessors.

  authors:
    - righi_mattia
    - andela_bouwe
    - schlund_manuel

  title: Test recipe for ESMValCore preprocessors.

  maintainer:
    - righi_mattia

  projects:
    - c3s-magic

datasets:
  - {dataset: MPI-ESM-LR, project: CMIP5, exp: historical, ensemble: r1i1p1}
  - {dataset: bcc-csm1-1, project: CMIP5, exp: historical, ensemble: r1i1p1, tag: group1}
  - {dataset: GFDL-ESM2G, project: CMIP5, exp: historical, ensemble: r1i1p1}

preprocessors:

  preprocessor_1:
    extract_levels:
      levels: bcc-csm1-1
      # This also works, but allows specifying the pressure coordinate name
      # levels: {dataset: bcc-csm1-1, coordinate: air_pressure}
      scheme: linear
    mask_landsea:
      mask_out: land  # if land (sea), will mask land (sea and oceans) out
    regrid:
      target_grid: bcc-csm1-1
      scheme: linear
    mask_fillvalues:
      threshold_fraction: 0.90
    multi_model_statistics:
      span: overlap
      statistics: [mean, median]
      exclude: [GFDL-ESM2G]

  preprocessor_2:
    custom_order: true
    regrid:
      target_grid: 1x1
      scheme: nearest
    extract_levels:
      levels: {cmor_table: CMIP6, coordinate: plev8}
      scheme: nearest
    mask_fillvalues: false
    multi_model_statistics:
      span: full
      statistics: [median]

  preprocessor_3:
    extract_levels:
      levels: [2000, 7000]
      scheme: nearest
    regrid:
      target_grid: 2x2
      lon_offset: false
      lat_offset: false
      scheme: nearest
    mask_fillvalues: false
    multi_model_statistics: false

  preprocessor_4:
    extract_levels: false
    regrid: false
    mask_fillvalues: false
    multi_model_statistics: false


  # Simple ensemble mean, median, and percentiles for multiple models
  preprocessor_5:
    ensemble_statistics:
      statistics:
        - mean
        - median
        - operator: percentile
          percent: 5
        - operator: percentile
          percent: 95
      exclude: [GFDL-ESM2G]

  # Calculate ensemble means, then multi-model mean
  preprocessor_6:
    regrid:
      target_grid: 3x3
      scheme: linear
    ensemble_statistics:
      statistics: [mean]
      exclude: [GFDL-ESM2G]
    multi_model_statistics:
      span: overlap
      statistics: [mean]
      keep_input_datasets: false
      exclude: [GFDL-ESM2G]

  # Multi-model mean with more complex groupby
  preprocessor_7:
    regrid:
      target_grid: 3x3
      scheme: linear
    multi_model_statistics:
      span: overlap
      statistics: [mean]
      groupby: [dataset, tag]
      exclude: [bcc-csm1-1, MPI-ESM-LR, GFDL-ESM2G]

  # First calculate ensemble mean and median, then multi-model min and max

  preprocessor_8:
    regrid:
      target_grid: 3x3
      scheme: linear
    ensemble_statistics:
      statistics: [mean, median]
      exclude: [GFDL-ESM2G]
    multi_model_statistics:
      span: overlap
      statistics: [max, min]
      groupby: [ensemble_statistics]
      keep_input_datasets: false
      exclude: [GFDL-ESM2G]

diagnostics:

  diagnostic_1:
    description: Test extract_levels, mask_landsea, regrid, mask_fillvalues, multi_model_statistics.
    variables:
      ta:
        preprocessor: preprocessor_1
        mip: Amon
        start_year: 2000
        end_year: 2002
      orog:
        mip: fx
      sftlf:
        mip: fx
    scripts: null

  diagnostic_2:
    description: Test custom_order, regrid, levels, multi_model_statistics.
    variables:
      ta:
        preprocessor: preprocessor_2
        mip: Amon
        start_year: 2000
        end_year: 2002
    scripts: null

  diagnostic_3_and_4:
    description: Test extract_levels, regrid.
    variables:
      ta:
        preprocessor: preprocessor_3
        mip: Amon
        start_year: 2000
        end_year: 2002
      pr:
        preprocessor: preprocessor_4
        mip: Amon
        start_year: 2000
        end_year: 2002
    scripts: null

  diagnostic_5:
    description: Test ensemble_statistics.
    variables:
      tas:
        preprocessor: preprocessor_5
        mip: Amon
        start_year: 2000
        end_year: 2000
        additional_datasets:
          - {dataset: MPI-ESM-LR, project: CMIP5, exp: historical, ensemble: 'r(2:3)i1p1'}
          - {dataset: bcc-csm1-1, project: CMIP5, exp: historical, ensemble: r2i1p1}
    scripts: null

  diagnostic_6:
    description: Test regrid, ensemble_statistics, multi_model_statistics.
    variables:
      tas:
        preprocessor: preprocessor_6
        mip: Amon
        start_year: 2000
        end_year: 2000
        additional_datasets:
          - {dataset: MPI-ESM-LR, project: CMIP5, exp: historical, ensemble: 'r(2:3)i1p1'}
          - {dataset: bcc-csm1-1, project: CMIP5, exp: historical, ensemble: r2i1p1}
    scripts: null

  diagnostic_7:
    description: Test regrid, multi_model_statistics.
    variables:
      tas:
        preprocessor: preprocessor_7
        mip: Amon
        start_year: 2000
        end_year: 2000
        additional_datasets:
          - {dataset: HadGEM2-ES, project: CMIP5, exp: historical, ensemble: r1i1p1, tag: abc}
          - {dataset: HadGEM2-ES, project: CMIP5, exp: historical, ensemble: r2i1p1, tag: abc}
          - {dataset: HadGEM2-ES, project: CMIP5, exp: historical, ensemble: r3i1p1}
          - {dataset: HadGEM2-ES, project: CMIP5, exp: historical, ensemble: r4i1p1}
          - {dataset: CanESM2,    project: CMIP5, exp: historical, ensemble: r1i1p1, tag: group1}
          - {dataset: CanESM2,    project: CMIP5, exp: historical, ensemble: r2i1p1, tag: group1}
          - {dataset: CanESM2,    project: CMIP5, exp: historical, ensemble: r3i1p1, tag: group2}
          - {dataset: CanESM2,    project: CMIP5, exp: historical, ensemble: r4i1p1, tag: group2}
    scripts: null

  diagnostic_8:
    description: Test regrid, ensemble_statistics, multi_model_statistics.
    variables:
      tas:
        preprocessor: preprocessor_8
        mip: Amon
        start_year: 2000
        end_year: 2000
        additional_datasets:
          - {dataset: MPI-ESM-LR, project: CMIP5, exp: historical, ensemble: 'r(2:3)i1p1'}
          - {dataset: bcc-csm1-1, project: CMIP5, exp: historical, ensemble: r2i1p1}
    scripts: null