climate.yaml

---
plugins:
  source:
    - module: intake_xarray, intake_excel
sources:
  Global_Carbon_Budget_2020:
    description: GCP 2020
    metadata:
      url: https://www.globalcarbonproject.org/carbonbudget/
      doi:
        - http://doi.org/10.18160/gcp-2020
        - http://doi.org/10.5194/essd-12-3269-2020
      plots:
        over_time:
          kind: line
    driver: excel
    args:
      urlpath: simplecache::https://data.icos-cp.eu/licence_accept?ids=%5B%226QlPjfn_7uuJtAeuGGFXuPwz%22%5D
      excel_kwargs:
        header: 20
        sheet_name: 'Global Carbon Budget'
        index_col: Year
        skipfooter: 4

  Global_Carbon_Budget_2021:
    description: GCP 2021
    metadata:
      url: https://www.globalcarbonproject.org/carbonbudget/
      doi:
        - https://doi.org/10.5194/essd-2021-386
        - https://doi.org/10.18160/gcp-2021
      plots:
        over_time:
          kind: line
    driver: excel
    args:
      urlpath: simplecache::https://data.icos-cp.eu/licence_accept?ids=%5B%220ST81nXCND5VfAQdOCSJDveT%22%5D
      excel_kwargs:
        header: 20
        sheet_name: 'Global Carbon Budget'
        index_col: Year
        skipfooter: 0

  NOAA_correlation:
    description: climate indices from psl.noaa.gov/data/correlation
    metadata:
      url: https://www.psl.noaa.gov/data/correlation/
    driver: csv
    parameters:
      index:
        description: name of the index
        default: nina34
        type: str
        allowed:
          - pna
          - wp
          - ea
          - nao
          - jonesnao
          - soi
          - nina3
          - censo
          - oni
          - nina1
          - nina4
          - nina34
          - np
          - hurr
          - pacwarm
          - swmonsoon
    args:
      urlpath: simplecache::https://psl.noaa.gov/data/correlation/{{index}}.data
      csv_kwargs:
        sep: '  '
        skiprows: 3
        skipfooter: 4
        dtype:
          year: object
        names:
          - year
          - Jan
          - Feb
          - Mar
          - Apr
          - May
          - Jun
          - Jul
          - Aug
          - Sep
          - Oct
          - Nov
          - Dec
        # index_col: 'year' not allowed by dd.read_csv but pd.read_csv
        na_values:
          - -999.90
          - '-999.90'
          - '-99.90'
          - -99.90
          - -99.0
          - '-99.0'
        engine: python

  NOAA_correlation_xr:
    description: NOAA_correlation as xr.DataArray
    driver: intake.source.derived.GenericTransform
    args:
      targets:
        - NOAA_correlation
      transform: "remote_climate_data.utils.convert.monthly_csv_to_DataArray"
      transform_kwargs: {}

  NOAA_SST_anomaly_timeseries:
    description: Sea Surface Temperature Anomaly Indices derived from ERSST
    metadata:
      url: https://www.psl.noaa.gov/forecasts/sstlim/timeseries/
    driver: csv
    parameters:
      index:
        description: name of the index
        type: str
        default: Nina4
        allowed:
          - CAR
          - EA
          - HAW
          - IND
          - Nina4
          - Nina34
          - NTA
          - STA
    args:
      urlpath: ftp://ftp.cdc.noaa.gov/Datasets/Timeseries/{{index}}_ersst.data
      csv_kwargs:
        sep: '   '
        skiprows: 2
        skipfooter: 8
        names:
          - Jan
          - Feb
          - Mar
          - Apr
          - May
          - Jun
          - Jul
          - Aug
          - Sep
          - Oct
          - Nov
          - Dec
        # index_col: 'year' not allowed by dd.read_csv but pd.read_csv
        engine: python