ENH: Add missing ECAD indices

[bzah]

icclim proposes an ECAD catalog but only ~49 indices are available.
With generic indices from #189 , it becomes trivial to implement new indices, as long as they follow a generic indicator pattern.

For the ECAD catalog we must add indices of radiation, pressure, cloud cover, sunshine, humidity
We should implement at least every generic indicator compatible index.

Missing ecad indices after merge of #230 :

• GSL - Growing season length
  Cannot be generified easily ; exist in xclim : growing_season_length
  Added by [ENH] Bind xclim indicators to missing ECAD indices #235
• PP - Mean of daily sea level pressure
  Generic indicator -> Average
• SPI3 - 3-Month Standardized Precipitation Index ;
  Cannot be generified easily ; exist in xclim: standardized_precipitation_index
  Added by [ENH] Bind xclim indicators to missing ECAD indices #235
• SPI6 - 6-Month Standardized Precipitation Index
  Cannot be generified easily ; exist in xclim: standardized_precipitation_index
  Added by [ENH] Bind xclim indicators to missing ECAD indices #235
• SS - Sunshine duration
  Generic indicator -> Sum
• SSp - Sunshine duration fraction with respect to daylength
  Cannot be generified easily ; does not exist in xclim
• PET - Potential EvapoTranspiration
  Cannot be generified easily ; exist in xclim: potential_evapotranspiration
• RH - Mean of daily relative humidity
  Generic indicator -> Average
• CC - Mean of daily cloud cover
  New Unit (oktas/octas) ; Generic indicator -> Average
  The conversion from % to okta might not be straightforward to add to the pint registry (to do in xclim or/and cf-xarray)
• CC2 - Mostly sunny days (cloud cover ≤ 2 oktas)
  New Unit (oktas/octas) ; Generic indicator -> count_occurrences(threshold="<= 2 oktas")
• CC6 - Mostly cloudy days (cloud cover ≥ 6 oktas)
  New Unit (oktas/octas) ; Generic indicator -> count_occurrences(threshold=">= 6 oktas")
• UTCI - Mean of the Universal Thermal Climate Index
  Cannot be generified easily ; exist in xclim: universal_thermal_climate_index
• TCI - Mean of the Tourism Climatic Index
  Cannot be generified easily ; does not exist in xclim
• TCI60 - Days where the Tourism Climatic Index ≥ 60
  Need a new TCI (unitless ?) StandardVariable ; Generic indicator -> count_occurrences(threshold=">= 60") ;
• TCI80 - Days where the Tourism Climatic Index ≥ 80
  Need a new TCI (unitless ?) StandardVariable ; Generic indicator -> count_occurrences(threshold=">= 60") ;
• HI - Huglin Index (index specifically aimed at grap growth)
  Cannot be generified easily ; exist in xclim: huglin_index
• BEDD - Biologically Effective Degree Days (grape growth again)
  Could be generified ; exist in xclim: biologically_effective_degree_days
• FXX - max wind gus
• FG6BFT - days with averaged wind > 6 Bft
• FGCalm - days with averaged wind <= 2 m s-1
• FG - mean a longer period of daily mean wind
• DDNorth - days with northerly wind
• DDSouth - days with southerly wind
• DDEast - days with easterly wind
• DDWest - days with westerly wind
• [ ]

As we can see from the list above, it would be quite useful to (re)do a binding between icclim standard indices toward xclim indicators so that we would have access to most ecad indices without having the burden of gentrifying them.

---

Related issues:

• RR: ENH: Implement RR (precipitation sum) #185
• CDDCold (not ecad!) ENH: Add Cooling Degree Days (CDDCold{xx}) #184

PR:

• RR: Enh/rr index #214
• Wind indices: Enh/ecad wind indices #230
• Easy xclim binding: [ENH] Bind xclim indicators to missing ECAD indices #235

xclim indicators (link to latest!): https://xclim.readthedocs.io/en/latest/xclim.indicators.atmos.html

[bzah]

@pagecp I was looking into adding the cloudiness indices but I have issues with the oktas (aka octas) unit.

The ATBD says oktas should be computed with ROUND((cloud cover in percents/100)*8) :

Daily mean cloud cover CC Whenever synoptical cloud cover data is available at 00, 06, 12 and/or 18 UT, mean daily cloud cover is calculated as the average of the available values.
This value in percent is converted to octas by ROUND((cloud cover in percents/100)*8)

But I wonder if we should keep the rounding operator because:

• It's fairly difficult to implement a rounding for automatic conversion with Pint, our unit library.
• I don't see the value in reducing the accuracy of the measure by rounding it.

Also, when we look at CF cloud_area_fraction which is based on CMIP6 Total Cloud Cover Percentage unit is percents.
We could simply convert the cloudiness indices [ CC, CC2, CC6 ] that are based on oktas to be based on percents, as long as we display the proper unit.

What do you think ?

[bzah]

conclusion: we should really try to add oktas as a unit in our pint registry