Subseasonal to Seasonal (S2S) Prediction Algorithms using Hybrid Machine Learning Techniques

< S2S dataset.zip >

1.ECMWF observations/hindcast realizations

• hindcast-like-observations_2000-2019_biweekly_deterministic.zarr
• forecast-like-observations_2020_biweekly_deterministic.zarr
• ecmwf_hindcast-input_2000-2019_biweekly_deterministic.zarr
• ecmwf_forecast-input_2020_biweekly_deterministic.zarr
• hindcast-like-observations_2000-2019_biweekly_tercile-edges.nc

2. External variables

• "nino" folder -> nino12.long.anom.data, nino34.long.anom.data : El Niño data
• "Oscillation" folder
  • -> ersst.v5.pdo.dat.text : PDO (Pacific Decadal Oscillation)
  • -> norm.nao.monthly.b5001.current.ascii.table.txt : NAO (North Atlantic Oscillation)
  • -> qbo.dat : QBO (Quasi Biennial Oscillation)
• "great_lake" folder -> N_seaice_extent_daily_v3.0 : Great lakes ice cover
• observed-solar-cycle-indices.json : Sunspot cycles (two variables: original value and smoothed value)

3. Region.txt : Region and its bound

4. Biweekly historical statistics data

• biw_stat_w34 folder -> data (mean, standard deviation, median, skewness, kurtosis) for Week 3-4
• biw_stat_w56 folder -> data (mean, standard deviation, median, skewness, kurtosis) for Week 5-6

< ML_code.zip >

• ML codes for training, testing, and calculating RPSS based on Python3
• Check run_val.sh and run_2020.sh