Jupyter Notebook Examples

To see the jupyter notebooks in action visit here:

Link to Jupyter Notebook Examples: Jupyter Examples.

Start up

Follow these simple steps to quickly get started in a jupyter notebook:

  1. ** READ THIS FIRST:** :

NOTE: You must first have your MJOcast environment activated.

This notebook shows the user how to drive the MJOcast on a single forecast file, and create desired output file.

In this example, the output file has lead-time dependent bias removed, based on the lead time dependent climo file.

This forecast file climo file was made with the Preprocessing_Tools., and MUST be completed prior to making this work on your files

  1. Install Packages: :
import MJOcast.utils.ProcessForecasts as ProFo
import MJOcast.utils.ProcessOBS as ProObs
import MJOcast.utils.WHtools as WHtools
import importlib
import xarray as xr
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib import image as mpimg

  1. Configuration: Customize your forecasting experience by configuring the toolbox:

    2.1. Manual Configuration: Edit the YAML configuration file directly to define your forecast parameters, data paths, and other settings. Jupyter With-YAML Examples.

    2.2. YAML Generator Tool: Alternatively, use the provided YAML generator tool in the preprocessor Jupyter Notebook for a user-friendly configuration process. Jupyter NO-YAML Examples.

# YAML file specifying output configuration
output_yaml = 'output.yaml'

# User information
user_ = 'wchapman'

# Base directory
base_dir = '/glade/work/wchapman/MJOcast/MJOcast'

# Flag to indicate whether to use ERA5 data
use_era5 = True

# Alternative name for user-defined observations (only applicable if user_era5=False)
usr_named_obs = 'alternative_name_of_obs.nc'

# Locations for observation and forecast data
obs_data_loc = './Observations/'

# Locations for observation and forecast data
forecast_data_loc = '/glade/campaign/cgd/amp/wchapman/MJO_S2S_CESM2/p1/examples'

# String pattern for MJO forecast data files
forecast_data_name_str = 'S2Shindcast_cesm2cam6vs_MJOvars_*.nc'

# Variable names for forecast data (and name of ensemble dimension)
forecast_olr_name = 'rlut'
forecast_u200_name = 'ua_200'
forecast_u850_name = 'ua_850'
forecast_ensemble_dimension_name = 'ensemble'

# Location for diagnostic plots
output_plot_loc = './output_plots/'

# Location for output files
output_files_loc = '/glade/campaign/cgd/amp/wchapman/MJO_S2S_CESM2/RMM_forecast/ERA5/'

# Naming convention for MJO forecasts
output_files_string = '/MJO_Forecast_Init'

# Flag to indicate whether to use forecast climatology (turn False if you don't want to remove lead time dependent bias).
use_forecast_climo = True

# Flag to indicate whether to use observed climatology
# (always false if "use_forecast_climo=true")
use_observed_climo = False

# Flag to regenerate climatology
regenerate_climo = False

# Flag to use Dask for climatology on a single node/machine
use_dask_for_climo = True

# Flag to regenerate climatology (single node/machine)
regenerate_climo = False

#create your driver yaml file:
WHtools.Create_Driver_Yaml('output.yaml', user_, base_dir, use_era5, usr_named_obs, obs_data_loc, forecast_data_loc, forecast_data_name_str,
           forecast_olr_name, forecast_u200_name, forecast_u850_name, forecast_ensemble_dimension_name,
           output_plot_loc, output_files_loc, output_files_string, use_forecast_climo, use_observed_climo,
           regenerate_climo, use_dask_for_climo)
  1. Run your example and create your processed file: : Begin forecasting:
#initialize the ObsProcessor:
MJO_obs = ProObs.MJOobsProcessor(output_yaml)

#Make the Observed MJO file:
OBS_DS, eof_list, pcs, MJO_fobs, eof_dict = MJO_obs.make_observed_MJO()

#Plot whatever day you would like for the obs phase space:
MJO_obs.plot_phase_space('2001-01-01',60)

#Now initialize the
MJO_for = ProFo.MJOforecaster(output_yaml,MJO_obs.eof_dict,MJO_obs.MJO_fobs)

#Now create the forecasts
DS_CESM_for,OLR_cesm_anom_filterd,U200_cesm_anom_filterd,U850_cesm_anom_filterd = MJO_for.create_forecasts(num_files=1)

#plot the phase space diagram for your forecast:
MJO_for.plot_phase_space(12,15)
Load the YAML configuration file:

[output]:
Number of forecast files to process: 1
expanding coords to include ensemble
ensemble dimension length: 11
there are 46 forecast lead days in these files
Initial look at forecast files passes the first test
 ----- Taking the EOF -----
...done making observed EOFS, check ./output_plots/*.png for verification metrics...
...attaching the BOM index for verification...
...saved OLR obs file...
Number of forecast files to process: 1
expanding coords to include ensemble
ensemble dimension length: 11
there are 46 forecast lead days in these files
Initial look at forecast files passes the first test
Using the forecast dependent climatology. Make sure you have generated it using ./Preprocessing_Scripts/*.ipynb.
reading forecasts file: /glade/campaign/cgd/amp/wchapman/MJO_S2S_CESM2/p1/examples/S2Shindcast_cesm2cam6vs_MJOvars_25jan2018.nc
expanding coords to include ensemble
ensemble dimension length: 11
there are 46 forecast lead days in these files
---- doing anomaly ----
im using the LT dp climo
---- done computing the anomaly----
--- filter out 120 days ---
--- done filtering out 120 days ---
--- project the EOFS ---
saved:  /glade/campaign/cgd/amp/wchapman/MJO_S2S_CESM2/RMM_forecast/ERA5//MJO_Forecast_Init_25jan2018.nc
--- done projecting the EOFS ---

A created dataset is then saved here:

DSdone = xr.open_dataset('/glade/campaign/cgd/amp/wchapman/MJO_S2S_CESM2/RMM_forecast/ERA5//MJO_Forecast_Init_25jan2018.nc')

Output figures are saved here:

"./output_plots/observed_eofs.png"

Please visit the provided links at the top of this page to see the running jupyter notebooks!

Follow these steps to make the most of MJOforecaster in your projects. Happy forecasting!