Dataset and source code for "Explanation and optimizing multi-model blending algorithm using random variables theory"

this dataset contain: 

1. 2m temperature de-biased model forecast data on station location, ECMWF, NCEP, JP and CMA
2. 2m temperature observaton data, obs_t2m
3. 24H QPF model forecast data on station location, ECMWF, NCEP, CMA-GFS, in raw_data_r24.zip
4. 24H precipitation data, in raw_data_r24.zip
5. source code (in python)

how to use it: 

1. prepare data and python environment
    1.1 if you want to run [Station_FCST_MMWB.py] or [Station_FCST_MMWB_r24.py] , please download the station forecast and observation data
    1.2 neet meteva package to read/write micaps-3 format data: https://github.com/nmcdev/meteva
    1.3 need cartopy to draw picture FigS01. 

2. try the 2m temperature blending methods <optional>
    2.1 unzip the [CMA.zip, ECMWF.zip, jp.zip, NCEP.zip, obs_t2m.zip] file into ./raw_data/
    2.2 run the Station_FCST_MMWB.py in python environment 

3. try the 24h QPF multi blending methods <optional>
    3.1 unzip the [raw_data_r24.zip] file into ./raw_data_r24/
    3.2 run the Station_FCST_MMWB_r24.py in python environment

4. draw figures
    4.1 run Fig01.py in python environment 
    4.2 run Fig02.py in python environment 
    4.3 run Fig03.py in python environment 
    4.4 run FigA01.py in python environment 
    4.5 run FigS01.py in python environment