-
prepare test data. 1.1 download data from url: https://www.scidb.cn/en/anonymous/TnpRYkVq 1.2 unzip the test data.
-
calculate the de-biased mulit-model forecasts 2.1 set path variables (in function "main_proc") in Station_FCST_BC.py 2.2 run Station_FCST_BC.py to de-bias the raw forecast.
-
do the traditional weighted multi-model blending and improved weighted multi-model blending 3.1 set path variables (in function "prepare_corrcoef" and function "main_proc") in Station_FCST_MMWB.py 3.2 run Station_FCST_MMWB.py to blend multi-model de-biased forecasts
-
test the blended forecast's expection and STD estimated equation (Fig. 1) run N01.py
-
test the MAE estimate equation (Fig. 2) run N02.py
-
show 72H blended forecast's MAE variation curve (Fig. 3) run N03.py
wangy1986 / cwmba Goto Github PK
View Code? Open in Web Editor NEWcorrelation coefficient improved weighted multi-model blending algorithm