/probabilistic-solar-forecasting

An archived dataset from the ECMWF Ensemble Prediction System for1 probabilistic solar forecasting

Primary LanguagePython

Probabilistic solar forecasting

An archived dataset from the ECMWF Ensemble Prediction System for probabilistic solar forecasting

Requirments：

The code is written in Python, and some packages should be installed before the scripts can be executed smoothly.

  • Package pandas is a fast, BSD-licensed library that provides high-performance data structures and data analysis tools.
  • Package numpy is the fundamental package for scientific computing.
  • Package pvlib provides functions for simulating the performance of PV systems.
  • Package scipy provides algorithms for scientific computing.
  • Package sklearn is the basic package for machine learning.
  • Package hydrostats is a library of functions for time series analysis.

Other Python packages that used for plotting the results include seaborn, matplotlib, and plotnine.

Data:

A total of 7 ENS_XXX.csv, Jacumba_ENS.csv, McClear_Jacumba.csv, ECMWF_HRES.csv, and 60947.csv files are provided. These files contain four years (2017--2020) of the ECMWF ENS forecast data for seven SURFRAD stations (xxx denotes the three-letter station abbreviations), four years (2017--2020) of the ECMWF ENS forecasting data for Jacumba solar plant, clear-sky irradiance for Jacumba solar plant, ECMWF HRES forecast data, and Jacumba solar power data whose Energy Information Administration (EIA) plant ID is 60947.

Code:

A total of three Python scripts, namely, post-processing.py, GradientBoosting.py, and ModelChain.py are provided for reproducibility. The file names are self-explanatory. In that, the post-processing.py provide the operational post-processing of NWP-based solar forecast at seven research-grade ground-based stations; GradientBoosting.py reproduces the irradiance-to-power conversion approach using gradient boosting; and ModelChain.py provides the irradiance-to-power conversion approach using model chain. To use these scripts, the user only needs to change the working directory.