Evaluation of WRF-Solar Cloud Forecast Using the NSRDB: Preprint

Jaemo Yang, Manajit Sengupta, Yu Xie, Aron Habte, Pedro A. Jimenez, Ju-Hye Kim

Research output: Contribution to conferencePaper


Cloud forecast is a crucial component in predicting solar irradiance from numerical weather prediction (NWP) models. Assessing cloud properties from the NWP models requires significant work due to the need for high-quality data, spatial analysis covering model extent, and detailed analysis of model performance for different types of clouds. This study presents an evaluation of the WRF-Solar cloud forecast using the National Solar Radiation Database (NSRDB). We propose an evaluation framework applied to a single model prediction as well as ensemble-based forecasts. Various cloud detection metrics are calculated when comparing with the satellite-derived dataset. The mismatched clouds from the WRF-Solar model are quantified using nine cloud types classified by cloud top height and cloud optical depth. The results based on the WRF-Solar forecasts covering the entire U.S. for the full year of 2018 shows mismatched cloud frequency in the range of 8% - 46% for thick and high-level (deep convective) to thin and low-level (cumulus) clouds.
Original languageAmerican English
Number of pages6
StatePublished - 2022
Event8th World Conference on Photovoltaic Energy Conversion - Milan, Italy
Duration: 26 Sep 202230 Sep 2022


Conference8th World Conference on Photovoltaic Energy Conversion
CityMilan, Italy

NREL Publication Number

  • NREL/CP-5D00-83925


  • cloud mask forecast
  • ensemble forecast
  • WRF-Solar EPS


Dive into the research topics of 'Evaluation of WRF-Solar Cloud Forecast Using the NSRDB: Preprint'. Together they form a unique fingerprint.

Cite this