Project Name: Probabilistic Cloud Optimized Day-Ahead Forecasting System Based on WRF-Solar
Funding Opportunity: Solar Forecasting 2
SETO Subprogram: Systems Integration
Location: Golden, CO
SETO Award Amount: $1,720,806
Awardee Cost Share:  $212,482
Principal Investigator: Dr. Manajit Sengupta

This project will develop an optimized ensemble-based solar irradiance forecasting system that will demonstrably improve current state-of-the-art solar forecasts from the Weather Research and Forecasting-Solar (WRF-Solar) model and provide probabilistic forecasts that improve grid operations. This probabilistic solar forecasting system will significantly enhance both the intra-day and the day-ahead solar forecasting capability for grid operations. To enable the widespread use of this operational solar forecasting model, this system will be publicly available as open-source software.

APPROACH

This project will develop a probabilistic WRF-Solar system that reduces computational burden by using an adjoint-based method, a mathematical technique, to identify the variables that significantly influence the formation and dissipation of clouds and solar radiation. After identifying the parameters that influence these solar forecast errors, the parameters will then be used to develop an optimized ensemble forecasting system which consists of multiple models and will produce multiple realizations of a single forecast. Using satellite-based solar radiation measurements, the system will be calibrated to ensure unbiased forecast trajectories to provide accurate estimates of forecast uncertainties under a wide range of meteorological regimes.

INNOVATIONS

The ensemble-based forecast will optimize calculations to provide a cost-effective, probabilistic forecasting solution. As the forecasting system will be calibrated using measurements of irradiance and other meteorological variables, it will reduce current inaccuracies and also produce forecasts in a form that can be used to improve grid operations. This approach to creating the forecasting system and the calibration method will help to address the need for improved accuracy in the forecasting of cloud properties, their formation, and their dissipation.