Development of an Integrated Framework for Solar Power Forecasting

Growing concerns over rising fossil fuel costs, energy security, resource depletion, climate change and sustainable economic development have stimulated the Renewable Energy (RE) sector. Solar energy is on the rise in the United States, but its variable nature and relative unpredictability remains a major obstacle for integration onto the national energy grid. The variability of the solar irradiance depends on several factors such as concentrations of gaseous species and atmospheric aerosols, solar geometry and cloud cover. The latter two are the most significant contributing factors to the solar variability and while the solar position can be easily determined throughout the day, solar variability due to clouds is considered primarily stochastic. Variable cloud amounts result in dominant sub-minute, hourly, multi-hour and daily variations of the solar irradiance, which result in intermittency and ramping of solar power generation. Our solar forecasting methodology uses near real-time cloud masks derived from a Cloud Mask Generator (CMG) model developed in house, which combines multi-satellite, multi-spectral remotely sensed imagery, numerical assimilations and conventional environmental observations to perform cloud cover analysis in both historical and real-time modes. The derived cloud cover is used to initialize our Cloud Propagation Forecast (CPF) model that is using a trailing window to generate training data for integrating morphological cloud changes, trajectory-based cloud propagation, and conditional cloud climatology into a single fused cloud forecast. For longer forecast times (beyond 6 hours) the NOAA NCEP's Short Range Ensemble Forecast (SREF) models cloud cover forecasts are seamlessly blended into the CPF. The CPF cloud cover forecast is then used by our broadband irradiance model to provide estimates of direct and indirect solar radiative energy at the surface. The simulated irradiance fields are validated against surface pyranometer observations and various user-defined solar metrics are derived to facilitate solar system operators' decision making with respect to short-term and day-ahead systems operations.