Cloud-tracking methodology for intra-hour DNI forecasting

• We create forecasting models for DNI using sky images.
• We develop a new method to track cloud motion towards the sun’s position.
• We forecast intra-hour DNI based on cloud cover.
• We compare the proposed method against PIV-based cloud motion tracking algorithms.

We present a novel method for cloud tracking based on total sky imaging to forecast intra-hour Direct Normal Irradiance (DNI). We introduce both a sector method used to detect the direction of motion of potentially sun-blocking clouds, and a adjustable-ladder method, which is based on a size-adjustable set of grid elements that focus on sky regions of greatest potential for affecting ground DNI values. Images taken every 20 s by a sky imager are processed to generate 1-min based DNI forecasts for up to 20 min ahead horizons. Our results show that the deterministic DNI forecasting using sector-ladder methods performs better than the Particle Image Velocimetry (PIV) methods, showing improved performance in both detection of cloud motion and DNI forecast under different sky conditions: broken-sky, clear-sky and overcast.