Shortest Term Solar Energy Forecast

Abstract

Due to the strong growth in photovoltaics (PVs), forecasting is gaining importance. At present, most forecasts are based on numerical weather prediction (NWP) models without cloud assimilation. These models lack the ability to forecast the exact position of clouds, which is needed for single-point forecasts. Nowcasting methods, based on current information about cloud positions, provide an opportunity to enhance solar forecasts. Here, we review two of four existing methods—sky camera and satellite image-based methods. Sky cameras are still at an early stage of development and much work is currently being performed. Uncertainties are in the range of 25 % for 30 s data and up to 10 min ahead. Satellite image-based methods have been in use for several years. It is possible to show that a simplified approach using NWP-based cloud vectors leads to similar accuracy as using cloud vectors based on multiple images. Satellite-based methods enhance forecasts by 30–40 % compared to NWP forecasts for forecast horizons of 15 min to 4 h. We also briefly discuss simple post-processing and aggregation methods.