Global horizontal irradiance forecast for Finland based on geostationary weather satellite data.

• We forecast irradiance with satellite imagery and cloud motion extrapolation. • Forecasts based on geostationary satellite imagery work well in the high latitudes. • We comparatively validated clear sky models and irradiance estimates in Finland. • Forecast performs well compared to smart persistence in changing cloud situations. Global horizontal irradiance (GHI) estimates and forecasts are necessary for the efficient use of a naturally fluctuating energy source like solar energy. However, few forecasting methods exist for high latitudes. In this study we present the development and validation of a satellite-based GHI forecast for southern Finland, called Solis-Heliosat. The forecast is formed by combining information from the clear sky (CS) model Pvlib Solis with consecutive geostationary weather satellite imagery, using the Heliosat method. Forecasts are produced up to 4 h with a 15-min temporal resolution. The CS model, the satellite-based all-sky estimate, and the GHI forecast have been compared and validated against other methods and in situ GHI measurements. An additional comparison was made for two datasets representing a changing cloud environment. The CS model estimates had an average rMBE (relative Mean Bias Error) of −6% to 1% and a rRMSE (relative Root Mean Square Error) of 6–10%. For the all-sky estimates the rMBE was −4% to −2%, and rRMSE 2–33%. With increasing forecast time the Solis-Heliosat rMBE descends to −9% and rRMSE reaches 50% at 4 h. Solis-Heliosat performs better than the persistence forecasts in most cases, particularly in a changing cloud environment. Our study indicates the use of satellite-based forecasts as a viable tool for forecasting GHI for the solar energy industry also in the high latitudes. In high latitudes geostationary satellite-based methods are at their limit; however, the information they can provide will enable efficient solar energy production. [ABSTRACT FROM AUTHOR]