Numerical weather prediction (NWP) model has been used for a solar irradiance and/or a PV power generation forecast as basic datasets. However, forecast errors are certainly included in the NWP models. Previous study investigated the relationship between cloud types and solar irradiance forecasting (e.g., Ohtake et al (2015)). Stratus clouds, cumulus and cirruls clouds are known as an element causing large forecast errors of solar irradiance at the surface. In some cases, large forecast errors of solar power forecasting (called, outlier events) are often caused even for a regional area forecasting.
Toward a high performance computing in a future status, it would be useful that whether more high resolution NWP model (sub-km scale) could provide better products or not for forecasting solar power and/or PV power generation for pin-point and/or regional area in a renewable energy field. Topography included in the models has a strong impact on cloud field forecasts, therefore, solar power forecasts would be received from the model topography.
However, a confirmation of model performance of NWPs with different horizontal resolution models is necessary as a basic validation for solar power forecasting. In this study, impact of horizontal resolutions for solar power forecasting are investigated.
Here, short term solar irradiance forecasts (up to 9 hours forecasts) are calculated using the different four horizontal grid spacing (or 5km, 2km, 1km and 500m) of NWPs in summer cases. Except for horizontal resolutions, other model parameters has not been changed. Product outputs were provided at every 2.5 minute in order to validate an estimated solar irradiance data from a geostationary satellite, Himawari-8, whose product output intervals is 2.5 minutes.
Model performances of both cloud field and solar irradiance (including direct and/or diffused solar irradiance) are investigated. A usefulness of the finer model with 500m grid spacing will be discussed in this presentation.