Under the condition of high penetration of PV system, large forecast error of solar irradiance can affect power supply and demand adjustment for example shortage of power supply and surplus power. Therefore, this study develops a model to foresee large forecast error of solar irradiance based on a variety of forecasted values of weather parameters such as low-level cloud cover caused by different combinations of physical schemes in a numerical weather prediction (NWP) model. Weather Research and Forecasting (WRF) is used as NWP model. The performance of the large forecast error foreseeing is evaluated by using Receiver Operating Characteristic (ROC) curves and False alarm rate (FAR) at Hit rate (HR) of 0.9. Target areas are Nagoya, Shizuoka, Nagano and Toyama. The results show that low-level cloud cover is useful for foreseeing of large forecast error of solar irradiance in summer at each area because of low false alarms with high hit rates. Furthermore, it is shown that useful weather parameters for foreseeing large forecast error of solar irradiance differ depending on areas and seasons, for example, in Nagano, Middle-level cloud cover is useful in seasons other than summer.
J-STAGEがリニューアルされました！https://www.jstage.jst.go.jp/browse/-char/ja/
