Improvement of the Japan Meteorological Agency Meso-Scale Model for the Forecasting the Photovoltaic Power Production: Sensitivity Experiments of Radiation-Related Schemes

Power production of a photovoltaic (PV) power plant varies according to weather conditions. Therefore, in association with the stabilization of the electric power system which takes account of varying PV power, it is important to predict the PV power production in a few days. The prediction of the PV power production is often executed by the engineering model. The input of the engineering model is from the output of the weather forecasting model so that the accuracy of the weather forecasting model affects the prediction of the PV power production. In particular, the forecast value of solar irradiance, which is calculated as downward shortwave radiation (DSWR) flux at surface in the weather forecasting model, is one of the major factors for the PV power production so that it is necessary to assess the accuracy of its value. In our research group, the output of the Japan Meteorological Agency Meso-Scale Model (hereafter MSM) is used for the input of the engineering model. MSM is an operational, non-hydrostatic and regional model used for a short-range forecast (33 hours). The model horizontal resolution is 5 km and the vertical level is 50. From our previous research, the MSM DSWR tends to be underestimated (overestimated) during summer (winter). Further investigation revealed that above error trends are associated with the MSM cloudiness forecast. So, in this study, to investigate the relationship between MSM DSWR and MSM cloudiness, sensitivity experiments associated with radiation-related schemes are executed. Two types of experiments are executed. One is the sensitivity experiment of the horizontal and vertical resolution (dx = 2 km, 90 levels) in order to change the cloud distribution and the other is the sensitivity experiment of the effective radius of cloud water and ice in order to change the cloud expression associated with the radiation process. The mean absolute error (MAE) of the daily integrated DSWR normalized by the value at the top of the atmosphere is used for the validation of the MSM DSWR error. Twelve cases (large overestimate: 3, slight overestimate: 3, slight underestimate: 3 and large underestimate: 3) are tested. The threshold of “large” is MAE > 0.2, in this study. For the sensitivity experiment of the resolution, the MAE is improved for overestimated cases. In overestimated cases, MSM tends to reproduce low-level cloud thinner than the actual cloud so that the higher vertical resolution improves the cloud distribution. However, the MAE is worse for underestimate cases. In underestimate cases, MSM tends to reproduce upper-level cloud thicker than the actual cloud so that the higher resolution reproduces furthermore thick upper-level cloud. For the sensitivity experiment of the effective radius, when the effective radius is small, the MAE is improved (worse) for overestimated (underestimated) cases. On the other hand, when the effective radius is large, the MAE is improved (worse) for underestimated (overestimated) cases. It is suggested that the property of cloud distribution or expression in the case of underestimate or overestimate affects the accuracy of the MSM DSWR.