Skip to Content
MeteorologyMeteorology
  • Article
  • Open Access

19 November 2026

20 Pages

Meteorological Models Evaluation for Optimized Photovoltaic Power Forecasting

,
,
,
,
,
,
,
,
and
1
Guodian Nanjing Automation Co., Ltd., Nanjing 210032, China
2
School of Atmospheric Physics, Nanjing University of Information Science and Technology, Nanjing 210044, China
3
National Demonstration Center for Experimental Atmospheric Science and Environmental Meteorology Education, Nanjing University of Information Science and Technology, Nanjing 210044, China
4
School of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, China

Abstract

Accurate meteorological forecasts are critical for optimizing photovoltaic (PV) power generation, yet the performance of numerical weather prediction (NWP) models at specific power plant sites remains underexplored. This study evaluates four meteorological models—a commercial model (named GD), the NUIST regional model, the European Centre for Medium-range Weather Forecasts-Integrated Forecast System (ECMWF-IFS), and the German Weather Service ICOsahedral Nonhydrostatic-seamless (DWD ICON-seamless) model—against 15-min observational data from three PV sites in China (coastal low-altitude ZJ, inland low-altitude JX, and high-altitude QH) over periods spanning 2023–2025. Standard error metrics (mean bias, mean absolute error or MAE, root-mean-square error or RMSE, index of agreement or IOA, and correlation) are applied to 10-day forecasts of surface pressure, temperature, humidity, wind, and 6-h forecasts of solar radiation components. The NUIST model substantially outperforms GD in 2-m temperature at QH (MAE reduction of ~52%) and in 10-m wind speed at ZJ (MAE reduction of ~16%), while GD shows slight advantages in wind speed at inland sites and in wind direction bias. For ultrashort-term solar radiation, the ICON model consistently achieves the lowest MAE for total shortwave radiation across all sites, with a 26.8% reduction at ZJ compared to GD. All models exhibit systematic temperature underestimation and large uncertainties in wind simulation, particularly at inland locations. These findings provide a quantitative basis for constructing optimized, site-specific ensemble forecast strategies to enhance PV power prediction.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.