Possibilities of using international solar radiation da-ta in Uzbekistan: approaches for energy and scientific research

Abstract

Relevance: The stability and efficiency of solar photovoltaic power plants (PV) largely depend on the accuracy of solar radiation data. In Uzbekistan, the lack of long-term and high-precision ground-based observations increases the necessity of employing international satellite- and reanalysis-based models. However, the accuracy of such datasets varies significantly depending on cloudiness, atmospheric conditions, and regional characteristics. Therefore, the verification of international datasets against local meteorological station observations is an urgent scientific and practical task. This approach enhances the accuracy of PV generation forecasting, strengthens the reliability of electricity supply, and reduces investment risks.

Aim: The main objective of this study is to improve the accuracy of solar radiation forecasting by comparing and verifying ground-based meteorological station data in Uzbekistan with international datasets (NASA POWER, ERA5, DWD Global, and DWD ICON). On this basis, the reliability of electricity generation forecasts for photovoltaic plants can be ensured, thereby supporting more informed decision-making in planning and operation.

Methods: The study utilized data from 22 solar meteorological stations of the Agency of Hydrometeorological Service of the Republic of Uzbekistan for the period 2022–2024. Measurements were carried out at 10-minute intervals, from which hourly and daily values were derived. International data were retrieved from the NASA POWER, ERA5, DWD Global, and DWD ICON databases. To assess accuracy, statistical indicators such as MAE, RMSE, MAPE, and R² were employed. In addition, daily weather conditions were classified into four categories: clear sky, full haze, partly cloudy, and overcast. This classification allowed accounting for atmospheric conditions when evaluating the datasets.

Results: The analysis showed that, in hourly verification, the ERA5 dataset demonstrated the lowest error values (MAE ≈ 74.8; RMSE ≈ 95.0), identifying it as the most reliable source. The highest errors were observed in the DWD Global dataset (MAE ≈ 106.4; RMSE ≈ 133.7). For daily cumulative values, ERA5 also produced the highest accuracy (MAE ≈ 298.3; RMSE ≈ 371.8). In cloudy conditions, errors increased sharply across all datasets, indicating the strong impact of cloud cover on forecast accuracy. The results demonstrate that, under Uzbekistan’s conditions, the verification of international datasets with ground-based observations significantly improves the reliability of solar radiation forecasts.