Short-Term Streamflow Forecasting for Run-Of-River Hydropower Plants using PCA-MLP and Local Linear Adjustment
J.D. Bautista-Izquierdo, L.A. Fernández-Jiménez Department of Electrical Engineering, E.T.S.I.I., University of La Rioja, Logroño, Spain 2026-06-27

Abstract

The participation of renewable energy sources in power systems requires reliable short-term generation forecasts to ensure the efficient operation of the electricity market. Small hydropower plants (SHPPs), which represent a significant share of renewable generation, are particularly affected by hydrological and meteorological variability. This impact is especially pronounced in basins with moderate streamflow, seasonal intermittency, and lack of upstream regulation, as occurs in several unregulated sub-basins of the Ebro River. A hybrid forecasting framework is proposed, combining physically based hydrological information and meteorological predictions within a machine learning model. Dimensionality reduction is applied to improve the robustness and stability of the model. In addition, a local linear post-processing adjustment is introduced, incorporating recent flow observations up to the day prior to the forecast, including intraday measurements from the same day. The proposed approach is computationally efficient and requires a low volume of data, showing a significant improvement in forecasting accuracy compared to standard methods. Its design makes it suitable for real-world applications as a preliminary step for forecasting electricity production and its use in electricity markets, where hydrological uncertainty is a common constraint.

Key words: Small hydropower plants, Streamflow forecasting, Machine learning, PCA, Hydrological uncertainty.

References

[1] Z. M. Yaseen, A. El-Shafie, O. Jaafar, H. A. Afan and K. N. Sayl, “A review of state-of-the-art artificial intelligence applications in hydrology,” Water Resources Management (2015). Vol. 29, pp. 4281–4305.

[2] O. Kisi and J. Shiri, “Improving accuracy of river flow forecasting using LSSVR with gravitational search algorithm,” Water Resources Management (2014). Vol. 28, pp. 2073–2090.

[3] J. Zhang, H. Chen and Y. Li, “Streamflow prediction using an integrated methodology based on convolutional neural network and long short-term memory networks,” Journal of Hydrology (2018). Vol. 558, pp. 1–12.

[4] Q. Duan et al., “A state-of-the-art review of long short-term memory models with applications in hydrology and water resources,” Hydrology and Earth System Sciences (2019). Vol. 23, pp. 1–25.

[5] V. Mockus, “Estimation of direct runoff from storm rainfall,” in National Engineering Handbook, Section 4: Hydrology, U.S. Department of Agriculture, Soil Conservation Service, Washington DC, USA (1964).

[6] Natural Resources Conservation Service (NRCS), National Engineering Handbook, Part 630: Hydrology, U.S. Department of Agriculture, Washington DC, USA (2004).

[7] S. Ahmad et al., “Short-term hydropower generation prediction model for run-of-river hydropower plants considering hydrometeorological factors,” Renewable Energy (2021). Vol. 170, pp. 613–627.