This study compares uniform and stratified sampling strategies applied to hourly solar irradiance signals. The analysis examines how each approach affects signal reconstruction, anomaly detection, and dynamic PV modelling. Using PCHIP interpolation and error metrics such as RMSE and MAE, results show that uniform sampling yields lower global reconstruction error (26.64 W/m2 vs. 32.98 W/m2), while stratified sampling captures instantaneous peaks more accurately under high-variability conditions. Stratified sampling also improves anomaly identification due to its more representative temporal distribution. These findings highlight a practical trade-off between minimizing average error and preserving extreme events, providing guidance for PV estimation, forecasting, and discrete-time control applications.
Bastidas et al. (Fri,) studied this question.