Impact of time interval on the Ångström-Prescott coefficients and their interchangeability in estimating radiation

The Ångström-Prescott (A-P) model, a highly rated model, has been widely used to estimate global solar radiation (H) at daily or monthly scales in different locations around the world. However, few studies focused on the interchangeability of the A-P coefficients at different time scales, especially finer scales. The objectives of this work were to determine the A-P coefficients at daily (TS1), 5-day (TS2), 10-day (TS3), and monthly (TS4) time scales and to investigate the variations of the A-P coefficients caused by the time intervals and how this variations impacts on H estimation. Long-term records of H (1961-2000) from a total of 15 stations in the Yangtze River basin were used in the present study. The model performance was evaluated using root mean squared error (RMSE), coefficient of determination (R2), mean absolute difference (MAD), and Pearson coefficient. Better fit were found at time scales of TS2 and TS3 (R2 = 0.86 and 0.84, respectively) between the observed and predicted H. For each station, slight differences existed between the coefficients at the studied time scales. In general, the difference in coefficients increased with the increased of time interval from TS1 to TS2, TS3 and TS4. The largest differences in the coefficients were found between TS1 and TS4. Nevertheless, further analysis demonstrated that the coefficients calibrated at finer time scale could be applied to estimate H at larger time scales with an acceptable accuracy, and vice versa. The results have significant implications to facilitate the calibration and choice of the A-P coefficients.