Stochastic Simulation of Daily Solar Radiation from Sunshine Duration

Abstract:

Solar radiation is a key component of the energy balance used for estimating evaporation. As solar radiation is not widely measured, many empirical models have been developed to estimate solar radiation using sunshine hours (SSH) data. Most of these models only provide deterministic estimates of monthly solar radiation and do not provide an estimate of the uncertainty in the predictions. This study developed five stochastic models which use daily SSH data to produce probabilistic simulations of solar radiation, and can be used to estimate historical daily radiation. The predictive uncertainty due to the timing of the SSH during the day (estimated using Monte Carlo simulation), as well as due to external errors (such as the variability in cloud type and atmospheric composition), were considered. The developed models differ in their parameterisation of the direct and diffuse components of the solar radiation, using either no scaling, linear or quadratic scaling of the radiation by the daily SSH fraction to account for cloud attenuation. For each model the simulated solar radiation was compared with the observed radiation. The performance of the five models was compared and the models were found to perform similarly well, with an average error of approximately 9% for all locations studied. The results suggest that the uncertainty due to the timing of the SSH does not dominate predictive errors in global radiation. Rather the external uncertainty is the dominant source of predictive error in the radiation estimates.