Stochastic differential equation (SDE)-based random process models of renewable energy sources (RESs) jointly capture evolving probability distribution and temporal correlation in continuous time. It enabled recent studies to remarkably improve performance of power system dynamic uncertainty quantification and optimization. However

considering the non-homogeneous random process nature of PV

there still remains a challenging question: how can a realistic and accurate daily SDE model for PV power be obtained that reflects its weather-dependent and non-Gaussian uncertainty in operation

especially when high-resolution numerical weather prediction (NWP) or sky imager is unavailable for many distributed plants? To fill this gap

this article finds that an accurate SDE model for PV power can be constructed only using the data from low-resolution public weather reports. Specifically

for each day

an hourly parameterized Jacobi diffusion process recreates temporal patterns of PV volatility. Its parameters are mapped from the day's public weather reports to reflect varying weather conditions using a simple learning model. The SDE model jointly captures intraday and intrahour volatility. Statistical examination shows that the proposed approach outperforms a selection of the latest deep learning-based time series models on real-world data collected in Macau.