SOLAR MONITORING, FORECASTING, AND VARIABILITY ASSESSMENT AT SMUD

摘要

The paper summarizes the deployment of a 71 station solar monitoring network in Sacramento, California, and its use in validating variability relationships as well as satellite based irradiance datasets. The data cleanup methods are described for eliminating shading artifacts in the ground-based solar monitoring data. The cleaned data is then evaluated to confirm theoretical relationships of spatial correlation between PV plants developed by Hoff and Perez. The relationships are confirmed for 1 minute, 5 minute, and 10 minute timeframes. Additionally, the ground-based datasets are compared to satellite datasets for determining error. Possible sources of error are discussed, and results show that for a half hour timeframe, error or difference in GHI is between 6 and 11%. For DNI, errors range from 17 - 22%. A portion of the errors can be attributed to bias, with GHI bias ranging from between -1 and -7% indicating satellite estimated slightly greater GHI resource and DNI bias ranging from between -1 and 11%, indicating generally that the ground-based RSR's measured slightly greater values than the satellite datasets.