Electronic International Standard Serial Number (EISSN)
Spatial correlation of solar radiation (SCSR) has a significant impact on the overall data quality when generating radiation time series for multiple sites. Currently, there are no known methods for integration of SCSR into synthetic data by using reduced and easily available inputs. Based on a hypothesis that at long timescales general and simple characterization of SCSR is possible, this paper addresses the problem of modeling monthly and daily SCSR. A regression analysis of satellite-derived radiation data covering over 300,000 locations pairs in 4 US regions is firstly described and general mathematical expressions for SCSR estimation are presented. A procedure for incorporating spatial correlation into conventional stochastic solar radiation models is then introduced by applying the obtained SCSR formulae and the existing methods of linear algebra. Finally, the underlying hypothesis is validated and the effectiveness of the proposed technique for creating spatially correlated monthly and daily solar radiation values is demonstrated based on numerical simulations and analysis of historical data.
regression analysis; solar resource; spatial correlation; synthetic generation