A statistical approach for estimating diffuse illuminance on vertical surfaces

This paper presents a statistical approach for estimating vertical diffuse illuminance in a tropical environment. A two-year period (2003–2004) of vertical and horizontal diffuse illuminance data from four measuring stations in Thailand were used for the model development. These data were divided into three groups according to sky conditions, namely clear, overcast and partly cloudy skies. Vertical and horizontal diffuse illuminances in overcast conditions were directly related using regression analysis and an equation relating vertical diffuse illuminance to horizontal diffuse illuminance was established. A similar set of equations was established for vertical sunlit and shaded surfaces under clear sky condition. To obtain the vertical illuminance model for all sky conditions, the correlation equations for clear and overcast skies were linearly combined using cloud cover as a weighting parameter. The performance of this regression model was evaluated using a one-year period (2005) of measured data. The discrepancies between the calculated and measured vertical illuminance in terms of RMSD range from 15.1% to 22.4%, depending on sky conditions, the orientation of the receiving surface and the positions of the sun relative to the surface. The performance of the proposed model was also compared to that of 12 models reported in literatures. It was found that the proposed model has the lowest RMSD for clear, partly cloudy and overcast sky conditions.