Many simulations and performance evaluations of solar energy systems require information on local solar irradiance as input and the most commonly required parameter is the global horizontal irradiance (GHI). Irradiance at differing geographical locations is frequently measured using sensors which are installed in the plane of existing photovoltaic (PV) arrays, whose tilt is optimized to the location. One is therefore often compelled to convert the irradiance from an arbitrary tilt to GHI as would be measured on a horizontal surface. The existing literature, however, focuses on using transposition and decomposition models to predict solar irradiance on tilted surface from solar irradiance data on horizontal plane. In this paper, we discuss the reverse process. We use Singapore data collected at various tilts and azimuths to perform the analysis. We first evaluate the performance of various transposition and decomposition models in tropical regions. We assess constraints on each decomposition model and select an optimal model using measured GHI data as a benchmark. (C) 2013 Elsevier Ltd. All rights reserved.