Predicting monthly mean daily diffuse radiation for India

An investigation on diffuse solar radiation for 23 stations of India is carried out exploiting the 15 years (1986–2000) recorded data. Seven generalized empirical models are developed correlating the diffuse radiation with the sunshine duration, temperature and relative humidity. The data of 18 sites out of 23 are employed to generate the models and the study is focused on both types of correlations linear and non-linear. For further validation and to check the accuracy of present models, statistical parameters mean percentage error (MPE), mean bias error (MBE), and root mean square error (RMSE) are used. Performance of proposed correlations is compared with the existing model. An exercise is also carried out towards the selection of most suitable equation. The study finds that a correlation of diffuse transmittance with percent sunshine, temperature and relative humidity gives the best fit. Moreover, the measured values of monthly mean daily diffuse irradiance of other five locations Bangalore, Kolkata, Nagpur, New Delhi and Pune are analyzed to evaluate the proposed generalized correlations.

► Seven generalized empirical models for the diffuse radiation are developed for India.
► The statistical parameters MPE, MBE and RMSE are applied to validate the correlations.
► The models are evaluated by estimating the monthly average diffuse radiation of five testing cities.
► An emphasis is made on different combinations of climatic parameters and on raising the power of variables.