Estimating solar radiation from incomplete cloud data

Abstract The performance of a cloud layer model in estimating solar radiation is examined for different intervals (1 to 6 h) between cloud observations. Data sets for these intervals were created from hourly meteorological data for two Canadian and three European stations. In the model calculations a total cloud transmission function is interpolated for hours without cloud observations. Model performance is assessed from mean bias and root mean square errors. Variations in these statistics are small for all interpolation intervals. Cloud analysis confirms that the small variation is due to persistence in cloud cover. Little error is introduced in solar radiation estimates from this model if three-hourly cloud observations, which are commonly made throughout the world, are used.