Satellite-based methods to predict daylight illuminance data and sky types under subtropical context.

摘要

Daylight data are of great importance for daylight design in buildings. Recently, researchers in Europe, Japan and Thailand proposed some satellite-based models to derive illuminance data and estimate sky patterns from geostationary satellite images. This study furthers the field by using geostationary satellite visible channel data to predict daylight illuminance and sky conditions under subtropical climate condition. This study used GOES-9 visible channel data and corresponding ground measurements collected in the IDMP station in The Chinese University of Hong Kong from June 2003 to May 2005. Statistical models both in indirect approach and direct approach have been proposed to derive global horizontal illuminance. The RMSE of hourly illuminance predicted by indirect approach and direct approach are 35% and 31% respectively. A new sky type index has been proposed to classify three sky conditions based on 15 CIE Standard General Sky. The sky type index is a function of cloud index derived from satellite data and the clear sky index that can be either calculated by normalizing global irradiance for an idea clear sky (model 1) or derived by the n - kc relation (model 2). Both model 1 and model 2 can predict three sky conditions well especially for overcast sky conditions that are in the majority in subtropical Hong Kong. The relative errors to predict overcast, partly cloudy and clear skies are 0%, 8%, and 9% for model 1 and 3%, 1% and 5% for model 2.;The results show that satellite-based methods can be applied under subtropical context to predict daylight illuminance and sky conditions using satellite visible channel data. Both indirect approach and direct approach to derive daylight illuminance have particular advantages and shortcomings. Direct approach has better accuracy, while indirect approach only needs global irradiance which is one of the most commonly measured meteorological parameters. The choice of which approach to use in the development of model depends, to a great extent, on the ground data available. The simple sky type index based on cloud index derived from satellite images can predict the sky conditions for places where ground measurements are not necessarily available.