Brightness-dependent Tarpley model for global solar radiation estimation using GOES satellite images: Application to Uruguay

A statistical model based on Tarpley’s original solar irradiation model is proposed and evaluated using data from seven new ground stations in Uruguay. The model estimates hourly global solar irradiation on a horizontal surface using GOES-East satellite images. We show that the introduction of a simple brightness dependence in the model parameters drastically improves the accuracy of the estimates. The implementations of the original and the improved models are explained in detail. For each case, the parameters are adjusted using controlled quality solar irradiation data and a comprehensive performance analysis, on an hourly and daily basis, is carried out using data from independent measurement sites. The relative RMS deviations for the estimates of the brightness-dependent model are 7.1% for the daily estimates and 14.0% for the hourly estimates, relative to the mean of the measurements. This represents a reduction of approximately 45% (daily) and 25% (hourly) in the relative RMS deviations of the original model. The improved model provides a good balance between simplicity and accuracy.

► Tarpley’s model for estimating solar irradiation from satellite images is revisited and improved.
► A brightness-dependent version is proposed and evaluated against independent data.
► An rRMS of 7.1% on a daily basis and 14.0% on an hourly basis is obtained.
► This represents a significant improvement over the standard model, preserving its simplicity.
► A detailed evaluation of both, the standard and the improved models, is provided.