New method for clear day selection based on normalized least mean square algorithm

Abstract

A new method is proposed to select clear days from data sets of solar irradiation recorded with ground-13 based instruments. The knowledge of clear days for a given site is of prime importance both for the study of 14 turbidity and for the validation of empirical models of Global Solar Radiation (GSR). Our innovative method is 15 based on the Normalized Least Mean Square (NLMS) algorithm that estimates noise according to a GSR model. The 16 developed method named Clear Day Selection Method (CDSM) is compared to the well-known clearness index 17 criteria (kt) taking data collected at Tamanrasset in Algeria during the period 2005-2009. The root mean square error 18 (rmse), the mean absolute percentage error (mape) and the dependence of model error (mbe) are considered for the 19 comparison. A different number of clear days is found with both methods, with additionally a kt dependency for the 20 clearness index criteria. The average values of rmse, mape and mbe between the daily average of the measured GSR 21 and its estimate using a model are better in case of CDSM for the period 2005-2009. Indeed, we found 25.28 W/m 2 , 22 4.61 % and 2.09 W/m 2 respectively for CDSM and 42.48 W/m 2 , 7.63 % and-5.91 W/m 2 for the clearness index 23 method with kt = 0.7. We also found that GSR of clear days is well correlated with the model in case of CDSM, 24 which gives good confidence in our results.

Study of Atmospheric Turbidity in a Northern Tropical Region Using Models and Measurements of Global Solar Radiation

“…However, only clear days are considered to study the turbidity. They are selected from the dataset according to an appropriate method [37], thereby reducing the number of useful days to 870, that is to say, that 40% of the observed days at Tamanrasset are determined to be clear.…”