Abstract
Sunshine duration (SD) is an essential atmospheric indicator which is used in many agriculture, architects and solar energy applications. In many situations where data of sunshine duration may not be available due to temporal and financial constraints, developing alternative indirect methods based on theoretical considerations for determining SD are essentially required. In this study, seven models were developed using stepwise regression technique to estimate monthly sunshine duration for Libya. The predictors which were used as inputs differ from one model to another and they included monthly cloudiness index, total day length, mean relative humidity, depth of precipitation, mean maximum temperature, altitude and longitude over 16 meteorological stations spread across Libya during the period of 1961 – 2000 . The evaluation of the developed models was performed using a set of data of four meteorological stations representing different physiogeographic and climatic zones during 2001 and against Abdelwahed and Snyder (2015) equations which were developed for estimating sunshine duration for Libya. The statistical parameters of evaluation criteria included mean absolute error (MAE), root mean square error (RMSE), (RMSE %) and Nash and Sutcliffe Efficiency (NSE). The linear regression equation relating predicted with measured data with intercept equals zero and determination coefficient (R2) were also used for evaluation purpose. According to the performance indicators, it was detected that six of the developed models were superior to the model with one parameter (cloudiness index) in estimating the sunshine hours. It was also found that all the developed models have better performance in estimating the sunshine duration as compared with Abdelwahed and Snyder (2015) equations. However, due to its few required variables, a model with two parameters (cloudiness index and total day length) is sufficient and can be used with confidence in estimating sunshine duration for Libya. Keywords: Sunshine duration, Stepwise regression, Statistical model. arabic 15 English 69