The measurment and the improving method for predicton of solar meteorological data

The aim of this thesis is try to improve statistical methods for prediction of five cites in southeastern region of Turkey which are Gaziantep with 12 years data, Şanlıurfa with 12 years data, Diyarbakır with 10 years data, Batman with 8 years and Mardin with 4 year data that are provided by Meteorological service of Turkey state. The methods exponentially weighted moving average (EWMA) and exponentially weighted moving average based Gaussian distribution (EWMA-GD) are used in literature of financial, engineering application. However the combination of EWMA and GD are novel method and used first time in literature. The data belong to five cities are divided into classes which are test class data consist of two years data and prediction class data to prediction for global solar radiation (GSR) and sunshine duration(SD). Exponentially weighted moving average (EWMA) is employed for prediction of GSR and SD as first method. This method is used test class years data to predict next years' for each five cities. The second method is called exponentially weighted moving average based Gaussian distribution (EWMA-GD). In EWMA-GD method test class years' data are used to construct Gaussian distribution and then calculate its parameters by using Caurane approaches. EWMA method is applied to the compute parameters of Gaussian distribution (GD) function to predict next years' Guassian distribution (GD) function parameters to construct GD function for predicted years. Both method EWMA and EWMA-GD are used long-term prediction for Gaziantep, Şanlıurfa, Diyarbakır, Batman cities and short-term prediction for Mardin city. The method EWMA and EWMA-GD are test by statistical tools which are determination coefficient R2 and mean absolute percentage error (MAPE). For both method EWMA and EWMA-GD R2 is computed close to 1 means prediction data are fitted by measurement data. Furthermore MAPE is computed between 0-10(kWh/m2) that is indicate perfect accuracy in prediction.