Statistical modeling and forecasting of weather data distribution using improved time series analysis

The current study is intended to investigate the applicability of a special class of time series models; autoregressive integrated moving-average (ARMIA) for the estimation of temperature distribution forecast model. Different transformations of ARMIA models such as differencing and smoothing are investigated, in addition to study the effect of each model parameters on the accuracy of the derived model. This study is applied at a temperature time series data of Riyadh city in KSA. By investigating a number of smoothing techniques, simple exponential smoothing (with = 0.2) is found to be the most adequate forecasting model for the case under study as it yields highest correlation factor (R2= 0.9337).