Forecasting of Metrological Parameters by Decomposition Method: Case of Karachi Pakistan

Abstract

Forecasting is an important subject for different applications in Metrology and Environment fields. The most simple and basic approach for analyzing time series is decomposition. This study is based on the trend component for significant parameters which were obtained through regression models fitted to the polynomial functions of power k for time t. In this regard monthly data of six metrological parameters precipitation amount (Rf), minimum temperature (Tmin), maximum temperature (Tmax), relative humidity (Rh), wind speed (Ws) and atmospheric pressure (Ap) for the period of 25 years (1990-2014 or 300 data points) of Karachi city, Pakistan were considered for forecast. The forecast values will be obtained by multiplying the values of these four components (trend, seasonal, irregular and cyclic) and were compared to the original values to justify the model. The reliability of forecasted values will be checked by using goodness of fit test by means of chi squire statistics χ2, which indicates that all the parameters follow the test except precipitation. After justification we found that our models are accurate and reliable