Modelling and Forecasting Meteorological Subdivisions Rainfall Data Using Wavelet Thresholding Approach

Abstract

Indian economy is critically dependent on monsoon rainfall. Its precise modelling and forecasting is extremely important for efficient planning and other policy implications. A variety of statistical techniques have been employed in the past to achieve the task. Wavelet analysis is a very powerful newly developing methodology for data analysis. In this paper, this methodology with thresholding is thoroughly discussed and applied to model and forecast meteorological subdivision's rainfall time-series data. Comparison with traditional approaches, like Autoregressive integrated moving average time-series approach and Nonparametric regression approach using local linear regression smoother shows superiority of proposed methodology for data set under consideration.