Forecasting of Summer Monsoon Rainfall over Gangetic West Bengal, India Utilising Intrinsic Mode Functions, Linear and Neural Regression
Abstract
The South West Monsoon rainfall data of the meteorological subdivision number 6 of India enclosing Gangetic West Bengal is shown to be decomposable into eight empirical time series, namely Intrinsic Mode Functions. This leads one to identify the first empirical mode as a nonlinear part and the remaining modes as the linear part of the data. The nonlinear part is modeled with the technique Neural Network based Generalized Regression Neural Network model technique whereas the linear part is sensibly modeled through simple regression method. The different Intrinsic modes as verified are well connected with relevant atmospheric features, namely, El Nino, Quasi-biennial Oscillation, Sunspot cycle and others. It is observed that the proposed model explains around 75% of inter annual variability (IAV) of the rainfall series of Gangetic West Bengal. The model is efficient in statistical forecasting of South West Monsoon rainfall in the region as verified from independent part of the real data. The statistical forecasts of SWM rainfall for GWB for the years 2012 and 2013 are108.71 cm and 126.21 cm respectively, where as corresponding to the actual rainfall of 93.19 cm 115.20 cm respectively which are within one standard deviation of mean rainfall.
Copyright (c) 2020 Pijush Basak
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors retain copyright of their work, with first publication rights granted to Tech Reviews Ltd.