Summer Monsoon Rainfall Forecasting over Gangetic West Bengal, India Using Linear and Neural Regression and Intrinsic Mode Functions

Authors

  • Vidhya Sagar

Keywords:

South West Monsoon (SWM) rainfall; Intrinsic Mode Function (IMF); Generalized Regression Neural Network (GRNN); Quasi-biennial Oscillation (QBO); Inter annual variability (IAV).

Abstract

It is demonstrated that the rainfall data from the South West Monsoon in India's sixth meteorological
subdivision, which includes Gangetic West Bengal, can be broken down into eight empirical time series, or
intrinsic mode functions. As a result, the first empirical mode is identified as a nonlinear component of the data,
whereas the following modes are identified as linear components. While the linear portion is logically described
using the straightforward regression method, the nonlinear portion is modeled using the Neural Network-based
Generalized Regression Neural Network model technique. As confirmed, the various intrinsic modes have a
strong correlation with pertinent atmospheric characteristics, such as the sunspot cycle, El Nino, and the quasibiennial oscillation. It has been found that the suggested model accounts for about 75% of the interannual
variability (IAV) of the Gangetic West Bengal rainfall series. Independent analysis of the actual data confirms
that the model is effective in statistically predicting the region's South West Monsoon rainfall. The actual rainfall
for 2012 and 2013 was 93.19 cm and 115.20 cm, respectively, within one standard deviation of the mean rainfall,
although the statistical predictions of SWM rainfall for GWB were 108.71 cm and 126.21 cm, respectively

Downloads

Published

2025-06-06

Issue

Section

Articles