Change-point detection in monsoon rainfall of Narmada River (central India) during 1901-2015

Since the middle of the last century, the so-called acceleration of the water cycle, due to global warming, has led to increased rainfall variability worldwide. In this research, change point detection in the monsoon rainfall of the Narmada River basin (central India) was analysed. This analysis, based on seven rainfall stations at an annual scale during the 1901-2015 period, utilised a combination of the Bayesian approach (BA) with the discrete wavelet transform (DWT). The analysis indicates a shift towards drier conditions starting in the 1960s, with a long-term trend beginning as early as the 1920s. It was revealed that the high variability of monsoon rainfall can be attributed to the dominance of intra-annual, multi-annual, and less-than-decadal cyclical phenomena (short- and medium-term phenomena), which mask existing change points, thus creating difficulties for the BA in identifying them. Overall, the BA-DWT methods effectively detected the change and multi-change points in the studied monsoon rainfall time series, thereby outperforming the BA method when applied to the original series.

Automated summary

In this research, change point detection in the monsoon rainfall of the Narmada River basin was analyzed using a combination of Bayesian approach and discrete wavelet transform. The analysis revealed a shift towards drier conditions starting in the 1960s, with a long-term trend beginning as early as the 1920s.