Multidecadal summer monsoon rainfall trend reversals in South Peninsular India: A new approach to examining long-term rainfall dataset

A long-term (1901–2020) gridded summer monsoon rainfall dataset for South Indian Peninsula has been examined to identify prominent rainfall trend reversals during the last 120 years. The current understanding of long-term rainfall variability in India is largely based on the linear monotonic trend analysis over a large geographical area in which multidecadal signatures and their geographical variations are missed out. Therefore, multidecadal rainfall trend variations are not clearly recognized. To overcome this limitation, an innovative, multipronged, and robust methodology based on mathematical and statistical tools is developed involving: 31-years moving average of percentage departure of summer monsoon rainfall for 120 years at district level; 15-years sliding trend analysis to identify the year of inflection point based on a change in direction of the trend; K-Means cluster analysis; Z-score based normality test of clusters; and determination of prominent timeframe of reversal of multidecadal rainfall trend. Using this methodology, long-term summer monsoon rainfall data (1901–2020) at a district level in South Peninsular India has been analysed. Three major timeframes of rainfall trend reversals have been identified during the last 120 years. These are: (1) from decreasing to increasing rainfall trend around 1934, affecting 67% of the area; (ii) from increasing to decreasing rainfall trend around 1969, affecting 45% of the area; and (iii) from decreasing to increasing rainfall trend around 1986, affecting 45% of the area. The significance of this study is that it identifies systematic reversals in rainfall trends in long-term data. The implication of this study is that it highlights the need to identify the corresponding meteorological, climatic, or anthropogenic causal factors behind the observed rainfall trend reversals.