Detecting Changes in Regional Rainfall Series in India Using Binary Segmentation-Based Multiple Change-Point Detection Techniques

Abstract

In this chapter, the rainfall patterns of five homogeneous regions in India namely Northwest, West Central, Central Northeast, Northeast and Peninsular India for the years 1871–2013 were analyzed using two change-point detection (CPD) techniques namely Binary Segmentation based on Cumulative sum and Likelihood ratio approaches. The CPD methodology involves estimation of threshold values based on Monte Carlo simulation. On applying the CPD techniques for five regions, the results showed that for most of the regions there were no significant change-points in annual/monsoon rainfall series except for monsoon rainfall in Northeast India. Further, the study extended to detect the changes in future rainfall for Northeast region by analyzing the outputs from global climate models. The findings of this study can help in better understanding the spatial variations and changing patterns of rainfall across India, and consequently in proper planning and management of water resources in the region.