Statistical evaluation of precipitation pattern via time series clustering: insights into the Bundelkhand region

Abstract

Rainfall is a critical climatic factor influencing ecosystems, agriculture, and water resources, particularly in India, where agriculture relies heavily on rain-fed conditions. The fluctuations in this climatic element hold significant importance for researchers and policymakers as this influence decision-making processes. This research aims to analyze the precipitation patterns in Bundelkhand region consisting of parts of Uttar Pradesh and Madhya Pradesh states of India, during 1951 to 2023 and project future trends using time series (TS) clustering methods. TS clustering, which identifies patterns in climatic data through techniques like Singular Spectrum Analysis (SSA), offers valuable insights into the trends, seasonality, and autocorrelations of rainfall data. By clustering these patterns, the study seeks to enhance understanding of rainfall variability, providing a basis for improved agricultural planning and decision-making. Additionally, the projections of the rainfall have also been done for future 24 months.  The result clearly identifies two different zones (clusters) in the Bundelkhand region and in both the zones (Southern and Northern), a decrease in rainfall patterns has been noted over time. Total rainfall is more in Southern part as well as fluctuations in rainfall is less as compared to Northern part. The yearly rainfall in the Bundelkhand region is strongly irregular throughout the period, unevenly spread across the months; and highly concentrated in one third of the year.