A Study of Trend and Change-Point Detection of Rainfall Using Non-Parametric Tests
R Manjula *
Department of Basic Science and Humanities, College of Forestry, Ponnampet – 571216, India.
*Author to whom correspondence should be addressed.
Abstract
Rainfall variability and abrupt shifts are important concerns in monsoon-dependent regions such as the Western Ghats. This study assessed monthly and annual rainfall trends in Ponnampet, Kodagu district, Karnataka, for 1993–2023 using non-parametric statistical methods. The Mann–Kendall test and Sen’s slope estimator indicated a positive annual rainfall trend with a slope of 30.276 mm/year; however, the reported annual p-value of 0.054 was marginal relative to the 0.05 significance threshold. At the monthly scale, August (7.222 mm/year; p = 0.0419) and September (9.159 mm/year; p = 0.0024) showed statistically significant increasing trends. June, July, October and November showed non-significant declining tendencies, whereas January, February, March and December showed no significant monotonic trend. Pettitt’s, Standard Normal Homogeneity, Buishand’s and Von Neumann tests were used to identify abrupt changes. Annual rainfall showed a change point in 2017, with mean rainfall increasing from 2029 mm to 2788 mm. September showed a change point around 2004, with mean rainfall increasing from 132.98 mm to 281.38 mm. August also showed a marked shift, with mean rainfall increasing from 348.26 mm to 828.78 mm. The results indicate notable late-monsoon variability and temporal shifts in rainfall, with implications for agricultural planning, water-resource management and climate adaptation in Ponnampet.
Keywords: Mann-Kendall test, change point detection, trend analysis, non-parametric tests