Trend Analysis and Change Point Detection of Temperature Patterns and Relative Humidity in Enugu, South-Eastern Nigeria

Abstract

This study examined the temporal variations and abrupt changes in temperature patterns and relative humidity in Enugu, South-Eastern Nigeria. A 31-year daily record of temperature (minimum and maximum records) and relative humidity were obtained from the Nigerian Meteorological Agency (NiMET). The data were processed to obtain annual maximum and minimum temperature and annual relative humidity data series for the climatic variables. Three trend detection methods (Linear regression analysis, Augmented Dickey-Fuller (ADF) test, and Mann-Kendall trend test) were employed for the trend detection in the time series data. Sen's slope estimator was used to quantify the magnitude of the trend after Mann-Kendall trend detection test was conducted. Abrupt changes in the temperature and relative humidity data distribution were investigated using Distribution-free cumulative sum (CUSUM) and Sequential Mann-Kendall methods. The results revealed that annual minimum temperature exhibited a statistically significant increasing trend through linear regression (slope = 0.068°C/year, p = 0.035, R² = 0.144) and Mann-Kendall test (tau = 0.284, p = 0.031; Sen's slope = +0.045°C/year). The ADF test confirmed non-stationarity in minimum temperature (ADF = -2.693, p = 0.075), implying a non-constant statistical parameter over time. However, no significant trend was observed in the annual maximum temperature and relative humidity using Linear regression and Mann-Kendall. Change point detection identified significant changes in minimum temperature around 2020-2022 (CUSUM value = 7, critical value = 6.79).