Statistical analysis of climate variability in Rwanda

Abstract

The present study explores climate variability and reflects climate characteristics in Rwanda using a robust methodology that combines advanced computation and classical statistical methods. The study uses a dynamic linear state space model, using a Bayesian analysis and Markov chain Monte Carlo simulations to estimate model parameters and account for uncertainty in a trend analysis. K-means clustering based on the Euclidean distance is used for climate data clustering. The Mann-Kendall test and Sen’s Slope estimator analyse monotonic trends and their magnitudes in time series. The variability was analysed using standard deviation. The standardised precipitation index assesses drought characteristics and logistic regression analysis behaviour. Temperature trend estimates have shown a high positive increase in annual mean temperature over eastern Rwanda. The mean seasonal minimum temperature shows a significant positive change across all seasons, with a notable rise in summer.

Among three identified distinct near-homogeneous zones, the southeastern zone of eastern Rwanda experiences notable rises in maximum temperature. The seasonal and annual mean minimum, maximum, and mean temperatures indicate relatively little variability (less than 1 °C) in the three zones. Rainfall over the region shows non-significant decreasing and increasing trends in the March to May and September to December seasons, respectively. The rainy season onset shows a significant change starting earlier than before. The season length indicates an increase across the region in both seasons, and it is significant for the September to December season. This study indicates that the drought frequency, duration, and severity have increased over the eastern region since 2010.

The study shows that knowledge about climate change and socioeconomic factors influence farmers’ decisions in choosing adaptation strategies. The study’s findings on climate variability in Rwanda have the potential to significantly impact the understanding of climate dynamics and inform future research and policy decisions.