Modelling Extreme Rainfall in Zambia using the Generalised Extreme Value Distribution
No Thumbnail Available
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Lusaka
Abstract
This paper delves into modelling Zambia's monthly rainfall using extreme value theory. A survey of the literature is included in Chapter Two, outlining the development of extreme value modeling and highlighting its multidisciplinary applications. The empirical review examines extreme value theory (EVT) from global, regional, and local viewpoints, demonstrating its varied uses in different climatic conditions. The study methodology is described in Chapter Three, which uses a hybrid approach with an emphasis on the explorative design. For model specification, the Generalized Extreme Value (GEV) distribution was selected. The analysis is based on data from the Climate Change Knowledge Portal of the World Bank Group, covering the period from January 1901 to December 2021. Chapter Four explores the results of the data analysis making use of the summary statistics and descriptive statistics to see what the data looks like. The monthly rainfall data's stationarity is confirmed by the Augmented Dickey-Fuller Test, which is an essential component of trustworthy time series modelling. After parameterizing the GEV distribution, diagnostic plots confirm that it is appropriate; the Fréchet type is found to be the best fit. Estimates of return levels and the likelihood of experiencing intense rainfall events augment the study's understanding.As thresholds are raised, the likelihood of excessive rainfall decreases, which is consistent with increased uncertainty in the prediction of more extreme storms. The trend of return levels is rising, suggesting that there may be more intense rains in the future even though the probabilities of these intense rains isn’t huge.
Keywords: Extreme value theory, GEV distribution, Fréchet distribution, return levels, stationarity, climate change.