Analysis of stationary and non-stationary hydrological extremes under a changing environment: A systematic review
Maximo Basheija Twinomuhangi, Yazidhi Bamutaze, Isa Kabenge, Joshua Wanyama, Michael Kizza, Geoffrey Gabiri, Pascal Egli
Abstract
Research on hydrological extremes has increased due to their increasing frequency and destructive power, with their non-stationarity attributed to human activities and climate change. To understand current advances in analyzing extremes, a systematic review of online literature was conducted using PRISMA framework. The review covered several aspects of analysis considered in literature like time series types, non-stationarity detection techniques, frequency analysis (FA) category, probability distribution types, covariates used, parameter estimation and model selection techniques. Results indicate that AMS (71.7 %), Mann-Kendall non-stationarity detection test (70.8 %), GEV distribution (41.4 %), ML parameter estimation (34.6 %) and model selection AIC (30.0 %) were mostly applied. Non-stationary alongside stationary FA was carried out most (82 %) and non-stationary models outperformed the stationary ones. Time was used as a covariate in most studies (50.5 %) compared to anthropogenic (7.1 %), local-scale (11.4 %) and large-scale (31.0 %) climate covariates. Effective hydrological extremes management requires an understanding of their non-stationarity in a changing environment. • Understanding non-stationarity in hydrological extremes is vital for water resources management. • Non-stationary frequency analysis models perform better than stationary ones in a non-stationary environment. • Choice of appropriate covariates is vital in non-stationary frequency analysis. • Estimation of uncertainty yields reliable results in non-stationary frequency analysis.