Spatiotemporal Trend Analysis of Flood Events Across Africa during the Historical Period (1927-2020)

Código

I-EBHE0008

Autores

Djanna Koubodana Houteta, Mouhamadou Bamba Sylla, Moustapha Tall

Tema

WG 1.01: REHYDRATE - REtrieve historical HYDRologic dATa & Estimates

Resumo

The spatiotemporal distribution of ground-based historical of the total flood events associated with the total deaths, the total affected people, and the total damages recorded in the Emergency Events Database (EM-DAT) was analyzed over Africa. A free software of Quantum Geographic Information System version 3.10.9 and Mann Kendall test was used to show their spatial and temporal distributions between 1927 and 2020. Furthermore, the total number of Consecutive Wet Day (CWD); annual total precipitation amount of very wet day (R95PTOT), Annual Maximum Precipitation (AMP), and Simple Daily Intensity Index (SDII) extreme precipitation indices were computed using ERA5 and CHIRPS bias corrected reanalysis products. The Pearson correlation indices were calculated between total flood events and extreme precipitation indices between 1981 and 2019 in order to detect the most trigger index of flood events occurrence in 16 capital city stations. Therewith, the Goodness-of-fit criteria and statistics were used to select the best fitting model within five different two parameters probability distribution models. Finally, the best suitable model was selected to predict flood events frequency up to 100-years return period in each city. Firstly, the results of spatial distributions showed that the countries such as Nigeria, and Ghana in Western Africa, Ethiopia, Kenya, and Tanzania in Eastern Africa, Algeria, and Morocco in Northern Africa, Angola and the Democratic Republic of Congo in Central Africa, and Mozambique, Malawi and South Africa in Southern Africa had had more recurrent and frequent total floods events, and were in consistence with their spatial distributions of the total deaths. Secondly, there were positive and significant trends of total flood events, total deaths, total affected people, and damages time series over each Regional Economic Communities, as well as in each country and capital city investigated. Furthermore, the CWD; R95PTOT, AMP, and SDII extreme precipitation indices were found to be in this order the main probable triggers of flood events occurrence in most of the capital cities. Finally, nonstationary flood return periods at site stations that derive from Gumbel, Weibull and Gamma best probability distributions have displayed low uncertainties at 95% confidence interval for return periods less than 20-years. These results have important implications for policymakers for formulating the effective flood disaster control strategies, for the planning for at least til 100-years resistant infrastructures, and for the flood forecasting under global warming conditions using the extreme precipitation indices drivers at the levels of African regional economic communities, countries, and capital cities.