6TH International Congress on Technology - Engineering - Kuala Lumpur3 - Malaysia (2018-07-19)

Flash Flood Warning Sub-systems For Rural Africa

This paper studies flash flood early warning systems and the need for sub-systems in rural Africa for real time warning delivery. It has been reported in previous studies, that sub-Sahara Africa lacks weather radars. Essentially, this means that there are no real time early warnings. This is a grave problem and presents a gap in knowledge that this study aims to address. This is done through the following objective; to examine the relationship between variables in the study and therefore establish whether sub-systems are a significant variable in flash flood warning systems for rural Africa. The variables to be examined are; the independent variable (existing flash flood warning system), the dependent variable (early warnings), the moderator variable (ancillary elements) and the mediator variable (sub-systems). This is investigated through a closed-ended questionnaire that is administered to a sample size of 708 meteorologists whose email addresses are available on the World Meteorological Organization’s expert database. These are experts from 17 river flooding countries in Africa. The target sample is 85 experts as determined through the G*Power application. 173 respondents register their feedback on the google form provided. The data is analysed on SPSS whereby a Cronbach alpha of .968 is achieved. Variables in the study are found to be correlated after conducting a Pearson’s correlation test. Using PROCESS allows for the testing of various models whereby moderation of existing systems (IV) and early warnings (DV) by ancillaries (MV) is confirmed. A moderated mediation model of existing systems (IV) and early warnings (DV) by ancillaries and sub-systems is also confirmed. The results wholesomely confirm that sub-systems are significant enough to be developed for rural Africa. It is further recommended that ancillaries, that are also found to be significant after testing of models, should be enhanced through community engagement.
Stella Mbau, Vinesh Thiruchelvam