Document Type


Publication details

Aliasgar, K 2010, 'Developing a geoinformatics based early warning system for floods, in the Caribbean, Trinidad and Tobago', PhD thesis, Southern Cross University, Lismore, NSW.

Copyright K Aliasgar 2010


In the Caribbean where hydrometerologically related disasters can be devastating, early warning can be useful in saving lives, livelihoods and in reducing the after effects of disasters. The main obstacle in creating early warning in developing countries is the suitability and availability of data for currently used methods of flood prediction and risk analysis. This research fills this gap by introducing methods which utilizes the data available to develop a scientifically based early warning for flooding in Trinidad. In this research, the development of early warning is addressed in three parts; quantification of thresholds for trigger, identification of vulnerability and risk and development of carrying capacity. This research shows that together, these parts can form early warning.

Quantification of thresholds for the trigger, rainfall, was done by developing intensity/duration threshold for flooding. Using TRMM satellite rainfall data the lower boundary intensity/duration threshold was calculated. Risk assessments were conducted using the Vulnerability Assessment Matrix for Developing Countries, developed as part of this research. Google Earth maps were used to assess socio-economic vulnerability whilst physical and environmental vulnerabilities were determined by using a road map, DEM and land degradation maps. These maps were combined with the previously developed Flood Susceptibility Map of Trinidad to determine risk.

In Trinidad, uncontrolled removal of the natural land cover and forest, within watersheds has led to increased flooding, suggesting a relation between the physical characteristics of the watersheds and flooding. This can be quantified by developing the carrying capacity for watersheds to determine the percentage forest which can be removed from a watershed before the probability of flooding increases. The geophysical terrain characteristics of watersheds were derived and binary logistic regression was used to develop the carrying capacity threshold.

In Trinidad the intensity/duration threshold for floods was I = 4.064D-0.267 where I is in mm/hr and D is in hrs. The flood risk map developed for Trinidad reveals that approximately half of Trinidad and most of the schools, hospitals and roads are at medium and high risk. For carrying capacity it was found that characteristics of the watersheds by themselves were not significant in predicting floods, but the interactions between these characteristics were. The developed algorithm consisted of landuse, geology and compactness ratio.

This research has shown that it is possible to develop early warning systems for developing countries by overcoming the obstacle of insufficient useable scientific data. By monitoring the intensity and duration of rainfall it is possible to determine the onset of a flood and warning can be issued to those at risk. The results of this research are useful to disaster specialists and planners as they can prepare for, mitigate and recover from flood disasters.