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 increasd 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. iii
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.