FLOOD SUSCEPTIBILITY MODELING USING ANALYTIC NETWORK PROCESS AND GEOSPATIAL TECHNOLOGY

Flood forecasting involves the use of spatial physical information and information on
decision maker's preferences. This research integrates Geographic Information
System (GIS) with appropriate Multi-Criteria Decision Analysis (MCDA) technique
known as Analytic Network Process (ANP) model and remote sensing to produce a
robust spatial flood forecasting model and to overcomes the drawbacks of Analytic
Hierarchy Process (AHP) model, which was criticized on the issue of its rank reversal
and for assuming criteria and alternatives to be independent which rarely occurs in
real life situation. Till date, there is a dearth of published reference materials that
utilizes GIS-based ANP model and remote sensing in forecasting flood susceptible
zones so as to give modelers a reliable flood forecasting tool. This research bridges
this gap by developing a hybrid of GIS-based ANP and remote sensing for flood
forecasting. The ANP mathematical model was used to calculate weights for the
various flood influencing factors/criteria. This involved the elicitation of experts'
preferences via ANP survey questionnaires. The outcomes from this process were
integrated into the GIS environment using a loose coupling approach. The flood
susceptible zones were subsequently simulated using the ArcGIS spatial analyst
functionalities. Results from the ANP model revealed the Very Highly Susceptible to
Flooding (VHSF) areas to fanned 38.4% (30924.612ha) of the total area. The results
were further verified using One At a Time (OAT) sensitivity analysis in order to
check its stability; where six out of the twenty two scenarios correlated with original
simulated spatial flood forecasting model produced.