Cyclone disaster mapping, monitoring and management using satellite remote sensing and spatial analysis

Abstract

Tropical cyclones are one of the most catastrophic natural hazards regularly affecting tropical coastal areas globally. The overall impacts of tropical cyclones can be minimized using a range of management approaches, in terms of: prevention/reduction, and preparedness. Accurate and detailed spatial information on the natural and built environment is required to support the development, implementation and evaluation of these cyclone management approaches. There is a lack of appropriate mapping approaches and spatial models at local scales (<1000 km2) to derive this information.This thesis developed and evaluated mapping approaches and spatial models at the local scale (<1000 km2) by integrating remote sensing, field data and spatial analysis for assessing, monitoring and managing tropical cyclone impacts. Three specific objectives were addressed: (1) developing and evaluating an approach for assessing the multiple impacts of tropical cyclones through object-based image analysis applied to moderate spatial resolution satellite imagery; (2) developing and testing a risk mapping approach for tropical cyclone impacts using an analytical hierarchy process (AHP) with data derived from remote sensing, field data and spatial analysis; and (3) producing and evaluating a simple risk modelling approach under present and future climate scenarios for tropical cyclone impacts using remote sensing, field data and spatial analysis. The developed mapping approaches and models were tested to justify the applicability in Sarankhola Upazila (151 km2), a local government area in coastal Bangladesh. Cyclone Sidr (2007) was used as a case study for objective one in the study site.After introductory, review and research approach chapters, chapter four developed a method for mapping the multiple impacts of tropical cyclones. The object-based approach was used to map land cover types in pre- and post-cyclone Sidr (2007) SPOT-5 satellite image data. A postclassification comparison technique was then employed to identify changes to features in the natural and built landscapes produced by the cyclone. The findings demonstrated the capability of a remote sensing approach for mapping the multiple impacts of a tropical cyclone. In particular, the objectbased approach provided accurate results for classifying features from pre- and post-cyclone satellite images with an overall accuracy of 95.43% and 93.27%, respectively. Mapped changes successfully identified the extent, type and form of cyclone-induced impacts in detail. The types of impacts identified were vegetation defoliation, destruction and damage of settlement/infrastructure, damage to cropland and the changes to landforms. The damage results were verified using local damage reports and high spatial resolution satellite imagery that were taken after the cyclone. Chapter five integrated remote sensing, field data and spatial analysis for developing a multi-criteria tropical cyclone risk mapping approach at a local scale. Three risk components: 1-vulnerability and exposure; 2- hazard; and 3-mitigation capacity, and their 14 relevant criteria were considered. Thematic raster map layers were prepared for every criterion and weighted using AHP. A weighted overlay technique was used for generating vulnerability, hazard, mitigation capacity and risk maps. The verified results revealed that detailed cyclone risk information can be successfully mapped using geospatial techniques and an AHP at the local scale. The study found that incorporation of mitigation capacity significantly influenced the risk results. The approach also solved several issues for local level multi-criteria data generation, processing and integration. The approach provided was validated by confidence level assessment and comparison to a map of past cyclone impacts.Chapter six developed a geospatial simple risk modelling approach at the local scale for estimating tropical cyclone impacts under present and future climate scenarios. Linear storm surge models were developed, using 100 year return periods from historical cyclone data and a bare-earth digital elevation model (DEM). A local sea level rise scenario of 0.34 m for the year 2050 was integrated with surge models to assess the climate change impact. The resultant storm surge models were used in the risk modelling procedures. The risk models identified the spatial extent and levels of risk, and demonstrated a verified simple and effective approach for modelling tropical cyclone risk at the local scale. The findings also revealed that incorporation of local climate change scenarios substantially increased the risk levels and areas in the study site.This thesis integrated remote sensing, field data and spatial analysis to deliver validated mapping approaches and spatial models for assessing, monitoring and managing tropical cyclone impacts at a local scale in data-poor areas. The presented approaches were useful to derive accurate and detailed information at local scales for cyclone disaster management. In a practical context, the produced information from these approaches can be used by planners and decision makers for developing mitigation plans and strategies to minimise the impacts of cyclones. The approaches presented in this thesis can be applied in other similar coastal environments to generate tropical cyclone disaster management information.