Risk and vulnerability assessment to tsunami hazard using very high resolution satellite data

1. Introduction

Vulnerability Assessment and Risk Modelling are important components for an effective End-to-End Hazard Early Warning System and therefore contribute significantly to disaster risk reduction. The knowledge about elements at risk, their susceptibility, coping, and adaptation mechanisms is a pre-condition for the setup of people centred warning structures, local specific evacuation planning and recovery policy planning. The work presented here is embedded in the Numerical Last Mile Tsunami Early Warning and Evacuation Information System (Last-Mile) project. The focus is on the capabilities of high resolution satellite data as basis to extract highly detailed information on the heterogeneous and highly structured urban area of Padang, Indonesia, to assess spatial vulnerability in case of a tsunami impact.

2. Approach

The approach to assess risk and vulnerability of the proposed paper is two-fold: On the one hand automated extraction of the highly structured and heterogeneous urban objects from very high resolution satellite data is presented. The IKONOS Sensor provides a high geometric resolution of 1m and enables highly detailed analysis of the urban morphology. An object-oriented hierarchical classification approach was implemented for an extraction of detached houses, of main street infrastructure, of vegetation areas separated in grassland and tree areas, of bare soil and of water areas. This classification of the current status of the built-up environment provides the necessary spatial information for substantial and sustainable management decisions.

On the other hand the approach focuses on the derivation of further indicators necessary to support risk management. Based on the classification result further indicators are calculated to contribute to the holistic perspectives assessing risk and vulnerability. The derivation of physical parameters of the structures, like building shape, height, size, or roof type enables a correlation between the vulnerability of the houses and the height of a potential tsunami wave impact. In addition the classification of the street network based on orientation with respect to the expected tsunami impact, the widths of the streets, the hierarchical order, as well as the street network density enable the assessment of evacuation routes. In combination with the spatial population density distribution an assessment of the carrying capacity can be performed. The population density is calculated using indicators like the built-up density, building heights or the land use using a top-down interpolation technique. Based on generalized census data the distribution of people with respect to the dynamic spatial shift within the course of a day is performed on a very local level.

In addition the digital elevation model from the SRTM (Shuttle Radar Topography mission) is used to model various tsunami impacts. Moreover, the derivation of surface roughness parameters, which are integrated in the inundation modelling, is investigated based on very high resolution remote sensing data. This enables a coarse assessment of tsunami-prone areas as well as the assessment of areas at high landslide risk using a slope calculation. This spatial knowledge in combination with the indicators derived on the urban landscape (population, houses, street network) makes the analysis of -how many- people are -where- at risk, possible.

3. Conclusion

The aim is support the preparation of an action plan for evacuation based on the spatial analysis of the derived layers. This spatial knowledge enables the identification of safe areas, and the planning of space-oriented evacuation routes based on very high resolution satellite data for different scenarios.