Abstract

Abstract:
Detailed information on the spatial pattern of land cover within and around urban areas is critical to address a wide range of practical problems relating to urban flood risk management. For example, flood loss estimation modeling requires an estimate of number of various types of urban structures, total square footage of residential and non-residential buildings for adequate estimation of urban flood damage.

Satellite remote sensing provides a powerful tool to acquire this information spatially consistent manner. The practical value of remotely-sensed data has increased significantly in this context with the advent of new, very high spatial resolution optical sensors. However, these new sensors demand new information-extraction methods in order to derive maximum benefit from the data that they produce. Techniques are required to infer land cover from the raw spectral signals that are recorded. There has been significant research in the use of remote sensing for deriving urban areas. However, several of these efforts were not able to handle the complexity of the urban landscape. Recent research points out the need for multi-band data co-registration and fusion using various sensors and ancillary data sources to increase the accuracy of land cover classification in urban environments.

The paper presents a methodology for classification and integration of high and medium resolution optical sensors such as IKONOS, SPOT and LANDSAT to characterize urban assets and environment and obtain information useful to analyze the specific natural hazard risks of urban runoff and flooding and loss estimation modeling. For the purpose, first airborne very high resolution data (of a few centimeter) is used in limited selected locations in urban watershed to resolve the complex spectral signals of the urban features and to derive the footprints of various inventories of built environment. The results of analysis of the airborne data are used for development of appropriate classification methodology for multi-spectral imagery and their integration to discriminate with a high degree of confidence the difference between built and natural environment and to quantify the important structural parameters associated with large-scale urban and suburban developments (floor areas, material or structural types and usage).