RAPID BUILDING DAMAGE MAPPING BY OPTICAL SATELLITE IMAGES

Earthquakes are widely regarded as major disastrous natural events claiming large numbers of lives and economical losses every year in Iran or other hazard prone countries. As a repeated experience, in every decade, some major such events have affected seriously urban settings in Iran. Rapid damage mapping associated to the early phase of post-event will help in managing search and rescue operations, resource allocations and preliminary economic loss estimation. Earthquake rapid damage mapping in disaster management phases using remote sensing technology is feasible and has been regarded as powerful technique for semi-automatic or automatic building extraction and damage mapping in urban areas. In November 12, 2017, a strong earthquake (Mw 7.3) occurred on the Iran-Iraq border. In this event, the province of Kermanshah was the most affected area with the city of Sarpol-e Zahab being the hardest-hit zone. For this study, the Chinese TRIPLESAT_2 and SV1 satellite data sets were acquired for both before and after the event. The object-based image analysis has been sought where a building mask map was utilized for extracting textural contents from all building footprints from all images. The building mask has been created as a layer from Google Earth images (Figure 1). A ground truth data is necessary in order to train the ANN and in order to evaluate the outcome of the algorithm. This data can be derived according to a manual process or using an already existing damage data. Since the ground truth information and the survey data of the study area was not available, ground truth and reference damage map was prepared according visual interpretation post-earthquake satellite image .the buildings were classified into two classes of "Changed" or "Unchanged"

کلیدواژه ها:

Optical satellite ، Building extraction ، Rapid damage mapping ، ANN algorithm