Road Damage Assessment from High Resolution Satellite Remote Sensing Imagery in Wenchuan Earthquake
基于高分辨率遥感影像的汶川地震道路损毁评估

The strongWenchuan Earthquake causes serious damage to the road infrastructure, which has bad influence on the disaster-reliefwork after the earthquake. Therefore, it is very mi portant to evaluate the damage loss of the road in the disaster area quickly and accurately, especially in the emergency tmi e ofdisasters. This paper introduces amethod to evaluate the damage of roads from satellite remote sensing mi ageswith high spatial resolution inWenchuan earthquake of8.0magnitudes. The study uses twenty nine high resolution satellite remote sensing mi ages after the earthquake, includingEROS-B, QuickBird, SPOT-5, ALOS data etc. These mi ages are allpanchromatic data. The spatial resolution is at least3 meters. The study area includes 20 counties in the damaged area ofWenchuan Earthquake. The terrain conditions are com- plex for all counties except several of the Chendu city. The studymethod includes three steps. First, remote sensing mi agery is geometrically corrected. Meanwhile, clouds are elmi inated by the threshold of high gray values. Second, road regions in high resolution remote sensing mi agery are segmented bygeographic information ofroad. Third, damaged roads are detection by the following steps. Each road is seg- mented by a fixed length interval into several segments. Then foreach segment, grey and texture feature in a neighborhood window are calculated. According to themodel of damaged road, grey and texture feature are used to identify damaged roads. At last, the resultofdamaged road detection is used to calculate assessment indexes including damage distribution and damage grade of each county. The grading standard is setted in three levels, which are severely damaged, moderately damaged and lightly damaged, according to the damage rate ofnational roads and provincial roads. The result shows that the earthquake affected road infrastructure in all the twenty counties of the study area. Six counties are severelydamaged, includingWenchuan county andBeichuan county; three counties aremoderatelydamaged; and eleven counties are lightly damaged. It is proved that, comparing the results of this study and the ground investigation data, the assessmentmethod of this paper could get a quick and relatively accurate evaluated resultof the damage loss of the roads in the disaster areas using the high resolution satellite remote sensing mi ages. The damaged road assessmentmethod isveryusefuland feasible in the serious situation, especiallywhenwe cannotget the ground investigation information.