Investigation of the Efficiency of Object-based Aerial Digital Images Processing Methods for Identifying and Estimating of Earthquake Damaging Impacts in Varzaghan City

Iran is known as one of the most susceptible countries in earthquakes hazard. Therefore, estimating of the damaged area, in context of earthquake hazard, provides valuable information for risk management. In this regard, the collapsed buildings and the degree of damage in the affected areas, and the type of damage caused by each building leads to assess the caused damage immediately. Technically speaking, this is part of the essential information for successful relief and rescue after an earthquake and reconstruction in disaster areas. In recent years, remote sensing technology has been used as a means of collecting information in crisis management in a large-scale disaster. Therefore, the remote sensing based damage assessment is an efficient technology for obtaining information from damaged buildings at short intervals, at a low cost, and with a vast field of view in urban areas. It has been widely using for assessing earthquake damage and monitoring the damaged buildings. The main objective of this research is to apply object based image analysis (OBIA) for damage detection and mapping in Varzaghan City.

On August 21, 2012, two earthquakes with a low distance of 6.4 and 6.3 magnitudes occurred, in 60 kilometers northeast of Tabriz. That earthquake caused the loss of 327 inhabitants, severe destruction of more than 20 villages and many buildings in both cities of Ahar and Varzaghan. The present study presents a basic object pattern for identifying the damaging effects of the earthquake that has been exploited using digital aerial photography and OBIA techniques. In this research, various satellite image object-based processing techniques have been tested and used to introduce to the most important spatial, geometric and spectral indices in the identification of damaged areas. At first, segmentation in two scales of 60 and 100 is performed using multi-resolution segmentation technique. We also used equal weight for all bands. The coefficient of compression was 0.5, and the coefficient was about 1/0. In doing so, 11 algorithms were applied for destructive detection.

The results of the algorithms were compared and verified. We compared the capability of 11 algorithms. The results with 5 meters buffer scale indicated that all algorithms covers the correctness of over 90%. In the scale buffer of 10 meters, the results indicate the accuracy of 93.93 percent. According to results, the geometric and rounded shapes with 96.07% and elliptic with 92.12% represent the highest efficiency. Depending on the destructive areas extracted, shown on separate maps for each indicator, buildings destroyed in the old city's texture on the edge of the river, the core of the town of Varzagan has been concentrated. Basically, it has residential, livestock and storage facilities for keeping livestock forage. New or almost new materials with industrial, residential and other uses are not damaged. Results of this research are important for risk management, and can be used for rapid monitoring of damaged area and zones after earthquake hazards.