Detection of Burn Area and Severity with MODIS Satellite Images and Spatial Autocorrelation Techniques

Sinasi Kaya, ITU, Istanbul, Turkey, Taskin Kavzoglu, Gebze Technical University, Geodetic and Photogrammetry Engineering, Kocaeli, Turkey and Hasan Tonbul, Gebze Institute of Technology, Geodetic and Photogrammetry Engineering, Kocaeli, Turkey
Abstract:
Effects of forest fires and implications are one of the most important natural disasters all over the world. Statistical data observed that forest fires had a variable structure in the last century in Turkey, but correspondingly the population growth amount of forest fires and burn area increase widely in recent years. Depending on this, erosion, landslides, desertification and mass loss come into existence. In addition; after forest fires, renewal of forests and vegetation are very important for land management. Classic methods used for detection of burn area and severity requires a long and challenging process due to time and cost factors. Thanks to advanced techniques used in the field of Remote Sensing, burn area and severity can be determined with high detail and precision.

The purpose of this study based on blending MODIS (Moderate Resolution Imaging Spectradiometer) satellite images and spatial autocorrelation techniques together, thus detect burn area and severity absolutely. In this context, spatial autocorrelation statistics like Moran’s I and Get is-Ord Local Gi indexes were used to measure and analyze to burned area characteristics. Prefire and postfire satellite images were used to determine fire severity depending on spectral indexes corresponding to biomass loss and carbon emissivity intensities.

Satellite images have used for identification of fire damages and risks in terms of fire management for a long time. This study was performed using prefire and postfire satellite images and spatial autocorrelation techniques to determining and analyzing forest fires in Antalya, Turkey region which serious fires occurred. In this context, this approach enables the characterization of distinctive texture of burned area and helps forecasting more precisely. Finally, it is observed that mapping of burned area and severity could be performed from local scale to national scale.

 

Key Words: Spatial autocorrelation, MODIS, Fire, Burn Severity

Back to: Remote Sensing of Forest Disturbance: Pushing the Frontier II Posters