A quantitative analysis of different change detection algorithms over rugged terrain MODIS sensor satellite imagery

Abstract

Remotely sensed data is only a key source of detection of Earth's surface changes or Land-Use/Land-Cover (LULC) monitoring. During past decades, a series of effective change detection techniques such as Principal Component Analysis (PCA), Change Vector Analysis (CVA) and Post Classification Comparison (PCC), have been developed to observe the LULC vicissitudes. All aforesaid techniques performed very well in some situations such as on horizontal land but very rarely experimented on rugged terrain satellite imagery because change detection procedures are very problematic for such study areas. From this perspective, this study comprises the quantitative analysis of different change detection techniques to study LULC changes linked with rugged terrain Moderate Resolution Imaging Spectroradiometer (MODIS) sensor satellite imagery. In addition to this, necessary pre-processing steps such as geometric correction, radiometric correction and topographic correction for flat surface as well as rugged terrain, have also been summarized to correct the estimated spectral reflectance value. Experiment outcomes confirms that CVA technique has greater potential (achieved accuracy assessment of 90% with Kappa coefficient of 0.8838) than PCC and PCA techniques (achieved accuracy assessment of 78-82% with Kappa coefficient of 0.6358-0.7537, respectively) to analyzed the overall transformed information over rugged terrain MODIS satellite imagery. It is projected that this will efficiently guide the natural hazard forecaster's or algorithm engineer's to precisely perceive the multi-temporal environment changes over Land-Use/Land-Cover rugged terrain.