Key Issues Of Change Detection Using Multi-temporal Satellite Data

Analyses of multi-temporal satellite imagery provide detailed information about changes and trends in urbanization development.Within the thematic framework,two major work packages are solved in the paper.One concentrates on the detection of clouds and cloud shadows in change detection study,the other one promotes and implements an approach to classify imageries in change detection study.In the end,a case study of change detection in Weihai city in China over the last 30 years is studied.In the 1st work package,a new method called Pmask(principle component analysis mask approach)for cloud and cloud shadow detection in Landsat imagery is provided.Landsat surface reflectance data and Brightness Temperature(BT)are used as inputs.Pmask first transports the information to the varimax coordinate axis using transferable principle component methods.Next,using change detection and Quasi-Accurate Detection of gross errors to label the outliers as potential cloud pixels.Finally,the cloud probability mask is checked with the BT band to eliminate the error pixels with the BT threshold.For the change detection datasets,the method improves the efficiency and the applicability of cloud and cloud shadow detection in change detection study.The 2nd work package presents a fast and reliable change detection approach and was processed with a total of 85 Landsat scenes of Weihai city,China that were recorded between 1984 and 2017.The strategy was developed based on a transferrable,segmented histogram-matching approach that resulted in a high overall accuracy of land-use classification of 90-96%.In contrast to the non-matched data,this matching approach significantly improved the general accuracy by up to 12.67%.The outcome of the study revealed a consistent urbanisation of Weihai City over the past 34 years that is related to policies,economic conditions and family households.The urban area grew at an average rate of 4.1 km2 per year,while the water,bare ground and vegetation areas decreased by 0.3 km2,2.7～3.6 km2 and 0.4～0.6 km2 per year,respectively.