Extraction of land use/land cover using multi-temporal sentinel-1A and Landsat integration: Case study of Hanoi

Satellite images is the major source for classification of land-use/land-cover (LULC). For optical satellite images, the classification is usually based on the spectral reflectance characteristics of the objects. However, optical data is affected by clouds and weather conditions. Radar remote sensing data is less affected by weather conditions but the information on SAR images is only backscatter of roughness of objects. Integration of Sentinel-1A and Landsat data has the main advantages of combining the all-weather capability of the radar sensor, rich spectral information in the visible-near infrared spectrum, with the short revisit period of both satellites. In this paper, a method of integrating multi-temporal Sentinel-1A data and Landsat data were proposed to classify LULC mapping. Normalized difference vegetation index (NDVI), a Normalized Difference Water Index (NDWI) andEnhanced Built-Up and Bareness Index (EBBI) were combinated with the standard deviation, the average of backscatter value of multi-temporal Sentinel-1A and phenology of double-cropped rice. A case study was Hanoi, Vietnam with data included 12 scences of Sentinel-1A, single-polarization VV, Interferometric Wide Swath mode (IW) and GRDH level acquired from Dec 2014 to Nov 2015 and 1 scence of Landsat 8OLI in May 2015. The integrated dataset was classified with Decision tree classification method, which showed the overall classification accuracy of 87%. © 2018 Proceedings - 39th Asian Conference on Remote Sensing: Remote Sensing Enabling Prosperity, ACRS 2018