SUMMARY

The land use practice in the Red River Delta is characterized by its mosaic spatial repartition and rapid land cover change. Red River Delta is an important rice producer of Vietnam where is living more than 20 % of Vietnamese population. In this context the land use mapping conducted periodically by the Land Administration Agencies can not satisfy land management needs at local scale as it is required at Duy Tien District level. Classifications of satellite image are used in this case study as an alternative to fill up information gap in the period between land survey operations. The image data used in our study is SPOT 4 scene acquired on November 2007. Due to their similar spectral nature and their size smaller than spatial resolution of SPOT4 (pixel size of 20m), different land cover types can easily be classified in confusion with each other. We applied two different methods, namely pixel based method (statistical approach) and cognitive method (object oriented approach), of classification for the same image set to compare their effectiveness and to reduce the spectral confusion of land cover types in Duy Tien. For Object Oriented classification we have used the ancillary information we have collected from the field observation and statistic data provided by the Service of natura