IDENTIFICATION OF CULTIVATED AREA BASED ON PHENOLOGICAL FEATURES IN ARID AND SEMI ARID AREA IN ASIA

Remote sensor data has long been provided new means of studying global patterns and dynamics in vegetation. Times series NDVI curves can reflect a real trend of vegetation growth and therefore have been used to interpret seasonal cycle and global land cover classification. Accurate identification of cultivated areas from natural vegetated area is not easy because of their similar characteristics in semi-arid area of Asia. This study proposed a method of discriminating between cultivated areas and naturally vegetated areas based on seasonal phenological features using NDVI data. 250 m spatial resolution, 16-day composite MODIS NDVI data (MOD13Q1) in 2000-2001 year acquired and several phenological parameters including beginning date, ending date and length of growing season are extracted. Non-cultivated natural vegetated area has similar profiles on beginning and ending dates of growing season with cultivated area while length of growing season and time difference between beginning of season to middle of season were longer. Even both parameter have same features, profiles of amplitude and maximum NDVI values were different. Phenological features of cultivated area differ on locations: west areas have early growing season, while east areas late. These results indicate that cultivated area could be identified using several phenological features and the time-series MODIS images have high potential for this analysis.