مقايسه روشهاي مختلف پايش پوشش گياهي در منطقه بياباني عين خوش دهلران با استفاده از GIS,RS

A comparition between different methods for monitoring the cover of Arid Ein-khosh region using RS and GIS

بيابانزايي ، تخريب سرزمين وكاهش توان بيولوژيك درمناطق خشك و نيمه خشك مي باشد منطقه دهلران منطقه اي خشك درجنوب تا جنوب شرقي استان ايلام است و بيابان زايي از معضلات اصلي اين منطقه است هدف اصلي از اجراي اين تحقيق دوساله ارزيابي روند بيابان زايي منطقه در دوره 1990 -2006 است كه مبتني بر درك تغييرات پوشش گياهي است درانجام اين پژوهش از نمونه گيري هاي صحرايي تصاوير ماهواره اي و نرم افزارهاي مربوطه بهره گرفته شد. نمونه گيري هاي صحرايي درنقاط مختلفي از منطقه انجام شد. پردازشهاي لازم برروي تصاوير صورت گرفت وتك بانهاي شاخصهاي گياهي توليد شدند. بين داده هاي پوشش زميني و ارزش طيفي پيكسلهاي متناظر اناليز همبستگي صورت گرفت. بهترين نتيجه همبستگي براي شاخصهاي WDVI SAVI وPVI1 ، به دست امدتصاوير به دو كلاس پوشش و غيرپوشش تقسيم شدند. سپس دقت نقشه ها ارزيابي شد كه بالاترين دقت مربوط به نقشه حاصل از PVI1 بود دراين نقشه سطح پوشش گياهي براي سال 1990 معادل 469/268 هكتار به دست امد كه براي سال 2006 ، 173/475 هكتار تقليل يافت.

Desertification is land degradation and loss of biological productivity in arid, semi-arid areas. Dehloran in an arid region in south and southeastern ilam of Iran that suffer from severe types of desertification and land degradation. To assess the trend of desertification during 1990-2006 period, we conducted a two years study in the region. The basic tools that employed in this study were in fieled sampled data, TM and LISS III images plus with RS,GIS and statistical softwares. All needed processes were done on the image and vegetation indexes were developed. Filed sampling was carried out in different parts of the region. To determine the best vegetation index, we evaluated 83 vegetation indexes, among them PVI1, SAVI, WDVI proved to be the best for discriminating the plant cover on the images. Then through maximum likelihood algorithm, the images were classified in two categories:vegetation and non-vegetation. Kappa coefficients of resulted classification maps for theers selected indices calculated using confusion matrix, suggesting that PVI1 map is the most accurate. Finally, change detection map was obtained by subtracting images, showing that 11/10%of the study area exoerienced changes in vegetation cover from year 1990 to 2006. of all changed area 75/55 belonges to desertified class while just 12/21 of hange dareaindicates therestoration