ارزيابي داده هاي سنجنده ASTERجهت تهيه نقشه تاج پوشش جنگل ﴿مطالعه موردي : جنگل هاي استان لرستان﴾

چكيده انگليسي :

Evaluation of the Potential of Aster Data for Mapping of Forest Vegetation Cover Case study Lorestan Forest Mehrdad Mahshanian Isfahani Mahshanian@of iut ac ir Date of Submission September 19 2011 Department of Natural Resources Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Languege FarsiSupervisor Seyed Jamaledin Khajedin Khajedin@cc iut ac irAbstractThe purpose of this study is to investigate the potential of ASTER images for mapping forest vegetationcover in Zagros s forests The study area is a sheet of topographic map scale 1 50 000 of Lorestanprovince with an area covering about 64 000 hectares In order to achieve our goal in this study a scene ofASTER sensor with dimensions 60 km 60 km Was selected The geometric correction of the image wasdone with nearest neighbor 1st degree polynomial method using 30 ground control points with root meansquare error of 0 46 pixel in ILWIS 3 4 software environment Different image analysis techniques wereapplied to the image including PCA analysis and preparation of various types and vegetation indices Slop based Distance based The stratified random sampling was used for ground sampling using the complexplot of 10m 10 m around the sampling points with a radius of 30 m from the centre of each point Investigation of correlation tables of nine sensor s bands and ground truth maps showed that band 4 had thehighest correlation 64 and band 3 had the lowest correlation 30 Also band 3 had the lowest correlationwith other bands Supervised unsupervised and hybrid classification approaches were used for creating land usemap of the study area using best false color composites FCCs In this analysis landuse map was obtained insix categories agriculture lands forest and rangeland water bodies bare ground rock and built up areas Theresults indicated that classification process provided overall accuracy and kappa values of 82 93 and 0 83 respectively Forward stepwised regression was used for statistical analysis and model creation In this test the dependent variable Y was the data from the ground truth map and the independentvariables X included the DN from the image bands PCAs and vegetation indices Data regression analysisincluded five groups major bands PCA components bands and PCA components vegetation indices and also combinations of bands PCA components and vegetation indices Among suggested models 17models were selected based on higher R2adj and lower Mallow coefficients repectively Final model indicatedthat bands 4 and 9 and MSAVI and NDVI indices resulted from various band components had the highestfrequency in models formula and the presence of band 3 in half of the indices indicated the vital role of thisband in model preparation process In order to select the most accurate model indicating vegetation condition classified maps were compared with the ground truth control maps When using Spectral raw bands theoverall accuracy and kappa values was the lowest about 56 08 and 0 34 respectively And The highestoverall accuracy was achieved when using Vegetation indicec 73 44 and 0 6 At the end Finallanduse cover map with overall accuracy and kappa values of 70 94 and 0 6 was obtained Comparingthe results of this study and Lorestan s forest maps showed a low compliance The highest overall accuracyand kappa coefficient was 22 7 0 5 in the best conditions The results also showed the importance of choosing asuitable method for field sampling in spare vegetation cover According to the regression results the sampling systematic transect which was used in the Lorestan s forest mapping was not an appropriate technique forproducing high accuracy maps In contrast the relationships between field and image data using stratifiedrandom sampling method was much higher The results of this study showed that despite the different spectraleffects of vegetation and soil background which reduced the accuracy of classification and forest covermapping using ASTER sensor data could manifest variati

ماهشانيان اصفهاني، مهرداد

ارزيابي داده هاي سنجنده ASTERجهت تهيه نقشه تاج پوشش جنگل ﴿مطالعه موردي : جنگل هاي استان لرستان﴾

سنجش از دور , شاخص هاي گياهي , نقشه كاربري اراضي , زاگرس