An investigation on the suitability of ETM+ imagery for discrimination of land units of coastal area

Abstract
   In order to study the capability of ETM+ imagery for discrimination of land units of coastal areas, apportion of Khalidj-o-Omani district was selected. For these studies ETM+ images of Landsat 7 of April 15, 2000 were used. The radiometric and geometric errors of images were investigated and they were not so significant for this study. The scanned topographic map   (scale = 1:50000) were georeferenced using corners georeference technique, and then they were used for georeferencing the images by tiepoints georeference technique. Also the reference points were checked in the field by GPS. Finally the results showed 0.31 pixel accuracy of RMSE.
   In this study the aerial photographs of the study area were visually interpreted and checked by field study. Finally a digital map was produced (scale = 1:50000). In the next blush, analyzing of map were performed in the Geographical Information System (GIS) and the final phenomena (classes) were separated to use as the ground truth for assessment of classified color composites. Principal Component Analysis (PCA), rationing, Linear stretch, and density slicing were applied to ETM+ images. During field study, training samples for classifying were recognized. In addition more training samples were introduced based on the data developed from aerial photographs.
   The different band combinations were classified by maximum likelihood classification method. Optimum Index Factor (OIF) was used to select the best color composites. The statistical parameters of training samples including: average, standard deviation, and also overlapping of samples were studied. Three final classes were mapped, including: Avicina marina, Halocnemun strobilaceum, and Sea.
   Maximum likelihood classification was used in this study. The best color composite was IPVI, IR2, and PC1 with overall accuracy and Kappa coefficient 90% and 88.5% respectively. Final results showed that ETM+ data can be used to discriminate the coastal phenomena located in the Khalidj-o-Omani area.
Key words: Landsat 7, ETM+ Images, Bushehr, Khalidj-o-Omanian region, Asaloyeh, Nayband,Saline Area, Halophyte Plants, Remote Sensing, Maximum likelihood,  Mangrove Forests.

 Discrimination of Coastal Sandy Areas in the West of Bushehr Province by ETM+ Images Of Landsat 7

Abstract
  In order to study the usability of landsat 7 data for discrimination of coastal sandy areas in the west of Bushehr province an area were selected which exactly located in land number NH39-10 of sari K55-1 on topographical map (scale = 1:250000) of Genaweh. In the first step, by using of topographical map, aerial photographs, field study, and GPS, the total surface of the sandy area were calculated 1766 hectares and this map was digitized. Geographical error of ETM+ images of April, 2000 were corrected and georeferenced. Different processing including color composites study was applied and the best color composites were visually selected. In the next step the ground truth and color composites were overlaid. Processing of images showed the surface of sandy areas 1500 hectares that in comparison to the ground truth map the overall accuracy was 84.9 %. The accuracy was low because the pixel value of low vegetated areas was similar to coastal sandy areas.
   This accuracy for discrimination of coastal sandy areas from other phenomena such as sea is satisfying. In order to get better results for southern coastal areas, more image processing including rationing, OIF calculation to find the best image combinations, Jeffrey's matzos method, Linear stretch, density slicing, unsupervised and supervised classification (different classification methods) such as maximum likelihood is needed.

Key words: Coastal Sandy Iand Discrimination, West Province of Bushehr, Coastal  Areas,  ETM+data, Landsat 7.