Nowadays the use of information and techniques in remote sensing has made it possible to have a remarkable development in acquiring information related to different biological phenomena as well as their optimal management. Remote sensing data along with other sources of information when available to researchers and managers can be used in environmental development, as well as in natural resources plannings. One of the salient and conspicuous objectives in this regard is soil conservation. Recognizing the regions affected by critical soil erosion reveals undisputable priorities for soil conservation. This can be achived in either one pf two ways. The first is making use of traditional methods of field evaluation which is time consuming and expensive while the second is the use of satellite images that is economical due to low costs as well as quickness. In the present research it is tried, to recognize and map erosion in Taleghan basin in Tehran province by using Image processing technics. Satellite images of LANDSAT 7 (ETM sensor) as well as Cosmos satellite (KFA — 1000 camera) were used. The research is devided into two sections: The first consists of” Image Fusion” technic so as to achieve the better images for recognizing then categorizing erosion. The second section comprises of categorizing then mapping the eroded regions. In the first section, five different methods were used for image fusion as follows: making color composite from principal components (PC-1,2 and 3); making color composite from PC
—2, and 3 as well as Panchromatic band of ETM sensor; providing images by IHS system and substitution of Intensity by ETM Panchromatic band then retransfering it to RGB system; multiplying Panchromatic band by other ETM sensor bands; finally, multiplying the intensity of Cosmos image by ETM sensor bands. Assessments at this stage indicate that the image obtained from fusion of Cosmos image and ETM band, posesses the best resolution and thus is the best image for recognizing erosion. Also these assessments indicate that the RGB image transfered from IHS presents the most clearness for recognition of phenomena. In the second part of the research, erosion is categorized by using images obtained from the fusion of ETM bands and Cosmos image. For this purpose, a map of ground turth from eroded regions is provided by using Cosmos image as well as by visual interpretation and field observations. The image is then categorized by using four different algorithms with “Minimum Distance To mean” showing the best results. Following categorization by ground truth, the accuracy of categorized image is designated by “ Overall Accuracy” index. Measurements indicate an approximate 80 percent accuracy for the categorization. In the last stage of the research, “ Spectral Signature” is provided from eroded regions indicating that the eroded regions in bands 3, 4, and 5 possess the most spectral reflections.
)