| Abstract |
Extraction of water bodies is one of the most important part of water resources management and for more than two decades many researches of detecting water related to remote sensing applied. the accuracy of monitoring water is an important and fundamental application in remote sensing. different methods have been improved to extract water features by using multispectral images. The main areas of interest for remote sensing research always had been concerned with studies related to environmental, especially detecting water bodies. Many methods which using near- an infrared and visible band of the electromagnetic spectrum had been already improved to monitor water surface. Moreover, the using of an image obtained in visible and infrared spectrum allows quality detecting of water bodies. Nevertheless, retrieval of water boundaries and mapping water surface with optical sensors is still quite demanding. Therefore, the microwave data could be the perfect complement to data obtained with passive optical sensors to monitor and detect and aquatic environment especially water surface.
The researchers use several methods to classify images which each one of them has an accuracy and efficiency by comparing them to each other to find an appropriate method to classify the remote sensing data. The aim of this study is to compare the accuracy of 6 classification methods for monitoring water surface in Tibet region. For this purpose, the images of Sentinel-2 satellite for optical method and Sentinel-1 satellite for Microwave method in 2016 were used. Traditional classification approaches are all pixel-based, and do not utilize the spatial information within an object which is an important source of information to image classification. In the pixel-based classification, the spectral angle mapper was used to classify the images. For supervised classification this study used the rule of maximum likelihood classification and also form signature file of sample from the area of each class of the four classes and the results showed that the calculated area from this method is (1736.79864 Square Kilometers). On the other hand, for the unsupervised classification using the two algorithms of ISODATA and K-Mean the results showed that the calculated area from this method is (1645.69206 Square Kilometers) and (1634.290647 Square Kilometers) Respectively. While the Object-based classification is a method that can achieve the same objective based on the segmentation of spectral bands of the image creating homogeneous polygons with regard to spatial or spectral characteristics. The segmentation algorithm does not solely rely on the single pixel value, but also on shape, texture, and pixel spatial continuity. The object based Classification was done by a nearest neighbor classifier, While the object based thresholding used NDWI and the results showed that the calculated area from this two method are (1748.261088 Square Kilometers) and (1764.683151 Square Kilometers) Respectively. Also different satellite-derived indexes are including Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Modified NDWI (MNDWI). As the NDVI uses the red band from the visible bands and near-infrared bands, the results showed that the calculated area from
iii
this method is (1741.620482 Square Kilometers), while the NDWI uses the green and Near-Infrared bands to enhance the presence of such features. The selection of these bands was done to (1) maximize the reflectance of water features by using green light wavelengths; (2) minimize the low reflectance of NIR by water features, and the result showed that the calculated area from this method is (1746.73507 Square Kilometers). And for the MNDWI was developed to modify the NDWI in detecting water features for water regions with backgrounds dominated by built-up land areas by using Short-wave Infrared (SWIR) band instead to Near-infrared (NIR) band as the water absorbs more light in the region of SWIR than in the region of NIR, but the two band used have different spatial resolution of 10m for green and 20m for SWIR, so we applied pan-sharpening method which is Intensity Hue Saturation (IHS) and the results showed that the calculated area from this method is (1752.69915 Square Kilometers). Also, this study presents a methodology to detect and monitor surface water with Sentinel-1 Synthetic Aperture Radar (SAR) data of Thresholding method where Any pixel (x,y) for which f(x,y) < threshold value which in this study is (-17) is considered as belonging to the water class, otherwise, it belongs to the background class and the results showed that the calculated area from this method is (1565.57894 Square Kilometers). And the total calculated area which is used to compare to all the results from all the different method is (1750.109264 Square Kilometers) from google earth image. Finally, from all the results of classification it shows that the object-based classification method gives more accurate and satisfying results. |
| Keywords |
The Normalized Difference Water Index (NDWI), Normalized Difference Vegetation
Index (NDVI), The Modified Normalized Difference Water Index (MNDWI), Pixel based
classification, Object-based classification, Thresholding method Synthetic Aperture Radar
(SAR), Sentinel-1 and Sentinel-2. |
