Satellite Used | Study area | Methods | Advantages |
Landsat 5 (TM) Landsat 7 (ETM+) Landsat 8 (OLI) Resolution: 30 m [35] | Lake Urmia located in the northwest of Iran | Normalized Difference Water Index (NDWI) and Principal Component Analysis (PCA) | NDWI performed slightly better than the NDWI-PCs. NDWI-PCs have an advantage over the NDWI, that it detects the surface water changes of two and three different times simultaneously by applying a single threshold to the selected PC. |
Landsat TM ETM+ Resolution: 30 m [36] | Rift Valley lakes in Kenya | Water Index (WI) using Tasseled Cap Wetness (TCW) index and NDWI | WI detected the shorelines with an accuracy of 98.4%, which was 22.3% higher than the TCW, and 43.2% more accurate than the NDWI. |
Landsat 4 - 5 Resolution: 30 m [37] | Huanghe river delta, China | B2 + B3 > B4 + B5 B4 < 60 B4 > B5 and B5 < B2 | The decision tree algorithm failed to extract small water bodies at scales below the sensor resolution. |
Landsat TM and ETM+ Resolution: 30 m [38] | Hebei, Jiangxi, Ningxia, China | A watershed segmentation method is adopted to detect mixed water pixels at the edges of lakes or rivers | Automatic method for extracting rivers and lakes (AMERL) successfully extracted most of the narrow rivers and lakes. |
SPOT 4 Resolution: 10 m [39] | Jiangning county of Jiangsu, China | Decision Tree (DT) model based on both spectral and auxiliary information of Digital Elevation Model (DEM) and Slope (DTDS). | It is difficult to extract water bodies effectively by applying a single technique due to effects of shadows. Unsupervised classification yields result with low accuracy. |
Landsat 5 TM Resolution: 30 m [40] | Murrumbidgee, Wagga Wagga, Australia | Single band density slicing and Maximum Likelihood (MXL). | MXL proved to be more accurate than density slicing to detect water bodies. Density slicing yielded a less speckled output image as compared to MXL. |
Landsat 5 TM Resolution: 30 m [41] | Denmark, Switzerland, Ethiopia, South Africa, New Zealand. | AWEISh = Blue + 0.25*Green − 1.5 *(NIR + SWIR1) − 0.25 *SWIR2 AWEINSh = 4 *(Green − SWIR1) − (0.25 * NIR + 2.75 * SWIR2) | Automatic Water Extraction Index (AWEI) successfully extracted surface water with high accuracy, particularly in mountainous regions where hills cast shadows on background surfaces and in urban areas with complex land cover. It is a simple technique to extract water in different environmental conditions. |
MODIS (250 - 500 m) and ASTER (15 - 90 m) [42] | Bihor, Romania | Threshold method and supervised classification | This approach is useful in providing information about water classification from different resolution data. |
ASTER (15 m) and MODIS [43] | Koros basin, Romanian-Humgarian border |
Band 1 (0.66 μm) - 250 m Band 2 (0.87 μm) - 250 m | Cloud shadow and water pixels are not completely separated out. |
ASTER Resolution: 15 m [44] | Beijing, China | 4 segmentation levels were created for differentiation between water, vegetation and non-vegetation | Accuracy of object-oriented classification is higher than the accuracy of MXL classification. |