Improved Wetland Classification Using Eight-Band High Resolution Satellite Imagery and a Hybrid Approach

Abstract

Although remote sensing technology has long been used in wetland inventory and monitoring, the accuracy and detail level of wetland maps derived with moderate resolution imagery and traditional techniques have been limited and often unsatisfactory. We explored and evaluated the utility of a newly launched high-resolution, eight-band satellite system (Worldview-2; WV2) for identifying and classifying freshwater deltaic wetland vegetation and aquatic habitats in the Selenga River Delta of Lake Baikal, Russia, using a hybrid approach and a novel application of Indicator Species Analysis (ISA). We achieved an overall classification accuracy of 86.5% (Kappa coefficient: 0.85) for 22 classes of aquatic and wetland habitats and found that additional metrics, such as the Normalized Difference Vegetation Index and image texture, were valuable for improving the overall classification OPEN ACCESS Remote Sens. 2014, 6 12188 accuracy and particularly for discriminating among certain habitat classes. Our analysis demonstrated that including WV2's four spectral bands from parts of the spectrum less commonly used in remote sensing analyses, along with the more traditional bandwidths, contributed to the increase in the overall classification accuracy by ~4% overall, but with considerable increases in our ability to discriminate certain communities. The coastal band improved differentiating open water and aquatic (i.e., vegetated) habitats, and the yellow, red-edge, and near-infrared 2 bands improved discrimination among different vegetated aquatic and terrestrial habitats. The use of ISA provided statistical rigor in developing associations between spectral classes and field-based data. Our analyses demonstrated the utility of a hybrid approach and the benefit of additional bands and metrics in providing the first spatially explicit mapping of a large and heterogeneous wetland system.

Random Forest Classification of Wetland Landcovers from Multi-Sensor Data in the Arid Region of Xinjiang, China

“…With the improved spatial resolution of remotely sensed imagery, the object-oriented strategy has also been proposed [22][23][24]. However, the accuracy and robustness of the existing classification methods are not yet satisfactory for wetland management, bearing big omission and commission errors due to the sparse yet variable vegetation and the hydrological fluctuation in the wetlands of arid areas [25,26]. The random forest (RF) algorithm, an integrative classifier, has shown to be able to achieve high classification accuracy even when applied to analyze data with stronger noise [27,28].…”

Using Remote Sensing and Gis Techniques to Detect Changes to the Prince Alfred Hamlet Conservation Area in the Western Cape, South Africa

“…The classification result and older topographic data were differenced and used to obtain the resulting change layer. Wetlands are one of the most difficult ecosystems to classify due to their high spatial heterogeneity and temporal variability, varying sizes and shapes, diversity of plants species, vegetation structures and types and water levels (Lane et al, 2014). There was no available data for wetlands within the CD: NGI topographic dataset, even though the topographic structure makes provision for this class.…”

“…The inclusion of image texture (Haralick et al 1973;Blaschke et al 2014) can assist with the classification process and has been found to increase classification accuracies (Aguilar et al, 2013;Franklin and Peddle, 1990). The texture measure GLCM (grey level co-occurrence matrix) contrast is effective in differentiating built-up areas from non-built-up areas (Duncan, 2013) and GLCM homogeneity is useful for wetland class discrimination (Lane et al, 2014). …”

“…Normalized Difference Vegetation Index and image texture are valuable for improving classification accuracy and for discriminating among certain habitat classes (Lane et al, 2014). The density-slicing tool was used to identify object feature threshold values for NDVI, as seen in Figure 5, where segments are shaded from blue to green.…”

Salt Marsh Monitoring in Jamaica Bay, New York from 2003 to 2013: A Decade of Change from Restoration to Hurricane Sandy

“…The Worldview-2 sensor collects eight spectral bands including the Coastal Blue, Blue, Green, Yellow, Red, Red Edge, Near Infrared 1 (NIR1), and Near Infrared 2 (NIR2). The Coastal Blue, Yellow, Red Edge, and NIR2 spectral bands of Worldview-2 have been shown to increase the accuracy of wetland vegetation classification [17]. This study used Quickbird-2 imagery data acquired on 10 September 2003 and 9 September 2008, and Worldview-2 data acquired on 15 September 2012 and 19 September 2013.…”