Fusion of VNIR Optical and C-Band Polarimetric SAR Satellite Data for Accurate Detection of Temporal Changes in Vegetated Areas

Abstract

In this paper, we propose a processing chain jointly employing Sentinel-1 and Sentinel-2 data, aiming to monitor changes in the status of the vegetation cover by integrating the four 10 m visible and near-infrared (VNIR) bands with the three red-edge (RE) bands of Sentinel-2. The latter approximately span the gap between red and NIR bands (700 nm–800 nm), with bandwidths of 15/20 nm and 20 m pixel spacing. The RE bands are sharpened to 10 m, following the hyper-sharpening protocol, which holds, unlike pansharpening, when the sharpening band is not unique. The resulting 10 m fusion product may be integrated with polarimetric features calculated from the Interferometric Wide (IW) Ground Range Detected (GRD) product of Sentinel-1, available at 10 m pixel spacing, before the fused data are analyzed for change detection. A key point of the proposed scheme is that the fusion of optical and synthetic aperture radar (SAR) data is accomplished at level of change, through modulation of the optical change feature, namely the difference in normalized area over (reflectance) curve (NAOC), calculated from the sharpened RE bands, by the polarimetric SAR change feature, achieved as the temporal ratio of polarimetric features, where the latter is the pixel ratio between the co-polar and the cross-polar channels. Hyper-sharpening of Sentinel-2 RE bands, calculation of NAOC and modulation-based integration of Sentinel-1 polarimetric change features are applied to multitemporal datasets acquired before and after a fire event, over Mount Serra, in Italy. The optical change feature captures variations in the content of chlorophyll. The polarimetric SAR temporal change feature describes depolarization effects and changes in volumetric scattering of canopies. Their fusion shows an increased ability to highlight changes in vegetation status. In a performance comparison achieved by means of receiver operating characteristic (ROC) curves, the proposed change feature-based fusion approach surpasses a traditional area-based approach and the normalized burned ratio (NBR) index, which is widespread in the detection of burnt vegetation.

Increasing the Detection Accuracy of Bi-Temporal Changes via Speckle Whitening of Single-Look Complex Synthetic Aperture Radar Images

“…Though speckle correlation is always a drawback, this inconvenience is mitigated for despeckling filters operating in a shift-invariant multiscale domain [29]- [32]. Despeckling is expected to preserve texture features, like [33], used for classification [34] and for merging SAR and optical data [35]- [38].…”

Spatial Resolution Enhancement of Satellite Hyperspectral Data via Nested Hypersharpening With Sentinel-2 Multispectral Data

“…1), are provided at 20 m spatial resolution. Thus, MS hyper-sharpening may improve the spatial resolution of the Sentinel-2 bands acquired at 20 and 60 m and the derived spectral indexes, specifically those computed from the RE bands, suitable for studies on vegetation [17], [18].…”

Characterizing Dust and Biomass Burning Events from Sentinel-2 Imagery

“…The frequency of revisit is reduced to 5 days when both satellites are in formation. Thanks to their decametric spatial resolution, Sentinel-2 MS images have been used in forest surveillance [27], assessment of changes in land cover [28], and management of natural disasters [29]. In recent years, Sentinel-2 imagery has also played a key role in atmospheric studies, assessing both qualitative and quantitative analysis of aerosol and cloud optical properties.…”