*4.3. Utility and Limitations of InSAR for Interpretations of Coastal Landscape Change*

Within the study area, the analyses of four specific sites show the potential of InSAR as a tool to help explain local changes to the landscape. Although somewhat expected, coastline accretion and the upward movement of the terrain on the updrift side of a pair of groins built in 2012 (Figure 8 and site 1 in Figure 6a) stresses the reliability of the technique and its effectiveness in detecting VLM. It also illustrates that VLM estimates may include upward VLM related to crustal motion and the effect of sediment deposition and surface accretion. In a similar manner, in addition to subsidence, the finding of LOS velocities smaller than −1.0 cm/yr seaward of El Torno lagoon and near the coastline (Figure 11 and site 4 in Figure 6a), might be influenced by the erosive regime of the coastline. Indeed, Shirzaei et al. [55] indicate that, in dynamic landscapes, VLM estimates do include changes in surface elevation due to erosion or deposition and must be separated from other types of VLM. Thus, in addition to the locations mentioned above, areas next to sand beaches, such as the coastal town of Ciénaga and the westernmost extreme of the study site, the VLM velocities might be influenced by the removal or deposition of sand associated with erosion or accretion.

In contrast to the aforementioned locations, the locations of the time series for sites 2 and 3 are over 1 km landward from the coastline (see Figures 6a, 9 and 10), in sectors that were populated by abundant mangrove forest until the 1980s. Therefore, this area is prone to gas emission related to the oxidation of organic-rich soils. VLM along these locations ranges between −0.5 and −1 cm/yr (Figures 5, 9b and 10b). As the sites selected for the time-series profiles are not in close vicinity to the coastline, their velocity values are not influenced by the erosive regime of the coastline and can be considered to reflect subsidence only. Deltaic areas with similar subsidence rates to the values reported above have been described elsewhere [12–14].

Caveats of the InSAR methodology in this study also arise in vegetated areas when the surface backscattering properties due to growing vegetation change through time [66]. This factor is especially the case for wavelengths of less than 10 cm (e.g., Sentinel-1 A/B C-band) that interact with leaves and soft-stemmed vegetation. The constantly changing vegetation conditions result in temporal decorrelation and a reduction of the coherence level [66,67]. This issue was addressed by contrasting the outcomes of Sentinel-1A/B with ALOS, a satellite with a longer wavelength (L-band), which penetrates deeper in the canopy or ground surface [54]. When comparing LOS velocities for Sentinel-1A/B and ALOS ascending orbit, we found that, even though the latter provided absolute larger subsidence values, the trends for both platforms were aligned for most of the study area (Figure 6).

Given the caveats mentioned above, the absolute values of VLM should be interpreted with caution in the context of an application demanding high accuracy levels. However, despite the lack of temporal overlap between the Sentinel-1A/B and ALOS datasets, the outcomes of ALOS and Sentinel-1A/B were consistent in revealing areas prone to upward and downward movements with respect to a reference point in Ciénaga.
