Coastal zones, defined as low-elevation areas at the land-water interface, are crucial for various ecological and societal functions. They support a significant portion of the global population and harbor vital wetland ecosystems that provide numerous benefits, including shoreline protection, storm buffering, and carbon sequestration. However, these zones are highly vulnerable to the impacts of climate change, particularly accelerated sea-level rise (SLR). SLR, exacerbated by factors such as ocean currents, coast morphology, and vertical land motion (VLM – subsidence or uplift), poses significant threats, including coastal flooding, erosion, and saltwater intrusion. Subsidence, driven by factors like fluid extraction and sediment compaction, significantly influences relative sea levels, amplifying the risks associated with SLR. The US Atlantic coast, one of the most populous in the US, is particularly vulnerable due to widespread land subsidence from sediment compaction, groundwater extraction, and glacial isostatic adjustment. This coast is home to a diverse coastal ecosystem, including extensive wetlands. Increased coastal flooding due to relative SLR has disrupted economic activities and caused prolonged wetland inundation, increasing the overall coastal vulnerability. Current assessments of coastal vulnerability are inadequate due to sparse and localized VLM measurements, highlighting the need for improved, high-resolution data on VLM to accurately assess coastal risk.

Existing literature extensively documents the vulnerability of coastal environments to sea-level rise, emphasizing the spatial variability of this vulnerability due to local land subsidence. However, a scarcity of high-resolution observations and models of coastal subsidence hinders accurate vulnerability assessments. Previous studies on wetland vulnerability often rely on sparse, point-wise measurements of vertical land motion (VLM), underestimating the true extent of subsidence and its impact on wetland stability. While field observations provide accurate in-situ data, they are impractical for large-scale monitoring. Comparisons between wetland accretion data from ground-based observations and relative SLR often fail to account for both shallow and deep subsidence, leading to underestimation of coastal wetland vulnerability. Interferometric synthetic aperture radar (InSAR) emerges as a valuable tool to overcome these limitations, offering high spatiotemporal resolution for elevation change monitoring across various land cover types, including dynamic wetlands. InSAR data offers cost-effective and up-to-date data, improving the accuracy and spatial density of VLM measurements.

This study utilized a combination of Sentinel-1A/B and ALOS satellite data from 2007 to 2020, along with 173 Global Navigation Satellite System (GNSS) observations over the same period, to create a high-resolution VLM rate map for the US Atlantic coast. The data covered an area extending approximately 100 km inland from the coast. The SAR datasets included multiple frames acquired in ascending orbit geometry. Data processing involved a multitemporal wavelet-based InSAR (WabInSAR) algorithm to generate 3D line-of-sight (LOS) velocities. A unified weighted least-squares joint optimization model integrated the LOS velocities with GNSS observations to determine the 3D deformation field. Model validation involved analyzing the standard deviation (SD) of InSAR pixels (precision) and comparing InSAR VLM with independent GNSS data (accuracy). Horizontal velocities showed the relative motion of the North American plate, consistent with previous findings. To assess uncertainties in sea level projections, the researchers compared their InSAR-derived VLM rates with those used in the IPCC Sixth Assessment Report (AR6) at 12 tide gauge stations. The study also utilized the 2019 USGS National Land Cover (NLC) map to estimate the exposure of different coastal ecosystems (wetlands, forests, agricultural areas, developed areas) to subsidence by interpolating VLM rates over NLC pixels. Wetland vulnerability was assessed using a modified vertical resilience (VR) index, incorporating VLM and SLR, and accounting for uncertainties through a 90% confidence interval. Low- and high-elevation marshes were differentiated based on elevation normalized to mean high water (MHW). The study also included a validation against previously compiled accretion rates from a limited number of discrete points and data from SET-MH stations.

The study generated a spatially semi-continuous VLM rate map for the US Atlantic coast at a resolution of ~50 m and mm-level precision. The map revealed widespread subsidence exceeding 3 mm per year across most coastal areas, including wetlands, forests, agricultural areas, and developed regions. Ninety percent of pixels showed subsidence, highlighting a broad-scale pattern. Major cities along the coast exhibited significant subsidence, with rates exceeding 5 mm per year in some locations (Charleston, Brunswick, Chesapeake Bay). A comparison with IPCC AR6 VLM projections showed that the IPCC underestimated VLM at seven tide gauge stations and overestimated it at one, with four stations within a 20% error range. The analysis of land cover exposure to subsidence revealed that wetlands, forests, cultivated crops, and developed regions exhibited the most significant exposure. Agricultural lands showed some of the highest subsidence rates, exceeding 11.8 mm per year in certain areas. The vulnerability assessment of coastal marshes indicated that 58-100% of marshes are losing elevation relative to SLR, significantly higher than previous estimates (43-53%). This discrepancy stems primarily from the inclusion of subsidence in the VLM data. The findings highlight the significant impact of subsidence on the stability and future of coastal marshes.

The findings of this study demonstrate the critical role of land subsidence in exacerbating the vulnerability of the US Atlantic coast to SLR. Subsidence can amplify the impacts of SLR, leading to increased coastal flooding, saltwater intrusion, and the decline of coastal ecosystems. The spatially explicit, high-resolution VLM data provide crucial insights into the regional variability of relative SLR, which is essential for accurate flood risk assessments and coastal planning. The substantial underestimation of marsh vulnerability in previous studies underscores the need to incorporate subsidence in future vulnerability assessments. The high rates of subsidence observed in agricultural lands and forests have significant implications for these sectors and their associated economic activities. The ongoing subsidence will amplify future inundation, especially in densely populated coastal cities, increasing the frequency and severity of flooding. The loss of coastal forests due to saltwater intrusion is also accelerated by subsidence, leading to ecological shifts and the loss of stored carbon.

This study provides unprecedented high-resolution data on VLM along the US Atlantic coast, revealing widespread subsidence that significantly exacerbates the vulnerability of coastal ecosystems and human populations to SLR. The findings highlight the critical need for incorporating high-resolution VLM data into coastal vulnerability assessments and sea-level rise projections to improve the accuracy of inundation models and inform effective coastal adaptation strategies. Future research could explore the various drivers of subsidence in more detail and develop improved models to project future subsidence rates under different climate change scenarios.

While this study provides high-resolution VLM data, it relies on satellite observations and may not capture localized variations in subsidence. The study primarily focuses on the current rates of subsidence and SLR and doesn't explicitly incorporate future accelerations in either rate. The accuracy of the VLM estimates is influenced by the density and distribution of GNSS stations, and data uncertainties are reflected in the confidence intervals reported. Furthermore, the validation with SET-MH stations was limited due to spatial mismatch.