Sea level rise and land subsidence are contributing to a rapid loss of coastline along the United States. Due to climate change and sea level rise, extreme weather events become more frequent and more hazardous which in turn increases flood risk (Lim et al., 2018; Catalao et al., 2020). These extreme weather events are eroding the coastlines further. Marshlands are essential to coastal ecosystems and are home to a variety of wildlife species (Scotch Bonnet Marsh Enhancement Project). They also provide a physical barrier from extreme weather events. It is important for the wildlife and coastal communities that these marshlands remain intact. The United States Army Corp of Engineers is tasked with mitigating these risks. They are using dredged material to build up endangered marshlands along the coast. The goal of this project is to analyze the effectiveness of these sediment enrichment sites using Interferometrically processed Synthetic Aperture Radar (InSAR) data to quantify the vertical displacement at the sediment enrichment sites. Light Detection and Ranging (LiDAR), multispectral data, and in-situ measurements are used to verify the elevation change.
Interferometrically processed synthetic aperture data has been proven reliable in many instances to help quantify the vertical displacement of coastal areas. In previous studies, InSAR data agrees with both in-situ GPS measurements and Global Navigation Satellite System (GNSS) measurements (Buzzanga et al., 2024; Zhong et al., 2022; Bui et al., 2021). It was also shown that InSAR data can show irregular deformation patterns due to local activities which is useful for this project that focuses on sediment enrichment sites (Buzzanga et al., 2024). Satellite imagery from Sentinel 1 provides frequently collected data that covers a large area of land and is precise in the vertical direction down to the centimeter to millimeter scale (Catalao et al., 2020, Bui et al., 2021). This is extremely important because other data such as in-situ measurements are much lower in spatial resolution and can be difficult to collect based on geographic features. UAV data of coastal areas depend on weather conditions, are expensive, and require more on-site presence. Sentinel 1 data on the other hand is easily accessible through the Alaska Satellite Facility Vertex. There are two methods of processing InSAR data to create time series of vertical displacement that are frequently employed: Small Baseline Subset (SBAS) and Persistent Scatterer (PS). The SBAS method uses many interferograms with small temporal and perpendicular baselines to create a time series of vertical displacement. The redundancy in interferograms is used to reduce the signal to noise ratio and results in an extremely precise vertical displacement signal (Bui et al., 2021; Zhong et al., 2022). The PS method optimizes cells with single point scatters; therefore, this method is best used in urban areas. The advantage of using the PS method is the lower computation burden since it uses fewer data, however the results are less precise (Bui et al., 2021, Zhong et al., 2022). This project will be focussing on SBAS considering radar data of coastal marshlands results in a lot of decorrelation and noise. In both methods, following the coherent pixel selection, the signal must be corrected for atmospheric effects, orbital errors and then unwrapped (Bui et al.). Atmospheric effects are due to propagation delay of the radar signal within the atmosphere. Unwrapping is the process of transforming the raw radar data from oscillations of 2π to elevation values (Buzzanga et al., 2024). Previous studies suggest that paired remote sensing techniques such as InSAR, LiDAR, and hyperspectral data are complementary (Zhong et al.). LiDAR data has a much smaller spatial resolution than the Sentinel 1 data which means it has the ability to show sub-pixel variation in vertical displacement (Zhong et al.). This is useful because the sediment enrichment sites are relatively small. Hyperspectral data can also provide additional information about elevation through biomass characteristics. Miller et al. found that spartina alterniflora, the most common species of biomass in marshlands around the sediment enrichment sites, has a parabolic relationship with elevation (Miller et al.). It was found that the biomass struggled when the elevation was either too high or too low (Miller et al.). Using multispectral imaging of the sediment enrichment sites, the relationship between elevation and biomass can be employed to validate the InSAR and LiDAR findings. While many studies have been performed using SAR data from Sentinel 1, the capabilities of the data have not been fully explored. This project aims to evaluate the capability of Sentinel 1 SAR data to be used in coastal marshlands in capturing vertical displacement of sediment enrichment sites over time. It is unknown how usable the InSAR data will be in these areas due to the high risk of decorrelation. Previous studies have been able to study coastlines with InSAR, however the presence of water poses a unique problem at each site and requires cautious analysis. Additionally, it is undetermined how well the sediment placements are performing over time. This project will show how much the sediment subsides after placement.
Using interferometrically processed Sentinel 1 Synthetic Aperture Radar (InSAR) data, can we quantify vertical displacement at sediment enrichment sites? We expect to see abrupt increases in elevation corresponding to sediment deposition, abrupt decreases in elevation due to extremely erosional events, and a gradual decrease after sediment deposition.
C-Band SAR data was acquired from Sentinel 1, a sun synchronous satellite with a revisit time of 12 days, a spatial resolution of 5 m x 20 m, and a swath width of 250 km (Podest). The spatial resolution may be resampled with a high resolution digital elevation model (DEM). C-Band SAR data has a wavelength of 5 cm which can penetrate 1-5 cm into the ground (Podest). SAR imagery captures complex signals which include information of the phase and amplitude of microwave signals emitted at the sensor and reflected back. The phase arrival contains information about distance from the sensor to the reflector (the ground) which allows for the quantification of vertical displacement between SAR image captures. The vertical displacement data collected over time will be used to create a time series of displacement corresponding to an area of interest (AOI). The SAR images used in this study were selected to minimize the temporal and perpendicular baselines.
SAR images, called Single Look Captures (SLC), can be downloaded and accessed using the open source packages asf_search and MintPy (Miami INsar Time-series software in Python) in Jupyter Notebook. The SAR images are fetched from the Alaska Satellite Facility (ASF), a data processing facility that provides open access to remote sensing data. For this project the InSAR products were generated using ASF Vertex, the graphical search interface for finding SAR images. The SLC’s were chosen based on their location, start date, end date, processing level, beammode, and flight direction. For the Scotch Bonnet/Maurice River study area, the specifications are as follows: the location was set by point at Maurice River, the date range was October 2022 to December 2024, the processing level was L1 SLC, the beam mode was Interferometric Wide Swath (IW), and the flight direction was ascending. For the Louisiana coastlines study areas, the specifications are as follows: the date range was January 2016 to June 2019, the beam mode was IW, and the flight direction was ascending.
Once this is complete, a data frame was created compiling the SLC’s and their baseline information. From this, a stack of SLC’s was created from the baseline results and find usable image pairs that satisfy the specified temporal and perpendicular baseline thresholds. Following this, the image pairs were clipped based on their overlap so they can be used to create an interferogram.
Using the Small Baseline Subset technique with MintPy interferograms will be created from the image paris. SBAS uses multiple SAR images which have relatively small temporal and perpendicular baselines in order to increase coherence (Bui et al., 2021). The process of SBAS includes SAR image coregistration, filtering, phase unwrapping, geocoding, and stacking of the interferograms (Hu et al., 2019). After this, the SBAS results go through a time-series analysis in which we select coherent pixels, and correct for atmospheric effects and orbital errors (Bui et al., 2021). In addition, in coastal areas, the water content can result in phase inconsistencies. These inconsistencies will be corrected using a phase closure bias. The ASF Application Programming Interface (API) will be used to generate the time-series of displacement data. This interface reduces the computation burden on local computers. After this, a time series of deformation will be plotted and the average deformation rate after sediment enrichment events will be calculated. In this stage of the project however, the SBAS tool on Vertex was used to generate the InSAR products. Once the Scotch Bonnet/Maurice River SLC’s were generated on Vertex, they were put into the SBAS tool with the following parameters: the temporal baseline was 24 days, the minimum overlap was 50%, no water mask was applied. This resulted in 121 InSAR products to be downloaded into Jupyter Notebooks. Once the SLC’s were generated, they were put into the SBAS tool on Vertex with the same parameters as the Scotch Bonnet site. This resulted in 76 InSAR products to be downloaded into Jupyter Notebooks.
Three Local DEM products of Scotch Bonnet were generated from LiDAR fly-overs by the Joint Airborne Lidar Bathymetry Technical Center of Expertise (JALBTX) for 2017, 2020, and 2024. Once downloaded into ArcGIS Pro, a mapping software, the DEM’s were processed to have the same geographic projection and coordinate system. Once complete, 3 different maps were created for each interval of time: 2017-2020, 2020-2024, and 2017-2024. These maps were exported as .tif files to be used in Jupyter notebooks in order to create a time series of deformation of the study area. This time series can help validate the InSAR deformation time series.
Multispectral products will be downloaded in ArcGIS Pro using Planet Labs, an Earth data archive. The multispectral images will be analyzed for biomass index which may be used to analyze the elevation of the marshlands considering that in marshlands across the United States, the local sea grasses exhibit a parabolic relationship with elevation.
Current sediment enrichment volumes were obtained for the Louisiana coastline sites. In future work, sediment enrichment volumes will be obtained for the Scotch Bonnet and Maurice River sites as well. This known volume of sediment will be used in the training and validation stage of the elevation change analysis
This project aims to quantify the vertical displacement of the USACE sediment enrichment sites. Using a linear regression model, the vertical displacement over time will be quantified by calculating the mean displacement for each polygon and analyzing the change over time. The independent variable for the linear regression is time and the dependent variable is vertical displacement. The linear regression analysis will allow us to quantify the change in vertical displacement over time. This time-series analysis will then be compared to the known volumes of sediment placed during sediment enrichment events by the United States Army Corp of Engineers (USACE). The known volume of sediment and the known area of the AOI allow us to determine the vertical displacement. This vertical displacement will be compared to the vertical displacement calculated through InSAR analysis. In addition, the InSAR vertical displacement time series will be compared to a sediment compaction model used by USACE. Both comparisons will analyze the differences in precision and bias.
3 DEM’s were created for the Scotch Bonnet site; they are shown in Figures 1-3. Currently, the sediment enrichment site can not be seen in these images since placement was in October 2024 and the most recent DEM corresponds to April 2024. However, the subsidence of the marshland prior to sediment enrichment is able to be seen. Figure 4 is a DEM difference map showing the difference in elevation from 2020 to 2024. It is easily seen that the channels of water are subsiding more than the rest of the marshland. It is probable this is due to the rise in relative sea level. The sediment enrichment site was placed at the end of the channel close to the road that runs diagonally from the top left to the bottom right. This area poses a significant risk of flooding and inundation.
Figure 1. DEM of Scotch Bonnet in 2017 acquired from LiDAR imagery (NOAA)
Figure 2. DEM of Scotch Bonnet in 2020 acquired from LiDAR imagery (NOAA)
Figure 3. DEM of Scotch Bonnet in 2024 acquired from LiDAR imagery (US Army Corp of Engineers, 2024)
Figure 4. Scotch Bonnet marshland elevation difference in feet between 2020 and 2024 using LiDAR imagery (US Army Corp of Engineers, 2024; NOAA)
The data from the sediment enrichment sites in Louisiana are shown in Table 1. The average vertical displacement in yards was calculated for each placement using the known dredged material volume and the known area of placement.
Table 1. Sediment Enrichment data collected for 17 areas along the Louisiana coastline including the calculated value for average vertical displacement after the placement (Ayala 2024)
The results in Figure 4 are in agreement with previous findings. The US coastlines are subsiding and relative sea level is drastically increasing which is threatening the integrity of the marshlands. This makes the U.S. coastlines more vulnerable to flooding, storm surges, and erosion due to extreme weather events. In addition, the inundation of marshlands causes extreme biodiversity loss. In some areas, the vertical change from 2020 to 2024 is multiple feet. It is assumed that some of this vertical displacement is due to tide changes between the two LiDAR images. However, it can be said that the regions around the channels in the marshlands are subsiding and eroding the most. The results mentioned in this paper represent the first step in a much larger objective. The next steps in the research include working directly with the InSAR products to build a time series and running the linear regression model. Simultaneously, I will be accessing the multispectral data to decide whether or not I will be able to use the biomass values to estimate elevation changes over time.
Relative sea level rise and extreme weather events pose significant danger to the U.S. coastlines. Marshlands act as a natural barrier to storm surges and provide a habitat for many species. The US Army Corp of Engineers (USACE) is tasked with mitigating the damages to these threatened areas. Sentinel 1 SAR data capabilities have not been fully explored. This project is the starting point for using interferometrically processed InSAR data to quantify vertical displacement at sediment enrichment sites placed by the USACE. Complimentary data such as DEM and multispectral data are being collected to verify the vertical displacement trends calculated from InSAR analysis.
Ayala, M. (2024). Measuring the Accuracy of InSAR Vertical Displacement in the Detection of Discrete Coastal Change. SUNY College of Environmental Science and Forestry. Bui, L. K., Le, P. V. V., Dao, P. D., Long, N. Q., Pham, H. V., Tran, H. H., & Xie, L. (2021). Recent land deformation detected by Sentinel-1A InSAR data (2016–2020) over Hanoi, Vietnam, and the relationship with groundwater level change. GIScience & Remote Sensing, 58(2), 161–179. https://doi.org/10.1080/15481603.2020.1868198 Buzzanga, B., Bekaert, D. P. S., Hamlington, B. D., & Sangha, S. S. (n.d.). Toward Sustained Monitoring of Subsidence at the Coast Using InSAR and GPS: An Application in Hampton Roads, Virginia. Geophysical Research Letters, 47(18). https://doi.org/10.1029/2020gl090013 Carter, E. (2023). Evaluating the Effectiveness of Sediment Enrichment Against Multiple Drivers of Coastal Erosion Using Satellite Data. Catalao, J., Raju, D., & Nico, G. (2020). INSAR maps of land subsidence and sea level scenarios to quantify the flood inundation risk in coastal cities: the case of Singapore. Remote Sensing, 12(2), 296. https://doi.org/10.3390/rs12020296 Hu, B., Chen, J., & Zhang, X. (2019). Monitoring the Land Subsidence Area in a Coastal Urban Area with InSAR and GNSS. Sensors, 19(14), 3181. https://doi.org/10.3390/s19143181 Lim, Y., Schubert, S. D., Kovach, R., Molod, A. M., & Pawson, S. (2018). The roles of climate change and climate variability in the 2017 Atlantic Hurricane season. Scientific Reports, 8(1). https://doi.org/10.1038/s41598-018-34343-5 Miller, G. J., Morris, J. T., & Wang, C. (2019). Estimating aboveground biomass and its spatial distribution in coastal wetlands utilizing planet multispectral imagery. Remote Sensing, 11(17), 2020. https://doi.org/10.3390/rs11172020 NOAA: Data Access Viewer. (n.d.). NOAA: Data Access Viewer. https://coast.noaa.gov/dataviewer/#/lidar/search/-8329209.207436156,4722652.23797108,-8318164.056848949,4733678.279301216 Podest, E. & National Aeronautics and Space Administration. (n.d.). Basics of Synthetic Aperture Radar. (National Aeronautics and Space Administration), Applied Remote Sensing Training. Retrieved from https://appliedsciences.nasa.gov/sites/default/files/Session1-SAR-English.pdf Scotch Bonnet Marsh Enhancement Project. (n.d.). US Army Corp of Engineers. Retrieved December 15, 2024, from https://www.nap.usace.army.mil/Missions/Civil-Works/Coastal-Dredging-Beneficial-Use/Scotch-Bonnet-Project/ Time Series Analysis with HyP3 and MintPy. Retrieved from https://hyp3-docs.asf.alaska.edu/tutorials/ US Army Corp of Engineers. (2024). 20240417_Scothc Bonnet_DEM_SPC [Dataset]. Yunjun, Z., Fattahi, H.. (2024). InSAR Time Series Analysis: MintPy + ARIA GUNW products. Retrieved from https://nbviewer.org/github/insarlab/MintPy-tutorial/blob/main/workflows/smallbaselineApp_aria.ipynb Zhong, W., Chu, T., Tissot, P., Wu, Z., Chen, J., & Zhang, H. (2022). Integrated coastal subsidence analysis using InSAR, LiDAR, and land cover data. Remote Sensing of Environment, 282, 113297. https://doi.org/10.1016/j.rse.2022.113297