ICESat-2 photons: strengthening shallow-water bathymetry

The Seabed 2030 initiative, anticipated to be a game-changer in ocean sciences and marine navigation, will map the ocean floor by 2030. Shallow-water bathymetry is a critical part of the Seabed 2030 ambition, but remains a gap that shipborne echosensors fail to address. Shallow water at the coasts is the interface between land and water, and its mapping is of the utmost importance for understanding ocean-land interactions and their interplay. A spin-off from the science objectives of NASA’s ICESat-2 is the potential to derive bathymetry for shallow waters with high precision; thus, ICESat-2-based seafloor data has the potential to connect the land and ocean maps.

The shallow waters along our coasts act as a buffer between the ocean and land. Depths of the shallow water get attenuated due to the wave energy dissipation through seabed friction. Mostly, they are in the epipelagic zone, permitting sunlight to reach the seafloor’s bottom. Shallow waters therefore provide a rich environment for many organisms, including fish, corals, sea turtles and mammals. They also regulate climate variables by processing nutrient effluence from land and anthropogenic emissions. For marine navigation, seafloor depth information in shallow water areas is crucial for manoeuvrability and safety issues. Therefore, high-precision bathymetric data for shallow waters is of immense use in many areas, including earth sciences and the understanding of benthic structures.

The challenges of mapping shallow water depths

Determining the depth of shallow water at less than 20 metres is generally challenging if relying on hull-mounted sensors, as the acoustic signals get distorted, impacting the measurement accuracy. Moreover, shallow water may pose manoeuvrability challenges for the ship. Applications of optical remote sensing-based methods for mapping shallow water depth are still in the validation stage, and the outputs are generally empirical; furthermore, these remote sensing methods need seed points as input during bathymetric modelling. Signals from non-imaging active sensors such as Lidar can penetrate water up to a certain depth before being attenuated, and can therefore aid in generating bathymetric information, especially in shallow waters. However, the more sediment in the water, the more the light attenuates due to scattering or absorption before reaching the seabed; thus, it is advised to acquire Lidar data during periods of reduced sediment load.

Harnessing the water penetrability of ICESat-2 green lasers for shallow water bathymetry

NASA’s ICESat-2, a novel satellite with a solo sensor – ATLAS – was launched in September 2018 and has been operational ever since. ICESat-2 is unique in providing measurements every 0.7m along the track. Combined with ICESat-2’s precision pointing and positioning information, the range measurements produce geolocation and elevations for all the successful laser photon returns from the Earth’s surface. A sequence of geolocated photons is of immense use to study the Earth’s surface features through profiling. In the first five years after the launch of ICESat-2, numerous research areas in the earth sciences reported the significance and application capability of geolocated photon data – in fact, the geolocated photon data from ICESat-2 not only filled gaps in the geospatial technologies but also unleashed novel applications for earth sciences. Even though the science objectives of ICESat-2 are oriented towards observing the cryosphere, researchers have successfully used the geolocated photons in a variety of disciplines related to land, ice, sea ice, vegetation/canopy, inland surface water and atmospheric studies in combination with optical remote sensing data. Researchers have successfully validated the application of ICESat-2 photon-based water-level measurements with the highest accuracy for various inland water bodies.

Researchers also observed green laser-based photons from the ICESat-2 penetrating the water column up to a depth of ~40m during clear water conditions; a phenomenon that successfully ushered in the spin-off application of ICESat-2 for shallow water bathymetry (Figure 1). However, night-time acquisition is advised for higher precision, due to reduced solar noise. The use of ICESat-2 data obtained during less sediment-heavy seasons is also advocated, to avoid photon attenuation and to reduce errors during bathymetric measurements.

The mode of data acquisition by ICESat-2 is along-track. Thus, to generate a bathymetry at a synoptic level, seafloor depths identified by the water-penetrated photons should either be used as seed points together with the optical remote sensing data or be interpolated to generate a surface from a set of multibeam photon clouds gathered from multi-date ICESat-2 acquisitions in conjunction with known depths accrued from electronic navigation charts (ENCs) or other sources.

ICESat-2 photons, once they hit the Earth’s surface, return from a variety of features such as land, water, canopy and snow. Thus, to identify the seafloor-returned photons, the returns must be classified based on the surface features. Numerous data science-based clustering algorithms (e.g. DBSCAN) can classify the return photons based on the surface features (Figure 2). By default, photons that have returned from the seafloor are apparent and require the application of a refraction correction to retrieve their actual depths. This is because of the change in speed of light at the air-water interface due to the differing refractive indexes of air and seawater. The default vertical datum of ICESat-2 photons is the height above the WGS84 ellipsoid, which needs to be converted to orthometric heights using geodesy utilities. Once a cloud of depth points accrues from seafloor return photons, it can be applied to the surface generation method using interpolation techniques. During the interpolation process, there may be a need for other known depth points (e.g. from ENCs or other sources) to satisfy the criteria of ‘well-distributed points’ during the interpolation process.

Successful mapping of submerged Adam’s Bridge using ICESat-2 geolocated photons

Adam’s Bridge joins Dhanushkodi, the south-eastern point of Rameshwaram Island in India, to Talaimannar, the north-western end of Mannar Island in Sri Lanka. The land bridge is an isthmus primarily submerged in shallow waters with occasionally exposed sandbanks. The southern part of Adam’s Bridge has the Gulf of Mannar, an arm of the Indian Ocean, and Palk Strait, an inlet of the Bay of Bengal, to the north.

Adam’s Bridge, regarded as a reefal assemblage, is a matter of scientific curiosity, and further understanding of its morphological structure based on a high-resolution Digital Bathymetric Elevation Model (DBEM) would provide information to reconstruct its evolution. In ancient texts, the land bridge is referred to as Ram Setu or Nala Setu. Until recently, with the advent of optical satellite imagery, researchers have reported only on the exposed parts of Adam’s Bridge, while very little information is available on the submerged structure due to the lack of high-resolution bathymetric information in its vicinity.

A high-resolution DBEM of Adam’s Bridge was generated successfully based on 133 tracks of ICESat-2 acquisitions between 2018 and 2023. Abiding by prerequisite conditions such as preferring night-time acquisitions and omitting data acquired during turbid load periods, 66 tracks qualified from the available 133 tracks, and these comprised 396 strong and weak beams of along-track data. All 396 data beams were processed to classify the returned photons from the water surface, water column, land and seafloor using clustering algorithms followed by manual correction using localized statistical algorithms to eliminate the outliers. The process produced ~0.2 million seafloor depths, which were used to generate a high-resolution (10m) bathymetric model for the study area. To understand the intricate details of Adam’s Bridge, elevation values (of those features that are above the mean sea level) and the seafloor depth both proved to be vital; a simple digital bathymetric model with only depth values (below mean sea level) would not provide sufficient information about the exposed features of the study area. The advantage of ICESat-2 photons is that they give elevation information above mean sea level and depth values of the seafloor, enabling the realization of both exposed and submerged features in the study area.

The results from this research provided intricate details of the submerged reefal assemblage of Adam’s Bridge (Figure 3). The first information that this DBEM provides is that, in its entire form, Adam’s Bridge is a submerged ridge with a submarine continuation of Dhanushkodi and Talaimannar Island from a rationalized depth of 8m. The current structure of Adam’s Bridge, being a barrier between two water systems – the Gulf of Mannar and Palk Strait – is influenced by the energy resulting from wave attacks from either side. At regular intervals along Adam’s Bridge, sudden narrow channels of depths of two to three metres probably permit the free flow or exchange of water between the Gulf of Mannar and the Palk Strait. Importantly, the narrow channels are accompanied by perpendicular ridges, in particular stretching towards the side of Palk Strait. These perpendicular ridges are likely the result of accumulated sediments/sands pushed by the dominant energy waves from the Gulf of Mannar over the years.

Volumetric analysis by fixing 8m water depth as a base for the study area yielded a volume of ~1km³ – equivalent to 1,000 times that of the Empire State Building’s volume. Of the total volume of the entire Adam’s Bridge, the volume ratio towards the Gulf of Mannar and Palk Strait is 44:56. Furthermore, the volume of Adam’s Bridge above 0m is 0.02km³, which is only 0.02% of the total volume; this is the same extent that is visible in the optical satellite imagery. The ICESat-2 photon-based DBEM is now being used as an input to various studies to understand its evolution.

Promising scope of ICESat-2 photons for mapping shallow-water bathymetry and its limitations

The multibeam approach of ICESat-2 acquisition enables an across-track span width of ~6km coverage with a spread difference of ~3km distance from the central beam pairs. Moreover, as each beam pair of ICESat-2 consists of a strong and weak beam with an across-track distance of 90m separation, it enables the accumulation of a good density of depth/elevation points. The high-resolution sampling for every 0.7m along the track provides detailed profiling of the Earth’s surface features. Recently, publications have shown the successful usage of ICESat-2’s geolocated photons as calibration data and seed data for empirical satellite-derived bathymetry. Coarse-resolution bathymetry datasets can be turned into high-resolution bathymetric models using the dense depth-point clouds accrued from the ICESat-2 water-penetrated geolocated photons. Open access global bathymetric data sources such as the General Bathymetric Chart of the Oceans (GEBCO) and Global Multi-Resolution Topography (GMRT) are available at ~450m and 100m, respectively. Research using the existing GEBCO depths and depths from the ICESat-2 photons enabled the creation of an improved resolution bathymetry dataset for the shallow waters at the coast of Point Pedro in Sri Lanka (Figure 4). Thus, there is ample scope to use the ICESat-2-derived seafloor depths to generate bathymetry for shallow waters at the global level.

There are limitations to using the ICESat-2 beams to derive the bathymetry for shallow waters. Computationally efficient clustering algorithms are available, which can aid in classifying the photons based on the returns from the water surface, water column and seafloor. However, not all the ICESat-2 acquisitions will enable seafloor detection. The sediment load in shallow water attenuates the ICESat-2 photon’s ray tracing mechanism to reach the seafloor, and daytime acquisitions generally fail to represent the seafloor returns. Furthermore, the 90-day temporal resolution of ICESat-2 permits only four acquisitions per year, which dents the availability of data for the study area. In terms of generating the surfaces by using the collection of ICESat-2 photons and other known bathymetry depth points (e.g. from ENCs), the preference for an interpolation method and reduced cell size can also be challenging due to the complex topography of the seafloor. If inconsistencies exist between the depth values accrued from the ICESat-2 water-penetrated photons and the existing bathymetric sources, the surface generation produced by fusing both these datasets may exhibit significant errors and artefacts. Despite these technical limitations, ICESat-2 is currently the best available space-borne Lidar sensor to aid in generating bathymetry for shallow waters, with the highest accuracy.

References

Dandabathula, G., Hari, R., Sharma, J., Sharma, A., Ghosh, K., Bera, A. K., & Srivastav, S. K. (2023). Prerequisite condition of diffuse attenuation coefficient Kd (490) for detecting seafloor from ICESat-2 geolocated photons during shallow water bathymetry. Hydrology, 11(1), 11–22.

Dandabathula, G., Ghosh, K., Hari, R., Sharma, J., Sharma, A., Padiyar, N., ... & Chauhan, P. (2024). Physical features of Adam’s Bridge interpreted from ICESat-2 based high-resolution digital bathymetric elevation model. Scientific Reports, 14(1), 14896.

Parrish, C. E., Magruder, L. A., Neuenschwander, A. L., Forfinski-Sarkozi, N., Alonzo, M., & Jasinski, M. (2019). Validation of ICESat-2 ATLAS bathymetry and analysis of ATLAS’s bathymetric mapping performance. Remote sensing, 11(14), 1634.

Babbel, B. J., Parrish, C. E., & Magruder, L. A. (2021). ICESat‐2 elevation retrievals in support of satellite‐derived bathymetry for global science applications. Geophysical research letters, 48(5), e2020GL090629.