Search

The Antarctic Ice Sheet, as the largest reservoir of frozen freshwater on Earth, has been losing mass at accelerated rates and contributing to global sea-level rise. The region where the ocean can directly interact with the grounded ice sheet is the grounding zone (GZ). Its location is directly linked with changes in ocean tides and ice sheet mass balance, therefore it is a sensitive indicator of the ice sheet instability and a key component in predicting the contribution of the Antarctic Ice Sheet to future sea-level rise. However, GZ migrations and dynamics have not been fully understood and the GZ is a region that is challenging for various satellite remote sensing techniques to derive geospatial information due to its complex morphology. To date, there is no single technique that can be used to map the GZ of the entire Antarctic Ice Sheet with sufficient spatial accuracy and temporal resolution. There are many regions where the GZs have not been mapped repeatedly during the past two decades and little is known about the current state of the GZ evolution. To address these research gaps, this thesis utilizes ICESat-2 laser altimetry data to identify the current GZ locations of the Antarctic Ice Sheet and improve our understanding of GZ dynamics. The first results chapter develops a new automated method for mapping Antarctic GZs using ICESat-2 laser altimetry and evaluates the accuracy of this method over Larsen C Ice Shelf with independent grounding line (GL) measurements. The results demonstrate the efficiency, density and high spatial accuracy with which ICESat-2 can image complex GZs and its clear potential for pan-ice-sheet GZ mapping. The second results chapter further improves this method and produces the first ICESat-2-derived high-resolution Antarctic GZ product. This new GZ product provides significant improvements in spatial coverage and spatio-temporal resolution compared with previous satellite altimetry methods. The results show pervasive landward GL migrations along the Amundsen Sea Embayment, up to 15 km GL retreat on the Crary Ice Rise on Ross Ice Shelf, and short-term GZ migrations inside an ice plain due to ocean tide variations. The third results chapter utilizes multi-source remote sensing datasets to investigate the GL migrations in relation to ice dynamics at the Moscow University and Totten Glacier Ice Shelves in Wilkes Land, East Antarctica. The GLs have been retreating along ice plains at the fast-flowing ice streams in this region, coincident with the high thinning rates during the past decade. This indicates a mass loss trend dominated by ice dynamics is underway. Additionally, this chapter identifies two tide-modulated ocean channels which might open pathways for the warm ocean water intrusion into the ice shelf cavities. These results not only provide significant advances in the Antarctica-wide GZ mapping, they have expanded our understanding of the GZ dynamics and should help to improve projections of future contributions from the Antarctic Ice Sheet to global sea-level rise.

Achieving coastal and shallow-water bathymetry is essential for understanding the marine environment and for coastal management. Bathymetry data in shallow sea areas can currently be obtained using SDB (satellite-derived bathymetry) with multispectral satellites based on depth inversion models. In situ bathymetry data are crucial for validating empirical models but are currently limited in remote and unapproachable areas. Instead of the measured water depth data, we used the ICESat-2 (Ice, Cloud, and Land Elevation Satellite-2) ATL03 bathymetry data at different acquisition dates and multispectral imagery from Sentinel-2/GeoEye-1 to train and evaluate water depth inversion empirical models in two study regions: Shanhu Island in the South China Sea (SCS), and Heron Island in the Great Barrier Reef (GBR) in Australia. However, different sediment types also influenced the SDB results. Therefore, three types of sediments (sand, reef, and coral/algae) were analyzed for Heron Island, and four types of sediments (sand, reef, rubble and coral/algae) were analyzed for Shanhu Island. The results show that consistency generally improved when sediment classification information was considered in both study areas. For Heron Island, the sand sediments showed the best performance in both models compared to the other sediments, with mean RMSE and R2 values of 1.52 m and 0.90, respectively, representing a 5.6% improvement of the latter metric. For Shanhu Island, the rubble sediments showed the best consistency in both models, and the average RMSE and R2 values were 0.65 m and 0.97, respectively, indicating an RMSE improvement of 15.5%. Finally, bathymetry maps were generated in two regions based on the sediment classification results. These results indicate that the SDB method with sediment classifications can improve the water depth estimation consistency. This technology can also provide an efficient way to generate large-scale bathymetry maps in remote and sensitive shallow water areas for marine applications.

This study investigates pine heights by age for actively managed stands in the southeastern U.S. using ICESat-2 ATL08-derived height data and maps derived from the Landsat time series. We intersected ICESat-2 ground tracks with locations of pine plantations and the Landscape Change Monitoring System (LCMS) Fast Loss product to identify previously clear-cut pine plantations. We subtracted the LCMS Fast Loss year from the date of the ICESat-2 acquisition to determine plantation age at the time of the height measurement. We stratified the data for management intensity, where stands that experience both thinning and harvesting were considered actively managed. The goal was to develop age-height relationships across the region to characterize better the impact of management on productivity and site index. This research involved the analysis of over 137,998 ICESat-2 ATL08 segments in actively managed pine stands in the U.S. Southeast. We compared a subset of ICESat-2 heights with heights derived from airborne laser scanning acquisitions (ALS) available through the USGS 3D Elevation Program. The resulting R2 was 0.82, giving us confidence in the ICESat-2 ATL08-derived forest heights. Then, through data processing and analysis, we successfully stratified the spatial patterns of ICESat-2 ATL08 heights in the southeastern region. These patterns provided insights into the distribution and variability of forest heights across the region, contributing to informed decisions in forest management. We identified some challenges in predicting pine stand age through Landsat-derived disturbance products. We found that LCMS Fast Loss labels some heavy thins as a ‘Fast Loss,’ in addition to stand-clearing disturbances like clear-cuts, adding noise to our estimation of stand age. To overcome this issue, we employed a robust model of the logarithm of heights with a reciprocal of age using a random sample consensus (RANSAC) model to calculate site indices at base age 25 (years). Our results showed the site index for the region at a base age of 25 years is 20.1 m with a model R2 of 0.91. We compared the ICESat-2-derived site index with the FIA-derived site index to see the robustness of our results. Then, the modeled site index values were used to produce a map at a base age of 25 years for the U.S. Southeast, offering insights into spatial differences in regional forest productivity. The results of this study have important implications for ecological research, forest management, and well-informed decision-making. Insights into the distribution and trends of actively managed forest heights in the Southeast are gained from studying the vast dataset, allowing for more efficient land management and conservation initiatives. In actively manage stands, our site index equation improves the ability to anticipate site productivity and estimate future timber outputs. The difficulties with age estimation that have been observed highlight the need for better methods for mapping disturbances using remote sensing in forests that use thinning as a silvicultural prescription.

Amery Ice Shelf (AIS) in East Antarctica acts as the primary buttress for the largest reservoir of potential sea level rise in the region: the Lambert-Amery Basin. Iceberg calving is an important mass loss process for ice shelf systems and is one of the ways that an ice shelf loses mass. While calving events themselves are not inherently concerning for an ice shelf, changes from the steady state such as increased rifting and shortening calving cycles can indicate a mass imbalance. Therefore, it is crucial to understand calving processes to improve our ability to detect potential imbalances. Rifting is a precursor to all tabular iceberg calving events, yet there are many aspects of this process that remain unclear; for example, the disputed role of mélange (the ice mixture which congeals inside the rift at sea level) in rift propagation (Fricker, Young, et al., 2005; Larour et al., 2004). We use advances in satellite remote sensing to develop a metric for characterizing the propagation regime of ice rifts to indicate increased potential for calving events: the Rift Growth Metric (RGM). We determine the RGM from geometric observations derived from Landsat 8 imagery and ICESat-2 laser altimetry. ICESat-2 data enables us to characterize the status and behavior of mélange within each rift investigating its impacts on propagation of ice rifts on AIS. This method determines propagation shifts and assess mélange thickness spatiotemporally using geographic information systems software builds on established behavior of rift system dynamics before and after iceberg calving.

The Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) is a second- generation laser altimeter satellite tasked with accurately measuring the change in ice sheet mass balance [1]. Errors in measurements of ice sheet elevation stem from errors in knowledge of where the spacecraft is pointing, which is a direct result of errors in the knowledge of the spacecraft's attitude. In order to meet the elevation change requirements, very accurate attitude determination is required. One way to improve the accuracy of attitude determination is to improve the error models and the calibration parameters of the sensors on the satellite, particularly the gyroscope. On ICESat-2, the gyroscope is the second-generation space inertial reference unit (SIRU-2). The SIRU-2 consists of four rate integrating hemispherical resonator gyros arranged with their axes forming an octahedral tetrad pyramid shape [2]. This four-axis mounting configuration of the gyroscope is the motivation behind this work. On the original ICESat spacecraft, data from only three axes of the gyroscope was collected and used in the attitude filtering process. Instead of implementing an attitude filter that requires the calibration parameters to be referenced in the spacecraft body frame [3],[4], this work investigates modeling the gyroscope errors in the native gyroscope frame. As a result of this modeling configuration, more data is available; however, more states in the attitude filter need to be estimated. This document develops an Multiplicative Extended Kalman Filter (MEKF) with the gyro parameters in their native frame. After numerical verification of the new model, a sensitivity analysis is performed to determine whether the orthogonal or the sense axis base model is more robust to increased gyroscope noise, increased gyroscope axis bias, scale factor error, and misalignment error, and axis failure. Ultimately, the increased number of states inherent in the sense axis based model introduces an observability issue that does not negatively impact the attitude solution, but also does not allow for current independent prediction of the estimated error parameters. More investigation on ways to separate the estimation of the states is necessary to determine if the added information from this sense axis model can be of use.

The Ice, Cloud, and land Satellite-2 (ICESat-2) is planned to launch in 2016 carrying the Advanced Topographic Laser Altimeter System (ATLAS). ATLAS will be the first space-based photon-counting laser altimeter to be put into operation, and is tasked with observing the Earth’s ice sheets, sea ice, and vegetation. The environment in which ATLAS will be operating is expected to introduce a significant amount of noise into the received signal; this necessitates that a set of onboard Receiver Algorithms be developed to reduce the data volume and data rate to acceptable levels while still transmitting the relevant ranging data. The algorithms make use of signal processing techniques, along with three databases, the Digital Elevation Model (DEM), the Digital Relief Map (DRM), and the Surface Reference Mask (SRM), to find the signal and determine the appropriate dynamic range of vertical data surrounding the surface for downlink. The focus of this study is the development of the DEM and DRM databases. A number of elevation data sets are examined for use as inputs for the databases. No global data sets of sufficient quality and resolution are available for the development of the project, so best-available regional elevation data sets were selected instead. Software was developed in MATLAB to produce the DEM and DRM data bases from the input data sets. A method for calculating relief from a gridded elevation data set along the flight path of a satellite was developed for the generation of relief maps used to create the DRM. Global DEM and DRM databases were produced by mosaicking individual DEM and DRM tiles from each input data set into global grids. A technique was developed to determine the accuracy of the DRM by using ICESat ground elevations to evaluate the accuracy of an input elevation data set. By comparing values of DRM accuracy to values of DRM relief, estimates of DRM accuracy as a function of relief magnitude were determined and used to define values of DRM padding in the receiver algorithm.