A Seamless Ice Sheet Digital Elevation Model using CryoSat

The Greenland and Antarctic ice sheets are contributing to a quarter of current sea level change and have the potential to raise sea level by several metres in the future. The surface elevation of ice sheets, and its temporal evolution, is one of the essential climate variables, it forms the basis observation for mass balance monitoring and the projection of sea level contribution under future climate scenarios. This work explores the creation of a seamless and gapless annual digital elevation model (DEM) derived from CryoSat radar altimetry measurements to aid in the ongoing study of their ever-changing topography. CryoSat-2 waveforms can be processed using two distinct techniques; (1) the conventional Point-Of-Closest-Approach (POCA), sampling a single elevation beneath the satellite, and (2) Swath processing which produces a swath of elevation measurements across the satellite ground track beyond the POCA, increasing spatial and temporal resolution. CryoSat operates in its Synthetic Aperture Radar Interferometric (SARIn) mode over the margins of the ice sheets allowing both processing techniques, however, within the ice sheet interior, CryoSat switches to its Low Resolution Mode (LRM), allowing solely the POCA technique for data processing. To achieve a comprehensive DEM encompassing the entirety of the ice sheet, whilst optimising data coverage, it is imperative to integrate and reconcile the outputs obtained from these distinct processing methodologies. This investigation uses two data sets provided by ESAs CryoSat thematic product range: the CryoSat-2 ThEMatic PrOducts (CryoTEMPO) land ice data set that applies the POCA processing technique and covers the entirety of the ice sheets and the CryoTEMPO-EOLIS (Elevation Over Land Ice from Swath) data set that provides a comprehensive point cloud data set specific to the ice sheet margins. In this investigation, the EOLIS and CryoTEMPO land ice datasets are aggregated into a spatial grid, utilising a Gaussian Radial Basis Function kernel to consider both, the spatial and temporal distribution of data points. To integrate EOLIS measurements from the margins of the ice sheet with CryoTEMPO land ice measurements from its interior, adjustments for variations in penetration are necessary to facilitate a seamless transition and mitigate the impact of anomalies. The combined and adjusted dataset is then post-processed to remove outliers while missing data is interpolated to generate a continuous DEM. Various spatio-temporal interpolation methods - such as External Drift Kriging, radial basis function, and DINCAE (Data Interpolating Convolutional Auto-Encoder) - have been explored and compared for their effectiveness. This poster will provide and summarise an overview of the gridding, merging, and interpolation methodologies. Additionally, an assessment of the performance of different interpolation methods and their accuracies will be presented with comparisons to existing DEMs.