NASADEM Overview and First Results: Shuttle Radar Topography Mission (SRTM) Reprocessing and Improvements

Abstract:

NASADEM is a significant modernization of SRTM digital elevation model (DEM) data supported by the NASA MEaSUREs program. We are reprocessing the raw radar signal data using improved algorithms and incorporating ICESat and ASTER-derived DEM data unavailable during the original processing. The NASADEM products will be freely-available through the Land Processes Distributed Active Archive Center (LPDAAC) at 1-arcsecond spacing.

The most significant processing improvements involve void reduction through improved phase unwrapping and using ICESat data for control. The updated unwrapping strategy now includes the use of SNAPHU for data processing patches where the unwrapped coverage from the original residue-based unwrapper falls below a coverage threshold. In North America continental processing, first experiments show the strip void area is reduced by more than 50% and the number of strip void patches is reduced by 40%. Patch boundary voids are mitigated by reprocessing with a different starting burst and merging the unwrapping results. We also updated a low-resolution elevation database to aid with unwrapping bootstrapping, retaining isolated component of unwrapped phase, and assessing the quality of the strip DEMs.

We introduce a height ripple error correction to reduce artifacts in the strip elevation data. These ripples are a few meters in size with along-track spatial scales of tens of kilometers and are due to uncompensated mast motion most pronounced after Shuttle roll angle adjustment maneuvers. We developed an along-track filter utilizing differences between the SRTM heights and ICESat lidar elevation data. For a test using all data over North America, the algorithm reduced the ICESat-SRTM bias from 80 cm to 3 cm and the RMS from 5m to 4m.

After merging and regridding the SRTM strip DEMs into 1x1-degree tiles, remaining voids are primarily filled with the ASTER-derived Global DEM. We use a Delta Surface Fill method to rubbersheet fill data across the void for a seamless merger. We find this to provide a more accurate fill than cut-and-paste patching. A new post-processing module creates DEM-derived layers from the void-free elevation data. The slope/aspect & plan/profile curvatures are found by fitting a local quadratic surface to each DEM post and computing metrics from the fit coefficients.