Geoscience Australia (GA) with the Commonwealth Scientific and Industry Research Organisation (CSIRO) are working to build a consistent synthetic aperture radar (SAR) capability in Australia. This will complement existing Earth Observation capabilities, supporting the Australian public and building international collaboration with global data providers. SAR provides value to Australia by enabling monitoring of floods, landcover change, coastlines, wetland inundation, soil moisture, vegetation classification, ground subsidence, deformation and elevation change. Regarding Australia’s interests in the Antarctic region, SAR enables monitoring of glaciers, coasts, and sea-ice by observing the continent through cloud cover and polar winter darkness.
To realise these benefits, GA has developed a CEOS-compliant Sentinel-1 normalised radar backscatter (NRB) product to be back-processed and delivered over Australia and Antarctica. Our chosen processing pipeline is an implementation of NASA JPL’s OPERA-RTC pipeline built on the ISCE-3 framework. The Sentinel-1 NRB product will be served along-side a static-layer product, allowing flexibility for users to convert easily between different backscatter conventions. This is critical over Antarctica, where dynamic terrain causes artifacts in NRB. To further mitigate terrain artifacts, the Digital Earth Antarctica program is collaborating with the Polar Geospatial Centre to integrate annual Digital Elevation Models (DEMs) into future ARD processing over Antarctica.
The ISCE-3 processor implements a new area-based projection algorithm (Shiroma et al., 2022) which produces differences in radiometry and geometry compared to the more widely used algorithms (Small et al., 2011). To enable Geoscience Australia to make an operational decision on which processor to use, we developed an evaluation framework to assess the quality and interoperability of NRB products produced by three SAR processors: SNAP, GAMMA, and ISCE-3. Differences in geocoding make this assessment challenging due to the dependence of radiometry on geometry. The evaluation framework reduces the complexity of the problem and provides linkages to NRB processing components to assist with the interpretation of quantitative results.
The evaluation study found ISCE3 (with atmospheric correction) provided the most precise geolocation and the most radiometrically consistent imagery over rugged terrain. Large deviations are observed occassionally at extreme local incidence angles, which can be easily flagged and masked by end-users. Our decision to implement ISCE3 is supported by these results as well as ISCE3’s ongoing operational use with the NISAR mission and overall processor performance. With ISCE3, GA can confidently produce consistent, high-quality, analysis ready SAR products over Australia and Antarctica.
Next, GA and CSIRO are exploring the development of an InSAR pipeline using NASA JPL Co-registered Single-look Complex (CSLC) products. Demonstrated for use over North America, CSLCs allow interferograms and backscatter can be produced quickly using band math, removing the pain of processing raw SAR imagery for scientists. A pilot study by CSIRO and the JPL OPERA program demonstrates a displacement product from CSLCs over four Australian cities. The results of this study will be used to assess the feasibility of operational InSAR processing over Australia and Antarctica.
