Towards the unification of radar altimetry processing approaches for different surface types

PI:  Christopher Buchhaupt, University of Maryland, College Park
Start Year: 2025 | Duration: 4  years
Partners: NOAA, NASA

Project Abstract: 

Satellite altimetry has revolutionized our understanding of ocean surface topography, offering unprecedented insights into global oceanographic processes. From Geos-3 in 1975 to the recently launched Jason-3, most satellite radar altimeters have employed the conventional or low-resolution mode (LRM) for data processing. The introduction of delay/Doppler altimetry in 1998 marked a paradigm shift, significantly enhancing the precision of satellite radar measurements. This technique, akin to unfocused synthetic aperture radar (SAR), drastically reduces the altimeter footprint along the flight direction and improves surface measurement accuracy. It also significantly enhances the signal-to-noise ratio in retrieving geophysical parameters. Recent advances have improved the agreement between LRM and SAR processing schemes, addressing discrepancies crucial for accurate ocean measurements. For example, in SAR mode parameter retrieval–also known as retracking–it has been shown that vertical wave particle velocities and the mean line-of-sight velocity need to be considered as otherwise the sea surface heights and significant wave height estimates will not align with LRM results. Conversely, LRM waveform retracking, and to some extent SAR stack retracking, requires a more accurate antenna pattern in the signal model. Although minor questions remain, the reconciliation of discrepancies between SAR and LRM measurements is largely complete. Despite these advancements, current methods face challenges with non-ocean surfaces, such as sea ice and coastal areas. Existing models often rely on empirical approaches or inadequately represent the unique characteristics of these surfaces. For example, sea-ice surfaces are typically handled with empirical methods or physical models that are not well-suited to their specific conditions. Coastal ocean surfaces are retracked using non-numerical signal models that approximate the point target response with a Gaussian function and overlook crucial vertical and mean-line-of-sight wave motions important for SAR mode retracking. Our project aims to bridge these gaps by integrating advanced retracking techniques across different surface types, including open ocean, coastal ocean, and sea-ice. We will develop a unified geophysical numerical retracking approach and utilize machine learning to enhance surface type classification and consistency.