Abstract's details
Uncertainty assessment of SWOT's systematic errors calibration algorithm: the complementarity of inland and ocean water elevation data
Event: 2025 SWOT Science Team Meeting
Session: Hydrology: HR SWOT Data (Data Validation & Enhancement)
Presentation type: Oral
The KaRIn instrument onboard the SWOT mission represents a technological breakthrough in radar altimetry, delivering high-resolution two-dimensional water elevation profiles over both oceans and inland waters. However, KaRIn’s innovative radar interferometry brings new challenges, as systematic errors largely dominate the mission’s error budget. Specifically, KaRIn data are particularly sensitive to our imperfect knowledge of the satellite’s roll angle, baseline dilation caused by changing thermal conditions, antenna group delay, and interferometric phase errors.
To address these issues, the level-2 cross-calibration (L2 XCal) algorithm was developed. It operates in two main steps: First, it estimates systematic errors from sea surface height (SSH) mismatches at crossover points; second, it interpolates between estimates to produce a continuous correction over time. Although primarily designed for hydrological applications, a global assessment of the algorithm’s performance over inland waters remains a challenge due to the limited amount of independent water elevation data records available for calibration inland and the differing characteristics of SWOT’s data over oceanic versus continental water surfaces.
The height correction produced by the L2 XCal algorithm inland relies on distant oceanic estimates. As such, it is crucial for hydrology to also produce an estimate of the uncertainty of this correction. This uncertainty is expected to depend on both algorithmic parameters and external factors such as the satellite’s beta angle, which characterizes its illumination conditions.
This work explores data-driven approaches to quantify the uncertainty of the XCal correction. We perform experiments to quantify how much signal may leak from ocean to continents. To complement this analysis, we introduce so-called “fake continents”―large areas in the open ocean where SSH data are artificially masked―to mimic inland conditions. By comparing the height correction produced by the algorithm with and without fake continents, we derive a model to estimate height correction uncertainty when propagated over land. We also examine crossovers between SWOT and other nadir missions and stable inland areas inland that regularly yield SSH; situations where natural variability cancels out or is small and observed variations are mostly due to residual systematic errors. These variations constitute empirical realizations of such errors, which we compare to the L2 XCal algorithm output. These results contribute to a better understanding of the dynamics of SWOT’s systematic error correction and help enhance the reliability of SWOT data over inland water surfaces.
The level-3 cross-calibration (L3 XCal) algorithm was initially developed for oceanographic applications but is also used for hydrology during the Calval phase since the scarcity of crossovers for this orbit leads to a degraded L2 calibration. Enhancements to the L3 XCal algorithm are planned to improve interpolation over continents for inland water applications: an abacus of the height correction dependence on beta angle and local time has been produced. This approach reduces the magnitude of dynamically computed corrections, thereby lowering calibration errors over landmasses. Moreover, an optimal interpolation scheme based on the measured covariances of KaRIn systematic errors has been implemented to further improve interpolation and generate uncertainties to be provided to users.
Back to the list of abstractTo address these issues, the level-2 cross-calibration (L2 XCal) algorithm was developed. It operates in two main steps: First, it estimates systematic errors from sea surface height (SSH) mismatches at crossover points; second, it interpolates between estimates to produce a continuous correction over time. Although primarily designed for hydrological applications, a global assessment of the algorithm’s performance over inland waters remains a challenge due to the limited amount of independent water elevation data records available for calibration inland and the differing characteristics of SWOT’s data over oceanic versus continental water surfaces.
The height correction produced by the L2 XCal algorithm inland relies on distant oceanic estimates. As such, it is crucial for hydrology to also produce an estimate of the uncertainty of this correction. This uncertainty is expected to depend on both algorithmic parameters and external factors such as the satellite’s beta angle, which characterizes its illumination conditions.
This work explores data-driven approaches to quantify the uncertainty of the XCal correction. We perform experiments to quantify how much signal may leak from ocean to continents. To complement this analysis, we introduce so-called “fake continents”―large areas in the open ocean where SSH data are artificially masked―to mimic inland conditions. By comparing the height correction produced by the algorithm with and without fake continents, we derive a model to estimate height correction uncertainty when propagated over land. We also examine crossovers between SWOT and other nadir missions and stable inland areas inland that regularly yield SSH; situations where natural variability cancels out or is small and observed variations are mostly due to residual systematic errors. These variations constitute empirical realizations of such errors, which we compare to the L2 XCal algorithm output. These results contribute to a better understanding of the dynamics of SWOT’s systematic error correction and help enhance the reliability of SWOT data over inland water surfaces.
The level-3 cross-calibration (L3 XCal) algorithm was initially developed for oceanographic applications but is also used for hydrology during the Calval phase since the scarcity of crossovers for this orbit leads to a degraded L2 calibration. Enhancements to the L3 XCal algorithm are planned to improve interpolation over continents for inland water applications: an abacus of the height correction dependence on beta angle and local time has been produced. This approach reduces the magnitude of dynamically computed corrections, thereby lowering calibration errors over landmasses. Moreover, an optimal interpolation scheme based on the measured covariances of KaRIn systematic errors has been implemented to further improve interpolation and generate uncertainties to be provided to users.