One Researcher's Noise is Another's Signal: Dynamic Azimuthal Correction of ESCAT/ASCAT Backscatter for Long-Term Land Surface Monitoring

Satellite scatterometers are active radar instruments that transmit microwave pulses towards the Earth's surface and measure how much of this energy is reflected back, a quantity commonly referred to as backscatter. Operating at C-band, the ERS-1/2 ESCAT (1991–2011) and MetOp-A/B/C ASCAT (2007–present) missions together provide more than 35 years of global land surface backscatter observations. These data are widely used for monitoring soil moisture, vegetation dynamics, cryospheric processes, and land cover change. Their coarse spatial sampling of 12.5 km enables very high daily global coverage, which reaches about 82 % during the ASCAT era.

A characteristic feature of scatterometer measurements is azimuthal anisotropy, meaning that backscatter depends systematically on the viewing direction. This arises because ESCAT and ASCAT use multiple fixed fan-beam antennas that observe the same surface under different azimuth angles as the satellite passes overhead. Oriented surface structures such as vegetation patterns, agricultural rows, sand dunes, or wind-blown snow features (sastrugi), interact differently with the radar signal depending on their orientation relative to the radar look direction. While physically meaningful, this directional dependence introduces additional variability when measurements from different viewing geometries are combined, complicating the interpretation of collocated time series.

We present an updated azimuthal correction scheme for ESCAT and ASCAT backscatter that aims to improve the temporal consistency of long-term land surface monitoring. Earlier approaches relied on a unique static reference polynomial derived from multi-year data and mixed viewing geometries to implement the desired azimuthal bias correction, thereby harmonizing the data. Our updated method applies yearly reference polynomials instead and explicitly distinguishes between ascending and descending satellite orbits. This approach thus accounts for long-term land cover changes and systematic diurnal differences between morning and evening overpasses, which are preserved in the corrected backscatter. In addition, the right-looking satellite swath is used as the reference geometry, reflecting the single-swath configuration of the legacy ERS scatterometers. Consequently, this also supports a consistent alignment of measurements across diverse satellite missions. Global comparison maps and local examples show that the updated correction effectively reduces azimuthal noise, as indicated by a lower estimated standard deviation of the corrected backscatter. At the same time, genuine geophysical signals, such as systematic morning–evening differences likely linked to moisture variability, are preserved.

Azimuthal anisotropy is generally undesirable for most monitoring applications, as quantities such as soil moisture do not depend on viewing direction. However, we show that the correction polynomials themselves can provide useful information in specific environments, including sand dunes, ice sheets, and areas dominated by strong point scatterers. This can, for example, enable studies of the temporal migration of sand dunes or sastrugi. We also show that even individual buildings can affect the full footprint of a 25 km ASCAT pixel, as metal–glass structures can produce strong corner reflections when aligned with the radar look direction. The proposed approach therefore supports robust long-term monitoring of natural processes by reducing azimuthal noise, while the correction parameters themselves provide physically meaningful directional information.