Measuring Long Wavelength Displacement with InSAR
Fattahi, Heresh; Amelung, Falk
University of Miami, UNITED STATES
Although Interferometric Synthetic Aperture Radar (InSAR) can be successfully used to measure crustal displacement in areas with localized deformation like volcanoes, earthquake epicenters and subsiding cities, its measurements of inter-seismic long-wavelength displacements are inaccurate. Orbital error is traditionally considered as the source of this InSAR's limitation. However, there is a lack of studies to assess the contribution of orbital error in InSAR displacement velocity maps. To properly evaluate the InSAR ability in measuring long-wavelength displacement, we present, for the first time, the mathematical formulation of orbital error contribution in InSAR LOS displacement velocity maps. This formulation enables us to analyze the potential of past, current and future SAR missions in measuring long-wavelength displacement.
Considering Envisat and ERS orbital accuracy, the evaluation of orbital error contribution in InSAR displacement velocity map shows that the contribution of orbital error in range and azimuth directions is less than 2 mm/yr over 100km, which is much less than observed ramp in InSAR velocity maps around 15 to 40 mm/yr over 100 km in range direction (based on our experience in several test sites using various satellite data). This contrast between the expected ramp due to the orbital error and observed ramp in the InSAR velocity map implies that other sources of systematic errors cause flattening residuals more than orbital error in InSAR products. In other words residuals from systematic errors and processing approximations have been misinterpreted as orbital error, limiting InSAR ability in measuring long-wavelength displacement.
We introduce several processing approximations, including geoid undulation and spherical approximations, as the possible sources of long-wavelength flattening residuals. We have been able to measure long-wavelength displacement using Envisat data without any orbital error correction in several test sites including Southern San Andreas fault, Baja California and Chaman Fault. Results from Southern San Andreas Fault are consistent with previous studies, which have used GPS to correct InSAR orbital error. Our preliminary results shows that Envisat and ERS SAR data can be used to measure long-wavelength displacement with an accuracy better than 2mm/yr over 100 km and Sentinels will provide an accuracy better than 1mm/yr over 100 km in InSAR velocity maps.