Abstract
Orbit errors usually form a minor contribution to the error budget of spaceborne repeatpass synthetic
aperture radar interferometry (InSAR). However, inaccurately determined satellite trajectories can occasionally
have a very signicant eect on interferometric products and distort the largescale component
of the deformation signal. It is thus indispensable to be aware of the underlying mechanisms when applying
InSAR to deformation monitoring and to eventually consider dedicated corrections. Against this
background, the impact of orbit errors on InSAR processing is comprehensively analysed.
Following a brief introduction to InSAR processing, a both quantitative and qualitative characterisation of
expectable orbit errors is provided. The accuracy of available orbit products is reviewed and evaluated by
gathering global quality indicators originating from validation campaigns. This survey is complemented
by a parametric characterisation of the interrelation between relative orbit errors or baseline errors,
respectively, on the one hand, and error signals in the interferometric phase or coregistration osets,
respectively, on the other hand.
Based thereupon, approaches to reversely infer baseline corrections from residual phase patterns are
reviewed and evaluated with particular attention to the approximation quality of dierent parameterisations.
As a result, two estimators with optimised properties are described in detail: a least squares estimator
requiring prior unwrapping and a gridsearch estimator that can handle the wrapped phase. Both are
based on the same functional model, accounting for baseline errors by two parameters: the error component
perpendicular to the line of sight and the error in the rate of change of the parallel component.
The methodology is generalised by adjusting baseline error estimates in an overdetermined network of
linearly dependent interferometric combinations of images. Thus, systematic biases, for instance due to
unwrapping errors, can be detected and iteratively eliminated. Regularising the solution by a minimumnorm
condition also enables the inference of quasiabsolute orbit errors that refer to individual acquisitions.
Testing this approach on a sample Envisat data set involves the evaluation of dierent stochastic
models and concepts of hierarchical organisation. Whereas the least squares estimator produces a consistent
solution, gridsearch estimates turn out to be unreliable in specic cases.
The study of orbit error correction approaches is concluded by an outlook on potential application scenarios.
It is further complemented by analysing some related error mechanisms that likewise stem from
inaccurate modelling of the acquisition geometry. Thus, the eects of timing errors and clock errors are
characterised, and the signicance of decorrelation due to orbit convergence is investigated. A whole
chapter is dedicated to the eect of unmodelled reference frame motion on InSAR deformation estimates.
The resulting bias is predicted for Envisat acquisitions at various locations on the globe, and three
correction approaches are proposed.
aperture radar interferometry (InSAR). However, inaccurately determined satellite trajectories can occasionally
have a very signicant eect on interferometric products and distort the largescale component
of the deformation signal. It is thus indispensable to be aware of the underlying mechanisms when applying
InSAR to deformation monitoring and to eventually consider dedicated corrections. Against this
background, the impact of orbit errors on InSAR processing is comprehensively analysed.
Following a brief introduction to InSAR processing, a both quantitative and qualitative characterisation of
expectable orbit errors is provided. The accuracy of available orbit products is reviewed and evaluated by
gathering global quality indicators originating from validation campaigns. This survey is complemented
by a parametric characterisation of the interrelation between relative orbit errors or baseline errors,
respectively, on the one hand, and error signals in the interferometric phase or coregistration osets,
respectively, on the other hand.
Based thereupon, approaches to reversely infer baseline corrections from residual phase patterns are
reviewed and evaluated with particular attention to the approximation quality of dierent parameterisations.
As a result, two estimators with optimised properties are described in detail: a least squares estimator
requiring prior unwrapping and a gridsearch estimator that can handle the wrapped phase. Both are
based on the same functional model, accounting for baseline errors by two parameters: the error component
perpendicular to the line of sight and the error in the rate of change of the parallel component.
The methodology is generalised by adjusting baseline error estimates in an overdetermined network of
linearly dependent interferometric combinations of images. Thus, systematic biases, for instance due to
unwrapping errors, can be detected and iteratively eliminated. Regularising the solution by a minimumnorm
condition also enables the inference of quasiabsolute orbit errors that refer to individual acquisitions.
Testing this approach on a sample Envisat data set involves the evaluation of dierent stochastic
models and concepts of hierarchical organisation. Whereas the least squares estimator produces a consistent
solution, gridsearch estimates turn out to be unreliable in specic cases.
The study of orbit error correction approaches is concluded by an outlook on potential application scenarios.
It is further complemented by analysing some related error mechanisms that likewise stem from
inaccurate modelling of the acquisition geometry. Thus, the eects of timing errors and clock errors are
characterised, and the signicance of decorrelation due to orbit convergence is investigated. A whole
chapter is dedicated to the eect of unmodelled reference frame motion on InSAR deformation estimates.
The resulting bias is predicted for Envisat acquisitions at various locations on the globe, and three
correction approaches are proposed.
Original language  English 

Qualification  Doctor of Philosophy 
Awarding Institution 

Supervisors/Advisors 

Award date  22 Jan 2013 
Publication status  Published  22 Jan 2013 