Satellite radar interferometry: Estimation of atmospheric delay

Shizhuo Liu*

*Corresponding author for this work

Research output: ThesisDissertation (TU Delft)

Abstract

Satellite SAR Interferometry (InSAR) is frequently promoted as an active, all­weather technique in comparison to optical remote sensing techniques that need solar illumination and cloud free conditions. However, the earth's atmospheric refractivity plays an important role in delaying the radar signal, resulting in a spatio-temporal phase shift in repeat-pass InSAR measurements (i.e., interferograms). The phase shift is also referred to as an atmospheric phase screen (APS). Therefore, repeatpass InSAR is not weather independent.
For C-band and X-band interferometry APS is mainly driven by the earth's tropo­sphere. The difficulty of correcting APS for surface deformation monitoring largely comes from the highly dynamic behavior of water vapor located in the lower ( < 2 km) troposphere. In this study we develop two methods for the APS correction and one method for the usage of APS for meteorological purposes. The first correction method is based on numerical weather models (NWM) and the second relies on In­SAR time series analysis. The method for meteorological purposes aims at retrieving water vapor spatial distributions at single SAR acquisitions from interferograms.
In the first correction method, we use the weather research and forecast (WRF) model to simulate delay at SAR acquisitions. To evaluate the simulation, we use atmosphere-only interferograms over different climatic regions varying from tropic to semi-arid. Some of the regions have strong topography variations and the others have flat terrains. We find that significant APS correction can be achieved only in mountainous regions. This is mainly the result of correcting the delay correlated with topography. However, the simulated topography dependent delay is not always reliable with a success rate of 80% (1 out of 5 cases in which the model failed).
In flat regions the correction hardly leads to APS mitigation and very often deteri­orates the original interfero-grams. This is because the model is not able to simulate the realistic water vapor lateral distribution at fine scales (between 1 and 50 km). We also assess the feasibility of using the model to simulate stochastic properties of APS such as spatial variograms. Unfortunately, we find the model constantly underestimates the real variability of APS at all spatial scales between 1 and 50 km.
Therefore, we conclude that contemporary weather models are useful for correcting the topography dependent delay but should not be considered reliable at all times. For flat regions, the value of the models is very limited and we recommend not to use them for delay correction.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Delft University of Technology
Supervisors/Advisors
  • Hanssen, R.F., Supervisor
Award date30 Oct 2012
Publication statusPublished - 2012
Externally publishedYes

Keywords

  • satellite radar interferometry
  • atmospheric delay
  • numerical weather models
  • Least-squares collocation
  • Deformation

Fingerprint

Dive into the research topics of 'Satellite radar interferometry: Estimation of atmospheric delay'. Together they form a unique fingerprint.

Cite this