Comparisons between L-band and C-band InSAR for large 2007-2008 earthquakes in Asia: Implications for future SAR missions and earthquake studies
Zhenhong Li(1), John Elliott(2), Eric Fielding(3), Barry Parsons(2) and Richard Walters(2)
(1) COMET, University of Glasgow, University Avenue, Glasgow G12 8QQ, United Kingdom
(2) COMET, University of Oxford, Oxford, Oxford, United Kingdom
(3) Jet Propulsion Laboratory, Caltech, Pasadena, California, United States
Vegetation over earthquake fault zones limits the time interval over which reasonable interferometric coherence can be maintained. Our eventual aim is to use InSAR with the longer wavelength of PALSAR to mitigate this difficulty in the measurement of interseismic strain accumulation. While waiting for a large enough number of PALSAR scenes to accumulate for this purpose, we have compared the SAR interferograms produced using ALOS PALSAR and ENVISAT ASAR in the process of determining the mechanisms for a number of large earthquakes. These include the 12th September 2007 Mw 8.4 and 7.9 Mentawai, Sumatra earthquakes; the 9th January 2008 Mw 6.4 Gaize, Tibet earthquake; the 20th March 2008 Mw 7.1 Yutian, Xinjiang earthquake; and the 12th May 2008 Sichuan earthquake.
Even in heavily vegetated areas such as Mentawai, coherence at L-band is generally good and deformation is well resolved, highlighting its main advantage over C-band, i.e. less temporal decorrelation due to its capability to penetrate more deeply in vegetation. Our examples also show that the longer wavelength of L-band makes it easier to unwrap PALSAR phase measurements than ASAR where the phase gradient is great.
On the other hand, ionospheric streaks have been observed in PALSAR pixel offset maps even in the mid-latitude areas such as Sichuan (China), indicating that caution needs to be exercised when interpreting L-band results (since ionospheric effects on InSAR measurements will be ~17 times greater at L-band than at C-band) and ionospheric correction models should be explored.
With the repeat cycles of ALOS and Envisat (46 and 35 days respectively), it may take nearly 2 months or more to acquire all the necessary post-seismic images to map completely an earthquake deformation field using either PALSAR or ASAR data alone; this interval can be shortened by combining both datasets, particularly when ASAR WS data are employed. Furthermore, PALSAR usually collects strip-map measurements in the ascending mode with an incidence angle of ~38 degrees, whilst ASAR can provide measurements in both descending and ascending modes (with a nominal incidence angle of 23 degrees for strip-map). 3D displacement maps can be resolved by integrating these different datasets especially when including SAR pixel offsets, and hence enable earthquake model parameters to be better determined.
Acknowledgements: ALOS PALSAR and ENVISAT ASAR data were provided through JAXA ALOS AO project (ID: P0610001; PI: I. Dowman), ESA ALOS ADEN AO project (ID: 3577; PI: B. Parsons), ESA AOE-621(PI: B. Parsons), ESA AOE-668 (PI: E. Fielding) and DRAGON 2558 (PI: Q. Zeng).
*COMET: Centre for the Observation and Modelling of Earthquakes and Tectonics
#: Co-authors are in alphabetical order