87
g 2008
REDUCTION OF ATMOSPHERIC WATER VAPOUR EFFECTS ON ENVISAT ASAR
INTERFEROGRAMS USING MERIS NEAR IR MEASUREMENTS
Zhenhong Li
Department of Civil, Environmental and Geomatic Engineering, University College London, London WC1E 6BT, UK
Now at Department of Geographical and Earth Sciences, University of Glasgow, Glasgow G12 8QQ, UK
Zhenhong.Li@ges.gla.ac.uk
Commission VII, WG VII/2
KEY WORDS: SAR, Change Detection, Radar, Integration, Monitoring, Mapping, Surface, Geodesy
ABSTRACT:
Atmospheric water vapour effects represent a major limitation of repeat-pass Interferometric SAR (InSAR) techniques including
InSAR time series approaches (e.g. permanent scatterers (PS) and small baseline subset (SBAS)). In this paper, it is demonstrated
that atmospheric water vapour effects greater than 4 cm can be observed even in desert regions (e.g. Southern California) and the use
of MERIS correction model can improve the accuracy of InSAR derived deformation signals from 9.9 mm to 4.1 mm. It is also
shown that, using an advanced integration technique titled InSAR Time Series with Water Vapour correction model (InSAR TS +
PWV) for reduction of water vapour effects with coincident MERIS near IR water vapour data, a time series of postseismic motion
after the 2003 Bam (Iran) earthquake is achievable with about 50% reduction in RMS model misfit. It is believed that this study not
only contributes directly to the ENVISAT mission, but also will benefit space agencies’ plans to design and launch InSAR missions
because it aids in the identification of necessary characteristics of their future InSAR missions.
1. INTRODUCTION
Land surface deformation is a major worldwide hazard which
can result from natural processes such as earthquakes,
volcanoes and landslides, or from anthropogenic processes
including extraction of groundwater, oil and coal. Surface
displacement causes many problems including economic loss
and human suffering. One estimate placed the direct annual cost
of subsidence damage and mitigation within the USA alone at
over $100 million (Panel on Land Subsidence, 1991). Indirect
costs of subsidence are even greater, such as the indication that
subsidence was a contributing factor to the levee failures in
New Orleans when hurricane Katrina hit in 2005 (Dixon et al.,
2006). On 8 October 2005 one of the deadliest earthquakes in
modem times struck the region of Kashmir in South Asia and
claimed 80,000+ lives. Therefore, land deformation mapping is
of crucial importance not only for sustainable economic
development but also for the safety of the general public.
Interferometric SAR (InSAR) has been already revolutionizing
our ability to image surface deformation at spatial resolutions of
a few tens of metres in the last two decades, which in turn has
led to many new insights into geophysical processes, such as
earthquakes, volcanoes and landslides (Massonnet and Feigl,
1998). It is well-known that InSAR techniques are limited by
atmospheric effects: any difference in radar signal propagation
delays between SAR acquisitions leads to additional shifts in
phase measurements, which largely result from changes in
water vapour content in the troposphere (Hanssen, 2001).
ESA’s ENVISAT provides a unique opportunity to investigate
atmospheric water vapour effects on SAR interferograms due to
its two main payloads, Advanced Synthetic Aperture Radar
(ASAR) and MEdium Resolution Imaging Spectrometer
(MERIS): (1) There is no time difference between acquisitions
of water vapour and SAR data; (2) MERIS full-resolution mode
has a high spatial resolution of up to 300 m; (3) MERIS and
ASAR have a virtually identical propagation path.
2. ATMOSPHERIC EFFECTS ON INTERFEROGRAMS
2.1 Atmospheric Effects on Interferograms - Theory
After removing the flat earth and local topography, the
unwrapped differential interferometric phase at pixel (x,r)
computed from the SAR acquisitions at times t M , the start time,
and t s , the end time, can be written as follows (Berardino et al.,
2002):
i>, uh (x,r) = 8<t>\°^ (x,r) + ôtfil {x,r)+S<f>“' u l (x,r) + (x,r)
,Sft T ( V ).fE^V, AZ +'->
usX ’ 1 r sin 6>
4 n
S K<s (■ x ’ r ) = ~r[ d - ¿(/ M ,x,r)]
CD
5 Kl ( x ’ r ) = X[ d *m (t s >x>r) - d a , n {t M ,x,r)~\
where  is the transmitted signal central wavelength (in mm,
e.g. c. 56.3 mm for ASAR), d(t s ,x,r) and d(t M ,x,r)
represent the cumulative deformation in the line of sight at
times t s and t M respectively, with respect to the reference
instant t 0 , i.e., implying d(t 0 ,x,r) = 0,V(x,r) ; AZ(x,r) is
topographic error present in the DEM used for interferogram
generation, and its impact on deformation maps is also a
function of the perpendicular baseline component B lt t , the
sensor-target distance r , and the look angle 0 . The terms
datm{ t M’ x ’ r ) an d d aM (t s ,x,r) account for temporal