The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi. XXXVII. Part B7. Beijing 2008
89
As shown in Figure 1, to reduce water vapour effects on the
interferograms, ZPDDM are inserted into the interferometric
processing sequence after removal of topographic signals by
use of a precise DEM such as that produced by the Space
Shuttle Radar Topography Mission (SRTM) (Farr et al., 2007);
The procedure used for InSAR water vapour correction has
been successfully incorporated into the JPL/Caltech ROI PAC
software by introducing two processing parameters:
ACMMODEL and ACMISZPDDM. ACMMODEL can be
set to NULL if no correction is applied; otherwise, the value
ACM IS ZPDDM is used to distinguish the type of difference
map(s) (currently, 0 = Slant Path Delay Difference Map
(SPDDM), 1 = ZPDDM).
After water vapour correction, atmospheric effects on the
corrected interferograms are limited and their residuals can be
considered as random noise, so that
50*1 (x,r) = ZPDDM + 5<j)^ ual (x,r) can be cancelled out in
Equation (1):
solid triangles represent GPS stations where the agreements of
InSAR-derived and GPS-derived range changes are within a
1-sigma range both before and after correction, white squares
with black borders imply improvement in their agreements after
correction, and red solid circles indicate deterioration in their
agreements after correction; (4) The circle and ovals in Figure
2(b) are believed to represent regions with real deformation
signals (see text).
Figures 2(a) and 2(b) show the interferogram spanning the
summer from 14 May 2005 to 27 August 2005 before and after
MERIS water vapour correction respectively. This InSAR pair
has a small baseline of c. 100 m, and the errors in the SRTM
DEM (c. 7.0 m in North America (Farr et al., 2007)) might lead
to a phase error of 0.49 rad (i.e. 0.22 cm in the satellite line of
sight range changes), which can be considered negligible. The
RMS of the unwrapped phase decreased from 1.95 rad before
correction to 0.79 rad after applying the MERIS water vapour
correction, indicating that the unwrapped phase was much
flatter after correction.
вф‘2 GO - £<C GO+c; GO+■sc’ GO
■SCG0=« i *UG0- 2>MW ( 2 )
¿C" 1 GO=¿C GO+<SC“ GO
where S<f>™T{x,r) represents corrected phase values and
8^ (x,r) residual water vapour effects after correction.
3.3 To what extent can water vapour effects be reduced
using MERIS?
-118*30’ —118"0Q' -117*30' -118*30' -118*00' -117*30' -117*00'
35*30'
35*00’
34*30'
34*00'
33*30'
33'00’
35*30'
35*00'
34*30’
34*00'
33*30'
(mm)
Figure 2. Interferogram (050514-050827) superimposed on a
SRTM DEM. (a) Original interferogram; (b) Corrected
interferogram using MERIS data. Note: (1) Grey within
interferogram coverage implies areas with low coherence due to
steep slopes and vegetation; (2) Negative values imply that the
surface moves towards the satellite; (3) In Figure 2(b), black
In order to validate the MERIS correction model, independent
3D GPS-derived displacements provided by the Scripps Orbit
and Permanent Array Center (SOPAC) (Nikolaidis, 2002) were
compared with InSAR results in the satellite line of sight (LOS)
direction, which showed that the RMS difference decreased
from 9.9 mm before correction to 4.1 mm after the MERIS
correction.
It is clear in Figure 2(b) that some real geophysical signals were
brought out after removing atmospheric water vapour signals:
(1) The black dashed oval indicates seasonal surface subsidence
in the Long Beach-Santa Ana basin (Argus et al., 2005; Bawden
et al., 2001); (2) The white solid oval implies surface
subsidence in Antelope Valley (Hoffmann and Zebker, 2003);
(3) The black solid circle represents geophysical signals in San
Bernardino (Lu and Danskin, 2001).
This case study has two important implications: (1) Water
vapour induced range changes in the LOS direction can be
greater than 4 cm even in a desert region such as Southern
California, and greater water vapour effects are expected in
tropic regions; (2) coincident water vapour products are
invaluable to assess and then reduce water vapour effects on
InSAR measurements. This successful integration of MERIS
and ASAR measurements not only contributes directly to the
ENVISAT mission, but also will benefit space agencies’ plans
to design and launch InSAR missions, because intensive
investigation of the major error source of InSAR techniques
will clearly identify necessary characteristics of future InSAR
missions which will in turn aid their ability to better achieve
space agencies’ science goals.
4. INSAR TIME SERIES WITH WATER VAPOUR
CORRECTION (INSAR TS + PWV)
4.1 InSAR time series with water vapour correction
Let t be a vector of SAR acquisition dates in chronological
order. For a dataset containing N interferograms constructed
from S acquisitions on different dates, assuming v kMl is the
mean velocity between the time-adjacent (e.g. the kth and
[k + \)th time) acquisition, the corrected phase can be written
as: