The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
91
Black ovals in Figure 4(a) indicate regions with water vapour
effects; (3) Positive values imply that the surface moves away
from the satellite; that is, the pixel exhibits subsidence in the
LOS, e.g. the red open rectangle labelled A; (4) Negative values
imply that the surface moves towards the satellite; that is, the
pixel exhibits uplift in the LOS, e.g. the red open square
labelled B.
Figure 5 shows another example on the date of 20060809.
Much stronger phase variations (e.g. those indicated by white
ovals) can be observed in Figure 5(a) than Figure 4(a), which
were reduced significantly after correction (Figure 5(b)).
(1) log(t) function:
,4 + 5xlog(i);
(2) exp(t/tao) function:
A + Bx( l-e' /r );
and (3) log-exp(P&A) function (Perfettini and Avouac, 2004):
S + Cxlog(l + i/x(e' /r -l)),
Figure 5. InSAR time series results: the LOS range changes on
date: 20060809 (relative to date: 20040211). (a) without
MERIS water vapour correction; (b) with MERIS water vapour
correction. Panels as in Figures 4(a) - 4(b) except that white
ovals indicate regions with water vapour effects in Figure 5(a).
4.3 Three years postseismic motions after the 2003 bam
earthquake
The distribution of the postseismic surface deformation
indicates that at least two different processes were involved,
with different spatial scales and at different depths in the crust
(Fielding et al., 2006). A narrow zone (roughly 500 m wide)
located where the surface ruptures of the 2003 earthquake were
observed south of the city of Bam (open rectangles in Figures 4-
5, labelled as A) continued to move away from the descending
satellite at least 3 years after the event (not shown in this paper).
Since signals with a similar magnitude can be seen on two
ascending tracks (156 and 385, not shown in this paper), the
displacement must be vertical. This can be interpreted as
localized and shallow compaction of material that dilated
during the earthquake (Fielding et al., paper in preparation for
Science, 2008).
A wider region moving towards the satellite can be observed in
the area indicated by open squares in Figures 4-5 (labelled as B),
i.e. the east to the south end of the main subsurface coseismic
rupture inferred from InSAR measurements (Funning et al.,
2005). Since the ascending tracks (156 and 385) show much
smaller signals over a smaller area (not shown in this paper),
this displacement must include both uplift and eastward
components. InSAR time series results reveal that this
displacement decays much more rapidly with time than the
shallow compaction. The displacement is believed to be due to
afterslip above and to the south of the main coseismic slip
asperity that ruptured during the 2003 earthquake (Fielding et
al., paper in preparation for Science, 2008).
Figure 6 shows that InSAR time series with and without MERIS
water vapour correction over Region B (see Figures 4(b) and
5(b)) can be fitted to three different functions:
where A, B, S, C, d and r are constants and t is time in years
since the earthquake. Strong variations can be observed in the
InSAR time series without MERIS water vapour correction,
with a misfit RMS of 0.25 cm. In contrast, the InSAR TS +
PWV time series appear to have a smaller misfit RMS of 0.14
cm (i.e. about 50% in reduction). For clarity, only lines showing
model fits to the time series with MERIS water vapour
correction are plotted in Figure 6. RMS values however are
given with respect to their own model fits. Region A (Figures
4(b) and 5(b)) exhibits a similar reduction in the RMS misfit
(not shown in this paper).
2004.0 2004.5 2005.0 2005.5 2006.0 2006.5 2007.0
Date (years)
Figure 6. InSAR time series of averages for rapid uplift region
B (2 km x 2 km, red and black squares in Figures 4 and 5
respectively). Note: (1) errors were estimated from RMS of the
area of interest; (2) Blue triangles represent time series without
MERIS water vapour correction whilst black squares indicate
time series with MERIS water vapour correction.
5. CONCLUSIONS
InSAR techniques can provide deformation measurements at
fine resolution (e.g. tens of metres) over wide areas (e.g.
100 km x 100 km). However, the accuracy of InSAR derived
deformation signals is usually limited to centimetre level due to
the spatiotemporal variations of atmospheric water vapour. This
paper has demonstrated the successful application of MERIS
water vapour correction model to ENVISAT ASAR data over
southern California: the RMS differences between GPS and