IEEE Geoscience and Remote Sensing Magazine - March 2020 - 121

{ = S 1 S )2, (1)

	

where S 1 is the complex SAR images obtained at t 1, and S )2
is the complex conjugate of the SAR images obtained at t 2 .
The interferogram contains the phase information [19],
[20] from the flat Earth-surface, topography, surface deformation, atmospheric effect, and noise, such that
	

{ = { flat + { topo + { def + { atmo + { noise, (2)

where { flat is the phase component of the reference earth
surface, { topo is the topographic component, { def is the
surface-deformation component, { atmo is the atmosphere
component, and { noise is the phase noise. Among the
phase components, { flat can be removed by the referenceellipsoid model, and { topo can be removed through an external digital elevation model (DEM), such as that of the
Shuttle Radar Topography Mission (SRTM) [21]. The { atmo
usually needs careful examination to separate it from
{ def . The technique of removing { topo through the external DEM is also known as differential InSAR (DInSAR). The
relationship between the Earth-surface deformation and
deformation phase, { def , can be expressed as

Acquisition
at t2
,B

ne

li
se

tio

Ba

t

gh

Fli

c
re

Acquisition
at t1

on

cti

re
Di

Di

B⊥

n

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

THE REPEAT-PASS THEORY
The geometry of repeat-pass SAR interferometry is given
in Figure 1. The antenna illuminates the ground surface
at different times of t 1 and t 2, respectively. The complex
SAR image obtained at t 1 is denoted as S 1, which is the
master or reference image. The image obtained at t 2 is
denoted as S 2 and is the slave image. The separation between the antennas at t 1 and t 2 is the baseline, B. The
component perpendicular to the line-of-sight direction
is the perpendicular baseline, B = . The interferogram is
generated by multiplying S 1 with the conjugate of S 2,
such that

ge

MARCH 2020

MONITORING WETLAND WATER-LEVEL CHANGES

n
Ra

ecosystem services to human beings, such as natural water
purification and flood mitigation, and they provide habitats
for wildlife [2]. The water flow in the wetlands is a key driver
of their ecology [3]. For example, the vegetation assemblage
structure and growth cycle are influenced by the water depth
and local-scale currents [4], [5]. A reduction of flood frequencies and duration and wetland geographic areas can result in
the decline of vegetation, water birds, and fish [4]. The structure and function of a wetland depend heavily on hydrologic
conditions [6]-[8].
However, the hydrologic monitoring of wetlands using
traditional in situ data is challenging. This is because the
water flow in wetlands has a laterally extensive 2D movement, unlike a river flow that is bounded inside a welldefined channel [9]. Therefore, the point-based in situ
data cannot provide adequate spatial sampling of the 2D
water flow in wetlands. Furthermore, many wetlands are
located in inaccessible areas where it is difficult to maintain in situ gauges.
Because SAR transmits radar pulses at an off-nadir looking angle, the smooth open-water surface reflects most of
the radar energy away from the antenna, resulting in little
energy being returned to the receiver [10]. If the open-water
surface is rough and turbulent, part of the radar energy can
be scattered back to the receiver. However, the SAR images acquired at different times are not coherent across the
open-water surface. Thus, it has been generally accepted
that InSAR, which utilizes the phase information from two
SAR images obtained at different times or different spatial
positions, is not an appropriate tool to study open-water
surface-elevation changes. However, the authors of [10]-
[12] found that interferometric analysis of L-band SAR imagery from the Shuttle Imaging Radar-C (SIR-C) and Japanese
Earth Resources Satellite 1 (JERS-1) can generate 2D water-level
changes (2h/2t) across the flooded Amazon wetlands with
centimetric accuracy. Their work confirmed that the primary scattering elements for L-band radar consist of the water surface and vegetation trunks, which enables "doublebounce" backscattering returns and, therefore, maintains
good coherence across inundated forests.
Consequently, InSAR has been used to provide direct
measurements of 2h/2t with great accuracy and a high
spatial resolution (+ 30 m) beneath flooded vegetation.
The approach provides simultaneous measurements of
water-level changes and corresponding areas, which facilitates the calculation of temporal water-storage variations. For example, InSAR was deployed to map 2h/2t
in Congo wetlands [13]-[15] and the Florida Everglades
[16], [17] using L-band images from the JERS-1 and Advanced Land Observing Satellite (ALOS) phased-array-
type L-band SAR (PALSAR). In addition to L-band SAR
images, the authors of [18] found that interferograms
generated from the C-band Radarsat-1 and European
Remote Sensing (ERS) satellite images can maintain adequate coherence across Louisiana swamp forests during
leaf-off season.

ρ + ∆ρ
ρ

FIGURE 1. The geometry of repeat-pass SAR interferometry.

121
IEEE Geoscience and Remote Sensing Magazine - March 2020

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2020
