IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 38, NO. 1, JANUARY 2000 105
Sensitivity to Soil Moisture By Active and Passive
Microwave Sensors
Yang Du, Fawwaz T. Ulaby, Fellow, IEEE, and M. Craig Dobson, Senior Member, IEEE
Abstract—The backscatter measured by radar and the emission
measured by a radiometer are both very sensitive to the moisture
content of bare-soil surfaces. Vegetation cover complicates the
scattering and emission processes, and it has been presumed that
the addition of vegetation masks the soil surface, thereby reducing
the radiometric and radar soil-moisture sensitivities. Even though
researchers working in the field of microwave remote sensing of
soil moisture are all likely to agree with the preceding two state-
ments, numerous claims and counterclaims have been voiced, pri-
marily at symposia and workshops, espousing the superiority of
the radiometric technique over the radar, or vice versa. The dis-
cussion is often reduced to disagreements over the answer to the
following question “Which of the two sensing techniques is less
impacted by vegetation cover?” This paper is an attempt to an-
swer that question. Using realistic radiative-transfer models for the
emission and backscatter, calculations were performed for three
types of canopies, all at 1.5 GHz. The results lead to two major
conclusions. First, the accepted presumption that vegetation cover
reduces the soil-moisture sensitivity is not always true. Over cer-
tain ranges of the optical depth of the vegetation canopy and
the roughness of the soil surface, vegetation cover can enhance,
not reduce, the radar sensitivity to soil moisture. The second con-
clusion is that under most vegetation and soil-surface conditions,
the radiometric and radar soil-moisture sensitivities decrease with
increasing τ, and the rates are approximately the same for both
sensors, suggesting that at least as far as vegetation effects are con-
cerned, neither sensor can claim superiority over the other.
I. INTRODUCTION
A
CTIVE and passive microwave approaches to sensing
soil moisture share certain physical processes, but they
are also markedly different with regard to the quantities they
sense and the image products they generate. Numerous papers,
and indeed books, have been published on the theory of radar
scattering from a soil surface, with and without vegetation
cover, as well as on analyses of experimental observations
and system imaging considerations ([1]–[17] comprise a
representative sample). Similar studies have been reported
for the passive-microwave case [18]–[26], but to date, only
one serious study has been conducted in which the active and
passive approaches were intercompared with respect to their
relative sensitivities to changes in soil-moisture content under
vegetation-covered conditions [27]. The comparison study,
which was published over a decade ago, provided useful and
interesting results, but the geometrical models it used in char-
acterizing the vegetation cover were somewhat simplistic. This
Manuscript received September 28, 1998; revised February 1, 1999.
The authors are with the Department of Electrical Engineering and Com-
puter Science, Radiation Laboratory, University of Michigan, Ann Arbor, MI
48109-2122 USA (e-mail: ulaby@eecs.umich.edu).
Publisher Item Identifier S 0196-2892(00)00015-2.
paper represents a focused revisit of the comparison question.
The study considers only a single set of wave parameters for
both active and passive simulations (L-band, incidence angle
, and an h-polarized antenna), but considers three types
of cover: a) a hypothetical layer of Rayleigh particles; b) a
grass-like vegetation layer; and c) a canopy of soybean plants.
II. SOIL-MOISTURE SENSITIVITY
In this study, we ignore atmospheric effects, terrain slope, and
imaging and calibration issues. We assume we have a pair of
sensors: a radar and a radiometer, both pointed at a soil surface
at an incidence angle of 30 . Both instruments operate at 1.5
GHz and use identical horizontally polarized antennas. The soil
is covered with a layer of vegetation characterized by an optical
depth τ and an albedo ω, as well as other geometrical properties
specific to the particular vegetation cover under consideration.
We shall start with two bare surfaces, one characterized by
a relatively smooth interface and the other by a medium-rough
interface (Section III). Appropriate surface scattering and emis-
sion models are then used to calculate the emissivity and the
backscattering coefficient as a function of the soil-volumetric
moisture content . (The zero subscript of and denotes
that these quantities pertain to the bare-soil case.) The radio-
metric and radar bare-soil moisture sensitivities are then defined
as follows:
(1)
(2)
where is in . For reasons that we will discuss later,
we shall also examine the behavior of the logarithmic radar sen-
sitivity defined as
(3)
where .
For the vegetation-covered case, we shall denote the emis-
sivity and backscattering coefficient by and , and the corre-
sponding soil- moisture sensitivities are defined as
(4)
(5)
0196–2892/00$10.00 © 2000 IEEE