Development of a Physically-based Soil Moisture Retrieval Algorithm ...

ῌ ,**3

Journal of The Remote Sensing Society of Japan Vol. ,3 No. + (,**3) pp. ,/-ῌ,0,

RSSJ

Original Paper

Development of a Physically-based Soil Moisture Retrieval Algorithm for Spaceborne Passive Microwave Radiometers and its Application to AMSR-E Hui LU*+, Toshio KOIKE*+, Hideyuki FUJII*,, Tetsu OHTA*+ and Katsunori TAMAGAWA*+ Abstract Many microwave radiometer algorithms for the retrieval of soil moisture tend to overestimate moisture in very dry cases, partly due to volume scattering e#ects. This study reports the development of a physically-based soil moisture retrieval algorithm for passive microwave remote sensing. The algorithm is based on physically-based radiative transfer, which simulates the radiative transfer processes in soil by a .-stream discrete ordinate method and the Henyey-Greenstein phase function. The multiple scattering e#ects of soil particles are calculated using the Dense Media Radiative Transfer Theory, and the surface roughness e#ects are simulated by the Advance Integral Equation method. The implementation of this algorithm consists of three steps : +) forward model parameters optimization ; ,) lookup table generation ; and -) lookup table reversion and soil moisture estimation. The algorithm was tested by retrieving soil moisture and temperature from AMSR-E Brightness Temperature data at a Coordinate Enhanced Observing Period reference site on the Mongolian Gobi. The retrieved soil moisture data was compared with in situ observations. The comparison shows that the performance of the new algorithm is satisfactory, with acceptable values of Standard Error of the Estimate and the square of the correlation coe$cient. Moreover, the algorithm estimates soil physical temperature accurately. Keywords : Passive Microwave Remote Sensing, AMSR-E, Soil Moisture Retrieval, Radiative Transfer Model, Lookup Table clude the long wavelength of microwaves and the independ+.

ence of the illumination source. These advantages have been

Introduction

recognized by many scientists, and extensive research acResearch on earth system modeling, global scale environ-

tivities have already been carried out worldwide. Models

mental process monitoring and climate change studies has

relating soil water content to soil dielectric properties were

been conducted using globally measured variables such as soil

first proposed by Wang and Schmugge,῍ and Dobson et al.-῍.

moisture, soil temperature and vegetation water content. Pri-

Several factors which a#ect the sensitivity of microwave emis-

mary among these is soil moisture, which links the land surface

sion to moisture have been estimated. For example, surface

and the atmosphere by influencing the exchange of energy and

roughness e#ects using some simple but empirically-based

material between the two.

However, due to their large

models were studied by Choudhury et al..῍, Wang and

variability, it is very di$cult to observe the spatial and tempo-

Choudhury/῍, Wigneron et al.0῍ and Urs Wegmuller and

ral distribution of soil moisture over a large scale using in situ

Matzler.1῍ A.K.Fung et al.2῍, Chen et al.3῍ and Shi et al+*῍

measurements, which are time consuming and expensive. As

studied surface roughness e#ects using physically-based

a result, in the past -* years much e#ort has been directed

models ; and Ulaby et al.++῍, Paloscia and Pampaloni+,῍,

towards observing soil moisture using satellite remote sensing+῍.

Jackson and Schmugge+-῍ studied them using vegetation. Al-

Fortunately, satellite passive microwave remote sensing

gorithms considering various other factors have also been

makes it possible to measure this important variable at the

proposed to retrieve soil moisture content from passive micro-

global scale by direct measurement of brightness temperature

wave remote sensed data. T. Jackson+.῍ developed a so-called

which are strongly related to the liquid moisture content.

single channel algorithm (SCA), in which the brightness

Further advantages of passive microwave remote sensing in-

temperature of the 0.3 GHz horizontal polarization channel

(Recievd July ,3, ,**2. Accepted December +,, ,**2) *+ River and Environmental Engineering Laboratory, Civil Engineering department, The University of Tokyo, 1ῌ-ῌ+, Hongo, Bunkyo-ku, Tokyo ++-ῌ20/0, Japan *, Earth Observation Research Center, Japan Aerospace Exploration Agency, Tsukuba, Ibaraki -*/ῌ2/*/, Japan

ῌ 253 ῌ

Development of a Physically-based Soil Moisture Retrieval Algorithm for Spaceborne Passive Microwave Radiometers and its Application to AMSR-E

was used. In this algorithm, ancillary data such as air temper-

its performance. Section / contains some concluding remarks.

ature, land cover, NDVI, surface roughness, and soil texture and porosity are needed. The algorithm of Njoku et al+/)

+0῏

,.

is

The Forward Model : A Fully Physically-based

a model-based iterative retrieval. It uses the brightness tem-

Radiative Transfer Model

perature observed by the lowest six channels of AMSR-E. Using their algorithm, the surface temperature, the vegetation

The objective of our research was to develop a physically-

opacity and the soil moisture are estimated simultaneously.

based soil moisture retrieval algorithm, which is able to esti-

The algorithm proposed by Paloscia+1) +2῏ is an experiment-

mate soil moisture content from low frequency passive micro-

based linear regression retrieval, in which soil moisture is

wave remote sensing data in a sparsely vegetated region.

estimated by using both the Polarization Index (PI) at +*.1 GHz and the brightness temperature at 0.3 GHz.

In our forward model, the downward radiation from vegetation and rainfall, which is reflected by the soil surface, is

Unfortunately, both the accuracy of retrieved soil moisture

neglected because the reflected radiation is much smaller than

measurement and the application range of those algorithms

the emission from the surface. The brightness temperature

still need to be improved. One key reason is that scattering

observed by spaceborne sensors is then expressed as :

processes are not well accounted for in their forward models.

TbῒTbs eῑtc eῑtrῐ῎+ῑwc῏῎+ῑeῑtc῏Tc eῑtr

In particular, the volume scattering e#ects of soil particles are

ῐΐ ῎+ῑwr῎R῏῏῎+ῑeῑtr῎R῏῏Tr῎R῏dR

generally neglected, leading to an overestimation of soil mois-

(+)

ture in dry cases. Shibata+3῏, for example, pointed out this

where Tbs is the emission of the soil layer, Tc is the vegetation

weakness. A second reason is that current models of surface

temperature, Tr is the temperature of precipitation droplets, tc

scattering e#ects of the soil surface lack physical meaning.

and wc are the vegetation opacity and single scattering albedo,

The application of those empirical and semi-empirical surface

and tr and wr are the opacity and single scattering albedo of

roughness models is strictly limited by the unavailability of

precipitation.

their parameters. Of first importance to solving such prob-

Emission from the soil is controlled by the soil properties

lems, the forward model, i.e. Radiative Transfer Model

and land surface roughness. Soil properties such as soil temper-

(RTM), needs to be improved (Mo et al,*῏).

ature, moisture content, and texture profiles are taken into

In this paper, we present a physically-based soil moisture

account through dielectric constant models and radiative

retrieval algorithm developed at the University of Tokyo. This

transfer processes inside the soil media. We used the Dobson

algorithm is based on a fully physically-based radiative trans-

model-῏ to calculate the dielectric constant of the soil ; and we

fer model. We calculate volume scattering inside soil layers

simulated the radiative transfer process inside the soil using a

using dense media radiative transfer theory (DMRT),+) ,,῏,

discrete ordinate method as proposed by Tsang and Kong,-῏.

and we simulate the surface roughness e#ect using the Ad-

We assumed that the soil has a multi-layer structure and is

vanced Integration Equation Model (AIEM)3῏. The forward

composed of many plane-parallel and azimuthally symmetric

model parameters are optimized using in situ observations

soil slabs with spherical scattering particles. The radiative

and lower frequency brightness temperature data. With these

transfer process in a plane-parallel and azimuthally symmetric

optimized parameters, we ran the forward model to generate

soil slab with spherical scattering particles can be expressed as :

a lookup table, which relates the variables of interest (soil moisture content, soil physical temperature, vegetation water

m

content and atmosphere optical thickness) to brightness temperature or other indices calculated from brightness tempera-

ῑ

ture data. Finally, we estimate soil moisture content by linear interpolation of the brightness temperature or index into the inverted lookup table. The performance of this algorithm was

ῌ῏ + d ῌIv῎tῌm῏῏ ῌIv῎tῌm῏῏ ῎ ῑῒ῎ ῑῑ῎+ῑw*῏B῎t῏῎ῑ dt ῍Ih῎tῌm῏ῐ ῍Ih῎tῌm῏ῐ + ῍ῐ + w* ῌPVV PVH῏ῌIv῎tῌmῒ῏῏ ῎ ῑ῎ ῑdmῒ ΐ , ῑ+῍PHV PHHῐ῍Ih῎tῌmῒ῏ῐ

(,)

where IP (t, m) is the radiance at optical depth t (dtῒKedz,

evaluated by means of retrieving soil moisture and physical

with extinction coe$cient Ke and layer depth dz) in direction

temperature data from the AMSR-E match-up data set for the

m for polarization status P (horizontal or vertical), w* is the single scattering albedo of a soil particle, B (t) is the Planck

Mongolia region. The paper is organized as follows. In Section ,, we present

function and Pij (i, jῒH or V) is the scattering phase function.

our physically-based radiative transfer model (RTM), em-

The .-stream model proposed by Liu,.῏ solves (,) by using the

phasizing the soil RTM, our so-called DMRT-AIEM model.

discrete ordinate method and assuming that no cross-

In Section - we describe the structure of our algorithm.

polarization exist. The Henyey-Greenstein formula,/῏ is used

Section . discusses the application of our new algorithm and

to express the scattering phase function.

῍ 254 ῍

Journal of The Remote Sensing Society of Japan

Vol. ,3 No. + (,**3)

For each slab, the extinction coe$cient Ke and albedo w

length (l), soil particle sizes (r) and vegetation parameters

used in equation (+) were calculated by the so-called dense

such as c and bῌ. These parameters are optimized by minimiz-

media radiative transfer theory (DMRT) under Quasi Crys-

ing the cost function :

talline Approximation with Coherent Potential (QCA-CP).

J῏hῌlῌrῌbῌῌcῌ῎ῐ

We simulated the land surface roughness e#ect using the

n

῕ῌ ῌ ῌ i῕+ f

Advanced Integral Equation Model (AIEM). AIEM is a

p῕H, V

ABSῑTBsim῏iῌfῌpῐ῔TBobs῏iῌfῌpῐῒ (/)

physically-based model with only two parameters : standard

where the subscript sim denotes the model simulated value

deviation of the height variations (or rms height) and surface

and obs is the observed value. n is the number of samples used

correlation length.

AIEM is an extension of the integral

in the optimization. p denotes the polarization status : H for

Equation Model (IEM)2ῐ. It has been demonstrated that IEM

horizontal polarization and V for vertical. f is some frequency

has a much wider application range for surface roughness

in the long wavelength region where the atmospheric e#ect

conditions than other models such as the Small Perturbation

may be ignored, such as 0.3, +*.1 and +2.1 GHz of AMSR-E,

Model (SPM), Physical Optics Model (POM) and Geometric

+.. GHz of SMOS and +3 GHz of SSM/I. Step ,. Lookup table generation. After Step +, the optimal

Optics Model (GOM). AIEM improves the calculation accuracy of the scattering coe$cient compared with IEM by

parameter values are then stored in the forward RTM. We

retaining the absolute phase term in the Green’s function.

then run the forward model by inputting all possible values of

By coupling AIEM with DMRT (DMRT-AIEM), this

variables used in Equation (+), such as soil moisture content,

radiative transfer model for soil media is fully physically-

soil temperature, vegetation water content and atmosphere

based. As such, the parameters of DMRT-AIEM, such as the

optical thickness. A family of brightness temperatures is then

rms height, correlation length and soil particle size, have clear

generated. Based on this brightness temperature database, we

physical meanings and their values can be obtained either

select brightness temperatures of special frequencies and po-

from field measurement or theoretical calculation.

larization to compile a lookup table or to calculate some

The e#ects of a vegetation layer depend on the vegetation

indices to compile a lookup table. For example, in an earlier

opacity tc and the single scattering albedo of vegetation wc

version of our algorithm, the index of soil wetness (ISW),0) ,1ῐ,

The vegetation opacity in turn is strongly a#ected by the

and Polarization Index (PI)+,ῐ were used to compile a lookup

vegetation columnar water content Wc. The relationship can

table. The ISW and PI were originally defined as follows :

be expressed as

+-ῐ

: ISW῕

bῌlcWc tc῕ cosq

(-)

where l is the wavelength, q the incident angle, Wc the vegetation water content.

c

PI῕

is +.-2 and bῌ is 3.-,.

The single scattering albedo of vegetation wc is small in the low frequency microwave region+,ῐ. In this research, wc is

TB῏f+ῌHῐ῔TB῏f,ῌHῐ + ῏TB῏f+ῌHῐΐTB῏f,ῌHῐῐ ,

TB῏fῌVῐ῔TB῏fῌHῐ + ῏TB῏fῌVῐΐTB῏fῌHῐῐ ,

(0)

(1)

where TB (f+, H), TB (f,, H) are the horizontal polarization brightness temperatures of frequencies f+ and f,, respectively.

calculated by wc῕w* · ῎ ῌ῍

TB (f, V), TB (f, H) are the vertical and horizontal polariza-

(.)

tion brightness temperatures of frequency f, respectively. Step -. Lookup table inversion and soil moisture estima-

-.

tion. The lookup table generated in Step , is reversed to give

The Algorithm

a relationship which maps the brightness temperature or As in other physically-based algorithms, such as that

indices obtained from satellite remote sensing data to the

developed by Njoku+/) +0ῐ and the SCA developed by

variables of interest. Finally, we estimate soil moisture con-

Jackson+.ῐ, the parameters used in our algorithm have clear

tent by linear interpolation of the brightness temperature or

physical meanings. This advantage derives from the strength

indices into the inverted lookup table.

of the forward radiative transfer model. The parameters of ..

the forward RTM should therefore be optimized before applying the algorithm.

Application to AMSR-E Data Set

The implementation of our algorithm

consists of three steps :

We tested our algorithm by retrieving soil moisture and

Step +. Forward model parameter optimization. The pa-

temperature input from AMSR-E TB data at a Coordinate

rameters to be optimized include rms height (h), correlation

Enhanced Observing Period (CEOP),2ῐ reference site in the

῍ 255 ῍

Development of a Physically-based Soil Moisture Retrieval Algorithm for Spaceborne Passive Microwave Radiometers and its Application to AMSR-E

Mongolian Gobi. The results with/without volume scattering e#ects were validated by comparing with in situ measurements by Soil Moisture Temperature Measurement System (SMTMS) and Automatic Weather Stations (AWS). .. +

CEOP Mongolia Reference Site and AMSR-E Matchup Data Set

The application region of this research is the AMPEX (ADEOS II Mongolian Plateau EXperiment for Ground Truth) area. AMPEX has joined the CEOP as the Mongolia reference site. AMPEX is designed to validate the AMSR and AMSR-E soil moisture algorithm(s). In this area, meteorological and land hydrological factors are measured with very densely installed instruments. AMPEX is located in the Mongolian Plateau, ,-/ km south of Ulan Bator. The area

Fig. +

stretches +0*km in the longitudinal direction (+*0ῐE῍

Distribution of ASSH and AWS in AMPEX study area

+*2ῐ-*῍E) and +,* km in the latitudinal direction (./ῐ-*῍N῍ .1ῐN) on the Mandalgobi, where 0 AWSs and +, ASSHs (Automatic Station for Soil Hydrology) were installed. Figure

that the maximum vegetation water content in our study area

+ illustrates the distribution of observation sites in this area.

was *.++ kg · m῏,.

For more details of AMPEX, please visit the following website

vegetation coverage.

: http : //home.hiroshima-u.ac.jp/~ampex/hm/index-e.htm.

.. ,

It is a small value reflecting the sparse

Best-fitting Parameters for AMSR-E

The ASSHs provide soil moisture and temperature profile

AMSR-E TB data obtained from low frequency channels

measured at two depths, - cm and +* cm below the surface.

(0.3,/, +*.0/ and +2.1 GHz) were used to optimize model

The AWSs provide soil moisture and temperature profiles

parameters. Since the wavelength of those channels is gener-

measured at four depths : - cm, +* cm, .* cm and +** cm

ally much larger than the diameter of atmospheric particles,

below the surface.

the atmospheric e#ect is negligible for the data measured with

The soil moisture measurements used

Time Domain Reflectometers (TDRs), and the soil tempera-

those channels.

ture was measured by platinum resistance thermometers.

As reported in the literature, it is reasonable to assume that

Through matching the AMSR-E footprints to in situ sta-

there is little or no volume scattering for soil moisture levels

tions, we generated a match-up data set consisting of bright-

over +*ῌ-*῎. So, we first used the data observed on wet days

ness temperature data observed by AMSR-E and in situ data

to estimate the roughness parameters, rms height and correla-

measured by SMTMS and AWS for the period from July to

tion length, in a best-fitting way. The particle size parameter

August, ,**-. The coverage of this data set is ,./ by ,./

could then be obtained by running a coupled DMRT-AIEM

degrees, with a resolution *.*/ degrees for all frequencies.

model to best fit the data observed on dry days.

The in situ data consists of soil moisture and soil temperature

In order to run DMRT, we used uniform soil moisture and

data. It is in the form of an image type and a text type. The

temperature vertical profiles with the value observed at - cm

text files record AMSR-E brightness temperature and in situ

depth. The bottom of the soil medium was set to be +.*m

data at each ground station. The in situ data include observa-

(layer thickness is +cm) and the brightness temperature at the

tions made within +, hours of the AMSR-E observation. In

bottom was assumed to be the soil physical temperature at

this research, the mach-up in situ data at each AMSR-E

that level, that is, the emissivity was equal to one.

satellite over passing is calculated by interpolating the in situ

downward radiation from each soil layer, reflected at the

data on the hour.

bottom boundary, was not considered in this study. The

The

Based on the AMSR ,**, field experiment results, the soil

interactions at the boundaries between neighboring soil layers

bulk density in this region is +.,/2 g · cm῏-. The soil texture

were also neglected because of the vertically uniform soil

is obtained from the Net Primary Productivity (NPP)

moisture and temperature profiles.

Database,3῎ : a sand fraction of *.0, a silt fraction of *., and

First, we used the AIEM model to best fit several wet day observations by changing rms height (s) and correlation

a clay fraction of *.,. As stated before, there is sparse vegetation in study area.

length (l). Second, employing this set of s and l, we could

The vegetation water content was measured in June and

obtain the surface emissivity for all observations. Third, with

August, ,**-. Based on this in situ observation, we found

some dry day observations, we could best fit the particle size

ῌ 256 ῌ

Journal of The Remote Sensing Society of Japan

Table + Data used for parameter optimization

Table , Best fitted particle size parameters

parameter using the DMRT-AIEM model. Finally, we calcuth

lated brightness temperature from April +* ,**- to April -*

Vol. ,3 No. + (,**3)

cal polarization and an index dTB calculated as follows :

th

dTBῒTB῏+2.1ῌHῐῑTB῏+*.0/ῌHῐ

,**. with best-fitting parameters.

(2)

Here, we use the A- station as an example, to introduce the

The lookup table of our AMSR-E algorithm is shown in

whole procedure and the result. Information about the data

Fig. -. It covers a region in which soil moisture content varies

we used to calibrate the model is listed in following table.

from ,ῌ to .*ῌ, and soil physical temperature varies from

The calibrated parameter values of AIEM with considera-

,1* K to -*- K. Compared with in situ observation values,

tion of shadowing e#ects are : sῒ*..0 cm ; lῒ+.*- cm. Then,

this range is large enough to include all of the actual soil

with this set of s and l, using data for .* dry days, we best fit

moisture and temperature states in Mongolia.

the particle size parameter as in table , :

Since the one-to-one relationship in our lookup table is very

As in Table ,, the best-fit particle sizes change at di#erent

clear, it becomes simple to reverse the lookup table, so that the

frequencies : longer wavelengths are matched with larger par-

soil moisture can easily be estimated from the AMSR-E data

ticle sizes. However, the ratio between the best-fit radius and

set. In this study, we retrieved soil moisture data for the

the wavelength in the sand is nearly constant. Therefore, we

period from July to August, ,**-. The estimation is shown in

call the best-fit radius the e#ective radius. The e#ective radii

Fig. . for (a) time variation and (b) accuracy comparison. It is clear from Fig. . that the algorithm gives a reliable soil

are generally larger than the physical values, consistent with similar results reported by Kendra and Sarabandi-+ῐ. .. -

moisture content estimate in both tendency and amplitude. The value of R-square is *.-3/-, and the Standard Error of

TB Validation

With the optimized parameters, we first ran the forward

the Estimate (SEE) is -.2ῌ. From Fig. . (a), we find some

model to simulate brightness temperature from 0.3 GHz up to

overestimation around Aug. ., +. and ,*, when moderate

-0 GHz using in situ observations of soil moisture and temper-

rainfall (/῎+* mm) occurred. Such errors can be attributed

The brightness temperature validation is

partly to the di#erence between the TDR sensor depth and the

shown in Figure ,, where the horizontal axis is the AMSR-E

penetration depth of the X band and Ku band. Moderate

observation and the vertical axis is our model output.

rainfall makes the soil surface much wetter than the soil -cm

ature as input.

As shown in Fig. ,, the simulation result of DMRT-AIEM

below the surface where the TDR sensors were located. Such

is generally good, with better results for vertical polarization

vertical heterogeneity of soil moisture in the first - cm of soil

than for horizontal polarization. The model overestimates

was not considered in our algorithm. On the other hand, the

brightness temperature at horizontal polarization for all

wet surface situation decreases the penetration depth dra-

frequencies, while the bias values for each frequency are

matically.

almost the same. This overestimation is partly due to the

algorithm estimate higher soil moisture content than the in

parameter optimization method, in which we set the same

situ observations for moderate rainfall periods.

The combination of these reasons makes our

weights for vertical and horizontal polarization. But, as we

One advantage of our proposed algorithm is that it esti-

know, the horizontal polarization is much more sensitive to

mates soil physical temperature and soil moisture simultane-

surface roughness than the vertical and the vertical TB is

ously. This is important for studies involving energy cycling

higher than the horizontal. As a result, our RTM cannot

and water cycling, such as studies of land surface processes

simulate horizontal TB as accurately as vertical TB. Further

and of weather forecasting.

work is needed to address this phenomenon. .. .

The average retrieved physical temperature for ASSH stations is shown in Fig. / (a) and Fig. /(b). As in soil moisture

Retrieval Soil Moisture from AMSR-E

Based on the TB validation shown in Figure ,, we build a

comparisons, the algorithm e#ectively retrieved physical tem-

lookup table composed of the soil physical temperature, soil

perature on average for +* stations. The value of R-square is

moisture content, brightness temperature at +*.0/ GHz verti-

*././2, and the value of SEE is ... K.

῍ 257 ῍

As with our soil

Development of a Physically-based Soil Moisture Retrieval Algorithm for Spaceborne Passive Microwave Radiometers and its Application to AMSR-E

Fig. , Comparison of simulated brightness temperature with the ones observed by AMSR-E

ῌ 258 ῌ

Journal of The Remote Sensing Society of Japan

Vol. ,3 No. + (,**3)

Fig. - A sketch of the lookup table for the AMSR-E soil moisture retrieval algorithm

Fig. .

(a). Time series of retrieved soil moisture, observed soil moisture and precipitation

Fig. / (a). Time series of retrieved soil physical temperature and in situ observation

Fig. .

(b). Comparison of estimated soil moisture with in situ observation

Fig. / (b). Comparison of estimated soil physical temperature with in situ observation ῌ 259 ῌ

Development of a Physically-based Soil Moisture Retrieval Algorithm for Spaceborne Passive Microwave Radiometers and its Application to AMSR-E

moisture analysis, the overestimation of daily temperature

and Technology Corporation for Promoting Science and

variation can also be explained partly as the results of di#erent

Technology Japan and by the Japan Aerospace Exploration

observation depths.

Agency (JAXA). The authors express their great gratitude to them. /.

Conclusion References

Spatially distributed soil moisture information is an essential parameter for hydrological, meteorological and ecological studies. This paper presents a physically-based soil moisture retrieval algorithm which reliably estimates soil moisture content from AMSR-E data observed at the CEOP Mongolia reference site. The algorithm is based on a radiative transfer model, DMRT-AIEM, which includes both volume scattering e#ects of dry soil media and surface scattering e#ects at the air/soil interface. Parameters used in this DMRT-AIEM are best-fitting.

They are : rms height, correlation length and

particle size. A lookup table generated by running the forward model for all possible situations is used to retrieve soil moisture and soil physical temperature. The algorithm has been developed in a way that does not involve any procedures or parameters related to specific radiometers or sensors, making its use independent of the choice of radiometers and sensors. The DMRT-AIEM algorithm was evaluated using the AMSR-E match-up data set at the CEOP Mongolia site. The TB validation results demonstrate that the modified radiative transfer model is able to represent the satellite TB observations accurately. By comparing retrieved soil moisture and physical temperature with in situ observations, it is clear that the algorithm estimates both soil moisture and soil physical temperature accurately. Even though our present algorithm can provide reliable estimates of soil moisture and physical temperature, it could be further improved in some respects : (+) The RTM considers the vegetation as a homogenous layer covering the soil surface. But, as we know, for sparsely vegetated areas, the vegetation covers only part of the land surface. This issue is currently under investigation by Fujii-,῍. (,) Our method of parameter optimization may fail for a region in which in situ observation is not available. A new parameterization method has been developed by Yang et al.--῍ using a land surface model and satellite observation to optimize the land parameters. E#orts in these respects will contribute to improved algorithm performance and a broader region of its applicability. Acknowledgments This study was carried out as part of the Coordinated Enhanced Observing Period (CEOP) and Verification Experiment for AMSR/AMSR-E funded by the Japanese Science

+῍ Eni G. Njoku and D. Entekhabi : Passive microwave remote sensing of soil moisture, J. of Hydrology, +2., pp. +*+ῌ+,3, +330. ,῍ J.R. Wang and B. J. Choundhury : Remote sensing of soil moisture content over bare field at +.. GHz frequency, J. Geophys. Res., 20, pp. /,11ῌ/,2,, +32+. -῍ M. C. Dobson, F. T. Ulaby, M. T. Hallikainen, and M. A. El-Rayes : Microwave dielectric behavior of wet Soilῌpart II : dielectric mixing models, IEEE Trans. Geosci. And Remote Sens., GE-,- (+), pp. -/ῌ.0, +32/. .῍ B. J. Choudhury, T. J. Schmugge, A.Chang, and R.W. Newton : E#ect of surface roughness on microwave emission from soils, J. Geophys. Res., 2., pp. /033ῌ/1*/, +313. /῍ J.R. Wang and B. J. Choundhury : Remote sensing of soil moisture content over bare field at +.. GHz frequency, J. Geophys. Res., 20, pp. /,11ῌ/,2,, +32+. 0῍ J.-P.Wigneron, A. Chanzy, J.-C. Calvet, and N. Bruguier : A simple algorithm to retrieve soil moisture and vegetation biomass using passive microwave measurements over crop fields, Remote Sens. Environ.,/+, pp. --+ῌ-.+, +33/. 1῍ U. Wegm¨ uller and C. Ma ¨tzler : Rough bare soil reflectivity model, IEEE Trans. Geosci. Remote Sens., -1, pp. +-3+ῌ +-3/, +333. 2῍ A. K. Fung, Z. Q. Li, and K. S. Chen : Backscattering from a randomly rough dielectric surface, IEEE Trans. Geosci. And Remote Sens., -*, pp. -/0ῌ-03, +33,. 3῍ K. S. Chen, T. D. Wu, L. Tsang, Q. Li, J. C. Shi, and A. K. Fung : Emission of rough surfaces calculated by the integral equation method with comparison to three-dimensional moment method Simulations, IEEE Trans. Geosci. And Remote Sens., .+, pp. 3*ῌ+*+, ,**-. +*῍ J. Shi, L. Jiang, L. Zhang, K-S. Chen, J-P. Wigneron, and A. Chanzy : A parameterized multifrequency-polarization surface emission model. IEEE Trans. Geosci. And Remote Sens., .- (+,), pp. ,2-+ῌ,2.+, ,**/. ++῍ F. Ulaby, M. Razani, and M. Doboson : E#ects of vegetation cover on the microwave radiometric sensitivity to soil moisture, IEEE Trans. Geosci. And Remote Sens., ,+, pp. /+ῌ 0+, +32-. +,῍ S. Paloscia and P. Pampaloni : Microwave polarization index for monitoring vegetation growth, IEEE Trans. Geosci. Remote Sens., ,0 (/), pp. 0+1ῌ0,+, +322. +-῍ T. J. Jackson and T. J. Schmugge : Vegetation e#ects on the microwave emission of soils, Remote Sens. Environ., ,0, pp. ,*-ῌ,+,, +33+. +.῍ T. J. Jackson : Measuring surface soil moisture using passive microwave remote sensing, Hydro.Process., 1, pp. +-3ῌ+/,, +33-.

ῌ 260 ῌ

Journal of The Remote Sensing Society of Japan

+/῍ Eni G Njoku : AMSR land surface parameters algorithm theoretical basis document (version -.*), Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA, +333. +0῍ Eni G. Njoku, Thomas J. Jackson, Venkataraman Lakshmi, Tsz K. Chan, and Son V. Nghiem : Soil moisture retrieval from AMSR-E, IEEE Trans. Geosci. And Remote Sens., .+, pp. ,+/ῌ,,3, ,**-. +1῍ S. Paloscia, G. Macelloni, E. Santi, and T. Koike : A multifrequency algorithm for the retrieval of soil moisture on a large scale using microwave data from SMMR and SSM/I satellites, IEEE Trans. on Geosci. And Remote Sens., -3, pp. +0//ῌ+00+, ,**+. +2῍ S. Paloscia, G. Macelloni, and E. Santi : Soil moisture estimates from AMSR-E brightness temperatures by using a dual-frequency algorithm, IEEE Trans. on Geosci. And Remote Sens., .+ (++), pp. -+-/ῌ-+.., ,**0 +3῍ A. Shibata, K. Imaoka, and T. Koike : AMSR/AMSR-E level , and - algorithm developments and data validation plans of NASDA. IEEE Trans. Geosci. Remote Sens., .+, pp. +3/ῌ,*-, ,**-. ,*῍ T.B. Mo, B. J. Choudhury, T. J Schmugge, J.R. Wang, and T. J. Jackson : A model for microwave emission from vegetation covered fields. J. Geophys. Res., 21, pp. ++,,3ῌ++,-1, +32, ,+῍ B. Wen, L. Tsang, D. P. Winebrenner, and A. Ishimaru : Dense media radiative transfer theory : comparison with experiment and application to microwave remote sensing and polarimetry, IEEE Trans. on Geosci. Remote Sens., ,2, pp. .0ῌ/3, +33*. ,,῍ L. Tsang and J.A. Kong : Scattering of Electomagnetic Waves : Advanced Topics, Wiley, New York, ,**+. ,-῍ L. Tsang and J. A. Kong : Theory for thermal microwave emission from a bounded medium containing spherical scatterers, J. Appl. Phys., .2, pp. -/3-ῌ-/33, +311

Vol. ,3 No. + (,**3)

,.῍ G. Liu : A fast and accurate model for microwave radiance calculations, J. Meteor. Soc. Japan,, 10 (,), pp. --/ῌ-.-, +332. ,/῍ L. C. Henyey and J. L. Greenstein : Di#use radiation in the galaxy, Astrophys. J., 3-, pp. 1*ῌ2-, +3.+ ,0῍ T. Koike, T. Tsukamoto, T. Kumakura and M. Lu : Spatial and seasonal distribution of surface wetness derived from satellite data, Proc. Inter. workshop on macro-scale hydrological modeling, pp. 21ῌ30, +330. ,1῍ H. Fujii and T. Koike : Development of a TRMM/TMI algorithm for precipitation in the Tibetan Plateau by considering e#ects of land surface emissivity, J. Meteor. Soc. Japan,, 13, pp. .1/ῌ.2-, ,**+. ,2῍ T. Koike : The Coordinated Enhanced Observing Period ῌ an initial step for integrated global water cycle observation, WMO Bull., /- (,), pp. +ῌ2, ,**.. ,3῍ Togtohyn Chuluun and D. S. Ojima : Simulation studies of grazing in the Mongolian steppe, Proc. /th Inter. Rangeland Congress, Society of Range Management, Denver, CO, USA, +330. -*῍ F. T. Ulaby, R. K. Moore, and A. K. Fung : Microwave Remote Sensing : Active and Passive, Artech House, Norwood, MA, +320. -+῍ J. R. Kendra and K. Sarabandi : A hybrid experimental theoretical scattering model for dense random media, IEEE Trans. on Geosci. Remote Sens., -1, pp. ,+ῌ-/, +333. -,῍ H. Fujii : Development of a microwave radiative transfer model for vegetated land surface based on comprehensive in situ observations, PhD Thesis, The University of Tokyo, ,**/ --῍ K. Yang, T. Watanabe, T. Koike, X. Li, H. Fujii, K. Tamagawa, Y. Ma and H. Ishikawa : An auto-calibration system to assimilate AMSR-E data into a land surface model for estimating soil moisture and surface energy budget, J. Meteor. Soc. Japan, 2/A, pp. ,,3ῌ,.,, ,**1.

ῌ 261 ῌ

Development of a Physically-based Soil Moisture Retrieval Algorithm for Spaceborne Passive Microwave Radiometers and its Application to AMSR-E

ῌ Hui

ῌ Katsunori

Lu

Dr. Hui Lu (born in March, +311) received his B. Eng. and M. Eng. from Tsinghua University, Beijing, China, in ,*** and ,**-, respectively. He received his D. Eng. degree in ,**0 from the University of Tokyo, Tokyo, Japan. Dr. Lu is currently a research fellow at the University of Tokyo. His research interests are : radiative transfer model development through field experiments and numerical simulations, passive microwave remote sensing of land surface parameters, and land data assimilation system development. He is a member of JSCE, IEEE, AGU and RSSJ. E-mail : lu῍hydra.t.u-tokyo.ac.jp ῌ Toshio

Koike

Prof. Toshio Koike (born in November +3/0) received his B. Eng., M. Eng. and D. Eng. from the University of Tokyo, Japan in +32*, +32, and +32/ respectively. He was a research associate at the University of Tokyo in +32/ and was appointed assistant professor, Tokyo University in +320. He was appointed Assoc. Prof. by Nagaoka University in +33,. He has been a Professor in the University of Tokyo in the department of civil engineering from ,*** to date. He is the lead scientist of the Coordinated Enhanced Observing Period (CEOP) project. His research interests are in Hydrology, Water Resources, Satellite Remote Sensing, Climate Change and Asian Monsoon. He is a member of JSCE, IEEE, AGU and RSSJ. E-mail :

Tamagawa Mr. Katsunori Tamagawa received Bachelor and Master degrees in the field of Engineering in +33- and +33/, respectively, from the Nagaoka University of Technology, Niigata, Japan. He has been working at the University of Tokyo since ,*** and is now a Research Associate with the Earth Observation Data Integration and Fusion Research Initiative (EDITORIA), in the University of Tokyo, Japan. He concurrently serves as Data Manager of the Coordinated Earth Observations Project (CEOP) and the Asian Water Cycle Initiative(AWCI) data archive center in the University of Tokyo, mainly to handle in situ observation data for the Asian region. He is a member of JSCE and IEEE. E-mail : ῌ Tetsu

Ohta

Mr. Tetsu Ohta (born in August, +31/) received his B. Eng. and M. Eng. from Nagaoka University of Technology, Nagaoka, Japan, in +332 and ,***, respectively. He is currently a Research Associate with the River and Environmental Engineering Laboratory, Department of Civil Engineering, the University of Tokyo. His research interests are passive remote sensing of soil moisture and data processing. He is a member of JSCE. E-mail :

ῌ Hideyuki Fujii Dr. Hideyuki Fujii (See introduction in this issue.) E-mail :

ῌ 262 ῌ