Estimate of Phase Transition Water Content in ... - Semantic Scholar
4248
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 12, DECEMBER 2010
Estimate of Phase Transition Water Content in Freeze–Thaw Process Using Microwave Radiometer Lixin Zhang, Member, IEEE, Tianjie Zhao, Student Member, IEEE, Lingmei Jiang, Associate Member, IEEE, and Shaojie Zhao, Student Member, IEEE
Abstract—Ground surface freeze–thaw cycles caused by changes in solar radiation have a great impact on soil–air water heat exchanges due to the phase transition of pore water. This influence should not be ignored in the land surface process and global environment change studies because of its large extent and the rapid changes in daily and seasonal frozen ground. The key index for evaluating the influence intensity is the content of water–ice phase transition in soil pores at the ground surface. In this paper, a data set was generated by observing field experiments and physical model simulations based on the configuration of the Advanced Microwave Scanning Radiometer–EOS (AMSR-E). The results showed that microwave radiation from freezing/thawing soil has an obvious correlation to the phase transition process of soil water. A large change in soil surface emissivity was shown after the freezing of soil. The magnitude of the difference in emissivity change is strongly related to the amount of water–ice phase transition. It can be shown that the higher the phase transition water content (PTWC), the greater the emissivity difference, and the higher the frequency, the smaller the emissivity difference. Based on an analysis of a large amount of random simulation data, an interesting characteristic was found, in that the emissivity difference in vertical polarization at each frequency is nearly proportional to the phase transition water content. Thus, a ratio index called Quasi-emissivity (Qe) was developed to eliminate temperature effects during retrieval. Using these clear rules, a physical statistical algorithm was put forth to estimate the phase transition water content. Finally, the inferred results by ground-based radiometer observation were compared with the ground truth. A satisfying agreement was achieved with a root mean square error of 0.0265 (v/v). This indicated that the microwave radiometer has a great potential in the measurement of PTWC. Index Terms—Advanced Microwave Scanning Radiometer– EOS (AMSR-E), freeze–thaw process, microwave radiometer, phase transition water content (PTWC).
I. I NTRODUCTION
F
REQUENT changes in the state of the soil surface (frozen or thawed) caused by changes in air temperature and solar radiation have a large impact on water and energy exchange Manuscript received September 3, 2009; revised February 23, 2010. Date of publication July 8, 2010; date of current version November 24, 2010. This work was supported in part by the Public Sector (Meteorology) Special Research Project GYHY200706044 and in part by the Chinese State Key Basic Research Project 2007CB714403. The work presented in this paper was performed at the State Key Laboratory of Remote Sensing Science, jointly sponsored by Beijing Normal University and Institute of Remote Sensing Applications, CAS, Beijing, China. This study was mainly supported by the special funds from China Meteorological Administration. The authors are with the School of Geography and Remote Sensing Science, Beijing Normal University, Beijing 100875, China (e-mail: lxzhang@ bnu.edu.cn). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TGRS.2010.2051158
processes between the soil and atmosphere. On one hand, it can be shown that the physical conditions of frozen soil and unfrozen soil are different. For example, compared to unfrozen soil, the heat transfer coefficient of frozen soil is high while the water transfer coefficient is low due to pore ice formation. On the other hand, phase transition of soil pore water during the freeze–thaw process causes the emission and absorption of a large amount of latent heat which delays the freeze–thaw process. These factors cannot be ignored in land surface process and global environment change studies when considering their large extents and the intensely dynamic changes in daily, seasonal, and permanent frozen ground. Phase transition water content (PTWC) is the amount of water that freezes or melts during soil freeze/thaw processes. This is an important indicator of the intensity of soil freezing/thawing and is the crucial parameter that influences soil freeze–thaw erosion and land surface energy balancing. Microwave remote sensing has the potential to monitor the soil freeze–thaw process and the accompanying water content of the phase transition at the upper layer. Ground experiments and model simulations of the characteristics of microwave radiation from the soil surface during freezing and thawing present a basis for monitoring the soil freeze–thaw process with a satelliteborne microwave radiometer. Wegmüller [1] measured the brightness temperature (Tb) change during several freeze–thaw cycles using a ground-based microwave radiometer and established a semiempirical model for frozen soil microwave emissions. Zuerndorfer et al. [2]–[4] developed a freeze/thaw status classification algorithm based on radiobrightness data from the Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR). Judge et al. [5] adjusted the Zuerndorfer’s algorithm using meteorological data from the American prairie corresponding to Special Sensor Microwave/Imager (SSM/I) data. Most of the related research work discusses qualitative monitoring of the freeze/thaw status of soil rather than a quantitative estimation of its physical parameters. The soil-freezing process is essentially a phase transition process of liquid water in soil pores. The microwave emissions of soil are sensitive to soil liquid water content mainly due to the large differences in the dielectric constants of liquid water, dry soil, and ice. Zhang et al. [6], [7] developed a dielectric constant model for frozen soil based on the semiempirical dielectric constant model for a soil–water mixture established by Dobson et al. [8], [9]. This dielectric constant model of frozen soil was integrated with the AIEM model that was improved by Chen et al. [10] to simulate the microwave emission of frozen soil. This paper further contributes an algorithm to obtain the PTWC at exposed soil surface during freeze–thaw processes.
0196-2892/$26.00 © 2010 IEEE
ZHANG et al.: ESTIMATE OF PHASE TRANSITION WATER CONTENT
For this purpose, we carried out field experiments using a new Truck-mounted Multifrequency Microwave Radiometer (TMMR). An irrigated agriculture land was measured during one winter season including three major freeze–thaw processes. It is the first time to present high temporal resolution radiometric multichannel (X, Ku, and Ka bands with dual polarization) data showing the daily freezing and thawing process of seasonal frozen soil. Several models were investigated for the microwave radiation simulation and phase transition water inversion. Its estimation with remote sensing technique has not yet been studied in detail before, although it is rather crucial in surface energy balance. The object of this paper is to demonstrate the sensitivity of microwave emission to the PTWC, to develop a retrieval method to estimate phase transition water in freeze–thaw processes using microwave radiometer, to promote the quantitative remote sensing of ground surface parameters in a permafrost region, and to provide information for the studies of freeze–thaw erosion and land surface processes. Section II introduces the field experiments using TMMR. Section III first briefly describes the theory used to investigate the microwave radiation from the freeze/thaw ground. Then, an algorithm used to estimate the amount of phase transition is shown. The measurement results are compared and discussed in Section IV. Finally, the conclusions are summarized in Section V.
4249
Fig. 1. TMMR and test field. TABLE I SPECIFICATIONS OF THE TMMR
II. F IELD E XPERIMENTS The main goal of the field experiments is to investigate the microwave radiation characteristics of soil during the freezing and thawing processes and to validate the method for phase transition water estimation. The most important instrument is the TMMR. The field experiments are introduced in detail below. A. Surface Preparation and Characterization The experiment area is located in Xiguzhuang, Hebei province, China (Latitude, 38◦ 35 ; Longitude, 115◦ 12 ), which is a typical seasonal frozen soil area in China. We choose a piece of farmland with a width of 30 m and a length of 50 m. It is large enough for the 3-dB beam width when scanning with the viewing angle from 20◦ to 70◦ . This experiment lasted intermittently from November 27, 2007 to Mar. 4, 2008 to cover nearly all of the freeze–thaw season. Before the experiment, the field was ploughed with the soil layer turned over about 30 cm deep. An adequate irrigation was conducted to keep the soil profile saturated. Then, the soil was allowed to dry by natural dehydration through evaporation and infiltration. These made the surface smooth and homogenous. The radiometer was mounted on the hydraulic elevator and was lifted to 8 m above the ground. It was set to work facing the south to avoid the shadow of the truck. Unfortunately, the C-band receiver failed to work before the experiment, and the X-band receiver encountered the same problem after a few days’ measurement. Finally, we succeeded to obtain continuous observation data of three complete freeze–thaw cycles including X, Ku, and Ka bands. We took soil cores for soil moisture measurements at 10:00, 14:00, 18:00, and 22:00 on every day. Soil cores were collected
randomly around the view field of the radiometer using a soil borer. Then, soil moisture was measured using the traditional drying method at 0–5 cm, 5–15 cm, 15–25 cm, 25–35 cm, and 35–50 cm. Four thermistor probes were placed below the soil surface. A DT-85 data logger was used to collect the physical temperature of soil profiles. Soil temperature was sampled approximately below the surface 2.5 cm, 12.5 cm, 22.5 cm, 32.5 cm, and 42.5 cm. A cutting ring is used to measure soil bulk density of the soil surface layer (0–5 cm). The mean bulk density was obtained as 1.02 g/cm3 . The soil texture is classified as silt loam (sand: 13.06%, silt: 61.84%, clay: 25.10%) according to the U.S. Department of Agriculture classification scheme. The roughness is confirmed using a grid plate and a digital camera. The pictures of the radiometer and test field are shown in Fig. 1. B. Microwave Radiometer System The microwave radiometer system used in this experiment is a four-band (6.925, 10.65, 18.7, and 36.5 GHz) dual-polarize radiometer designed and made by Radiometer Physics Gmbh and Beijing Normal University. The receiver modules are fixed on a positioner, by which the azimuth and elevation angle of the viewing direction could be changed at a precision of 0.1◦ . The whole system is controlled by a host computer. Table I shows some of the specifications of TMMR. The receivers of each frequency are calibrated through a four-point calibration procedure before the experiment. In each receiver, there is an internal hot load, of which the physical
4250
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 12, DECEMBER 2010
temperature is measured by internal thermometer. The cold load is gained by pointing the antennas to a liquid nitrogen cooled black body target. The physical temperature of liquid nitrogen is calculated according to the air pressure. There is also a stable noise diode in each receiver, which is used to correct detector diode nonlinearity. By select or not select this noise source, another two calibration points could be gained. Therefore, four formulas could be used to solve for the gain and system noise of each receiver as well as the system nonlinearity and the noise temperature of the noise diode. Then, the internal hot load and the noise diode are used to calibrate the fluctuation of system gain and noise automatically and continuously during measurements. The system nonlinearity and the noise temperature of the noise diode are considered constant during automatic calibration. Fig. 2. Simulated unfrozen water content (y-axis) of different soil types versus soil physical temperature (x-axis).
III. M ETHODOLOGY In this section, related models are used to simulate the various freeze–thaw processes. The analysis shows that the simulation models can accurately predict the characteristics of the change, thus providing a reliable foundation for the following algorithm. An algorithm used to estimate the PTWC is developed according to the Advanced Microwave Scanning Radiometer–EOS (AMSR-E) configuration and characteristics of microwave radiation of soil in freeze–thaw processes. A. Theory 1) Physical Properties of Frozen Soil: Even when cooled below 0 ◦ C, some water in soil remains unfrozen. This unfrozen water plays an important role in the freezing and thawing processes of the active layer in permafrost area. It directly affects the processes of water migration during freezing and thawing. It is also related to the thermal properties of the permafrost. The amount of unfrozen water is mainly dependent on the characteristics of soil and physical temperature. The adsorption forces and curvature at the soil particle surfaces is the main reason why water does not freeze at below freezing temperature. These processes can be correlated to the specific surface area (SSA) of the soil, which is the total surface area of a 1-g soil. The greater the SSA is, the greater the binding force on water is caused by soil particles. This makes the soil not prone to freeze. On the other side, with the temperature deceases, the water overcomes the adsorption forces from soil particles to further freeze, and the remaining unfrozen water content decreases. The relationship that the unfrozen water depends on SSA and temperature can be expressed using an empirical model [11] mu = a|Ts |−b
S = 0.042 + 4.23clay% + 1.12silt% − 1.16sand%.
(1)
where mu is the unfrozen gravimetric soil moisture (g/g), a and b are constants determined by soil properties (correlated to SSA here), Ts is soil temperature in degrees Celsius. S is soil specific area (SSA, m2 /g). The soil freezing and thawing processes occur at the interface between the liquid and solid phases. The adsorption forces are directly determined by the total surface area of soil particles.
(2)
In Fig. 2, we demonstrate three curves that the unfrozen water varies with temperature. At the same temperature, the unfrozen water content behaves clay soil > silt soil > sandy soil. In other words, the unfrozen water increases with the SSA of soil. As we are able to predict the liquid (unfrozen) water content of frozen soil, it is not adequate to simulate the microwave radiation from frozen soil. At microwave bands, electromagnetic signals from the ground surface with a certain roughness are mainly determined by the dielectric constant of soils. An important model should be used to calculate the dielectric constant of the frozen soil. Based on the previous studies, Dobson developed a semiempirical model to calculate the dielectric constant of soil [9]. It was concluded that the dielectric constant of wet soil is a function of its volumetric content and of the soil’s textural composition. The model has been widely used in the field of microwave remote sensing. The expression [9] is α β α εα m = 1 + (ρb /ρs ) (εs − 1) + mv εf w − mv
ln a = 0.5519 ln S + 0.2618 ln b = −0.264 ln S + 0.3711
However, SSA varies greatly among soils due to differences in particle-size distribution, mineralogy, and organic matter content. It is primarily associated with the clay fraction because of the high degree of subdivision. Ersahin [12] studied the soils with the texture varied from clay to sandy loam. It is concluded that clay% was the most successful variable in estimating SSA. To evaluate the combined effect of soil textural components, the sand, silt, and clay contents were used together in a multiple regression analysis. An equation for predicting SSA was established [12]
(3)
where εm , εs , and εf w are the relative permittivity of wet soil, soil solids, and free water, respectively. ρb and ρs are bulk density and specific density, respectively (g/cm3 ). mv is the volumetric soil moisture (v/v). α is a constant shape factor (optimized as 0.65), and β is a soil textural composition dependent coefficient. As described above, not all liquid water transforms into ice. Frozen soil is a mixture of soil, water, and ice. Zhang [7] added a new term to describe the contribution of ice fraction
ZHANG et al.: ESTIMATE OF PHASE TRANSITION WATER CONTENT
4251
Fig. 3. Simulated brightness temperatures (y-axis) versus soil physical temperature (x-axis) during freeze–thaw processes corresponding to increasing soil moisture, the dashed line represents the freezing point.
to the dielectric constant of frozen soil. It can be used for various soil types with inputs of soil texture, bulk density, soil moisture, and temperature. The simulated results were validated by experimental data obtained by an Agilent PNA Network Analyzer E8362B. The expression [7] can be written as α εα mf = 1 + (ρb /ρs ) (εs − 1) α + mβvu εα f w − mvu + mvi εi − mvi
(4)
where the subscripts s, i, f w, vu, and vi refer to solid soil, ice, free water, volumetric unfrozen water, and volumetric ice content, respectively. 2) Microwave Radiation from Frozen Soil: Understanding of the interaction between soil surface and electromagnetic waves has progressed significantly in recent decades. The simplest model of soil surface is the model of uniform halfspace with a flat surface, which is given by Fresnel formulas. However, natural soil surface always have a roughness. In general, we use root mean square height to measure the vertical roughness and correlation length to measure the horizontal roughness. In this case, the total reflectivity should be expressed in terms of the bistatic scattering coefficients. Several models have been developed to describe the microwave scattering behavior of random rough soil surfaces, including the Physical Optical Model (POM), Geometric Optical Model (GOM), Small Perturbation Model (SPM), and Integral Equa-
tion Model (IEM) [14]. Among the models, POM, GOM, and SPM are applicable for a limited range of roughness and frequency [13]–[16]. The IEM model can simulate the emissions of microwave with the information that considers the roughness correlation function of the soil surface [15]. It has been shown that the IEM with a transitional function can provide very good backscattering coefficients results for a wide range of surface roughness parameters, by comparing them with the Monte Carlo simulations and laboratory-controlled microwave experimental measurements [10], [15]. However, it is also shown that the original IEM produces larger errors in the order of tens of degrees Kelvin in brightness temperature, which is unacceptable for passive remote sensing [15]–[17]. A modified IEM, known as Advanced IEM (AIEM) [10] has been developed in recent years to calculate the emissivity. The results simulated at L-band with small root mean square height and correlation lengths agree with those simulated using 3-D method of moment Monte Carlo [10]. Shi [17] compared AIEM results with experimental data at AMSR-E frequencies and over a wider range of roughness. The results showed that the predicted surface emissivity is very similar to that of the experimental data over a large number of observations and the characteristics of the surface emission signals in their frequency and polarization dependences. Thus, the volumetric liquid water and ice content of soils with various textures can be calculated using the unfrozen water
4252
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 12, DECEMBER 2010
content predicting model. Then, the permittivity of frozen soil is estimated using the soil–water–ice mixture dielectric constant model. By assuming the frozen soil performs as a homogeneous semiinfinite space, we could use the AIEM model to simulate the brightness temperature observed by the ground radiometer. It can be approximately described by the following expression: Tb = Tgnd (1 − rgnd ) + Tatm,↓ rgnd + 2.7 K · tatm rgnd
TABLE II WIDE RANGES OF GROUND KEY PARAMETERS
(5)
where rgnd is the ground surface reflectivity calculated by the AIEM, Tgnd is the effective physical temperature of ground, tatm is the atmospheric transmissivity, and Tatm,↓ is downwelling atmospheric brightness temperatures. 2.7 K refers to the cosmic background radiation. These atmospheric parameters are calculated using formulas presented in the HUT snow emission model [18]. The brightness temperature of freeze/thaw soil was simulated with the in situ parameters of soil texture and roughness as basic inputs. Because the goal is to investigate the influence of soil moisture and physical temperature on the microwave radiation, the inputs of soil moisture and physical temperature are set to cover a wide range, with soil moisture changing from 0.1 to 0.4 (v/v), and soil temperature from −15 ◦ C to 15 ◦ C. The viewing angle was set as 55◦ according to the configuration of AMSR-E. The simulated results are shown in Fig. 3. When the initial soil moisture is as low as 0.1 (v/v), brightness temperatures changes are mainly determined by the physical temperature. When the soil moisture is large enough, there are obvious changes in brightness temperatures after soil freezing. This is caused by the decrease of unfrozen water during the soil freezing process. The changes of H polarization are larger than those of V polarization at the same frequency and with the same initial soil moisture. With the increase in initial moisture content, the changes at every channel also increase. In fact, the amplitude of changes is related to the PTWC. Before and after soil freezing, the absolute brightness temperature of V polarization is higher than that of H polarization at each frequency. When freezing occurs, the brightness temperature at V polarization of 36.5 GHz has a best stability. It has the best correlation with physical temperature. From the above simulations, we could see that the H polarization is more sensitive to phase transition water than V polarization, and that low frequencies are more sensitive than high frequencies. The phase transition of water has the smallest effects on changing the brightness temperature at 36.5 GHz for V polarization. B. Inversion The AMSR-E instrument measures radiation at six frequencies in the range 6.925–89 GHz, all with dual polarization. The antenna scans conically at a fixed incident angle of 55◦ , providing near-global coverage in two days or less. The spatial resolution varies from approximately 60 km at 6.925 GHz to 5 km at 89 GHz. The Aqua orbit is sun-synchronous with equator crossings at 13:30 (ascending) and 1:30 (descending) local solar time. This is advantageous when monitoring the freeze–thaw cycle and is the base for algorithm development to estimate the PTWC using two phase data that correspond to the ascending and descending data.
Fig. 4. Difference in quasi-emissivity (x-axis) versus PTWC (y-axis) at 10.65 GHz of V-polarization.
To estimate the phase transition water, the emissivity difference between frozen and thawed soils should be observed. However, this is very difficult when the surface physical temperature is uncertain. McFarland [19] carried out a number of studies of surface physical temperature inversion using microwave data and showed that Tb37V is the most suitable channel to estimate surface temperature. This is also approved by the above simulations. With the Tb36.5V indicating the change of physical temperature, a ratio of Tb at the other channels and Tb36.5V can be considered as a Quasi-emissivity (Qe). It can be used to establish the algorithm instead of actual emissivity Qef,p =
Tbf,p . Tb36.5,v
(6)
So, the Quasi-emissivity change in the freeze–thaw process can be calculated using AMSR-E data according to the equation ΔQef,p =
TbF,f,p TbT,f,p − . TbF,36.5,v TbT,36.5,v
(7)
Here, p = h or v polarization, f is frequency, F and T represent frozen soil and thawed soil, respectively. Tb is brightness temperature observed by the radiometer. The freeze/thaw status of soil can be obtained by the ground stations records, or estimated by MODIS land surface temperatures (LST) products. Using a freeze/thaw discrimination algorithm is also an effective method [5], [20].
ZHANG et al.: ESTIMATE OF PHASE TRANSITION WATER CONTENT
Fig. 5.
4253
Physical temperature measurements of soil in the field experiment at Xiguzhuang. The dashed line represents the freezing point.
Mass data is simulated randomly with the variation ranges of parameters in Table II, which contains the greatest number of soil conditions in China. The relationship between the Quasiemissivity change and PTWC is analyzed using the simulation database. The V-polarization data is found much suitable for retrieval due to uniform sensitivity. A good linear relationship exists between phase transition water and the difference in Quasi-emissivity. It is well known that 6.925 and 10.65 GHz (low frequency) are the channels with more sensitivity to soil moisture. However, a survey deployed by Li [21] showed that the radio frequency interference is widespread in the C-band (6.925 GHz) channels. So, in this paper, the retrieval algorithm is established at 10.65 GHz with V-polarization. Based on mass data simulated using physical models, a statistical retrieval model is presented through regression to describe the relationship between PTWC and differences in Quasi-emissivity as follows. The correlation coefficient can reach 0.9061 (see Fig. 4). mvpt = 3.0185ΔQe10.65,v + 0.0008
(8) Fig. 6. Soil moisture measurements in the field experiment at Xiguzhuang.
where mvpt stands for volumetric phase transition water during the freeze–thaw process, and ΔQe10.65,v is the Quasiemissivity difference at 10.65V. IV. R ESULTS AND D ISCUSSION The measured physical temperatures of soil profile are shown in Fig. 5. We can see the surface soil (0–5 cm) occurred freeze–thaw cycles which would determine the microwave radiation. The soil below the surface about 10 cm was always frozen, and the deeper soil was always thawed. The averaged soil moisture of soil profile was plotted in Fig. 6. It can be seen that the water was migrated to the surface soil due to the effect of the continuous freeze–thaw processes since the moisture of surface soil was bigger than the deeper soil. The continuous measurements from December 19, 2007 to 21 at Xiguzhuang are shown in Fig. 7. As the soil moisture content was as large as 0.28 (v/v). Thus, there must be a great deal of water phase changed to alter the dielectric properties of soil. It can be seen that the brightness temperatures of every channel went down rapidly when the physical temperature rose above the freezing point. This is because the thawing of soil causes the permittivity to increase and emissivity to decrease significantly.
When the ground surface was completely frozen or thawed, the brightness temperatures only varied with physical temperature. The reliability of the above retrieval method was confirmed based on the continuous radiometer measurements in the field experiments. The soil moisture used here is an average of the measurements at 0–5 cm. The soil physical temperature was processed corresponding to the step of Tb measurements. With the measured initial soil moisture and physical temperature, the PTWC can be calculated as the ground truth. And, the phase transition water can also be inferred from observation Tb using function (7) and (8). The measured and estimated PTWC are compared in Fig. 8 during three days. It shows that the PTWC can be detected clearly using the developed physical-statistical model with a root mean square error of 0.0265 (v/v). This validation results are satisfying, and confirm that the microwave radiometer have the ability to measure the PTWC in soil freeze–thaw process. A physically based statistical method for estimating the PTWC during freeze–thaw process is presented in this paper. Our objective is to develop a simple approach that could make use of two phase observation from radiometer, and to demonstrate its sensitivity to the phase transition water, which
4254
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 12, DECEMBER 2010
Fig. 7. Brightness temperature measurements in the field experiment at Xiguzhuang. The dashed line represents the freezing point.
Fig. 8. Comparison of measured and estimated PTWC. The dashed line represents the freezing point.
can help us further enhance our understanding of frozen soil. This method also has limitations: the input parameters for modeling are site-specific and requires auxiliary data or other methods to obtain the freeze–thaw state of soil. To further verify the method, the next step is to find a relatively large and flat exposed surface, and carry out simultaneous measurements with AMSR-E. So, we can see the applicability when using satellite data. Researches that should be promoted in the future include: 1) evaluating land cover (especially the vegetation in the winter and alpine meadow) effects on the detecting of phase transition water in soil; 2) using the ancillary data to consider the effective PTWC under a more complex ground surface situation over a pixel-scale; and 3) applying retrieved PTWC to soil freeze–thaw erosion and surface energy balance researches. On the other hand, the soil moisture and ocean salinity mission [22] is available for providing L-band microwave measurement. It is considered optimum for soil moisture monitoring due to greater sensitivity. Also, this will be more helpful for phase transition water monitoring that needs further work.
mounted radiometer. Several interesting conclusions related to the PTWC are found.
V. S UMMARY
ACKNOWLEDGMENT
A randomly simulated data set of microwave emission from soil during the freeze–thaw process was established by integrating the dielectric constant model of frozen soil and AIEM with confirmed reliability. Based on the regression analysis, a simple statistical method was developed for estimating the PTWC. The method was successfully applied over a farmland using a truck-
The authors would like to thank J. C. Shi and K. S. Chen for their support with the model codes. Special thanks to S. Chang, B. Li, W. P. Xing, Y. Zhen, Z. Y. Zhang, L. M. Zhang, and H. X. Qi for their contribution to the field experiments. The authors also wish to thank the three reviewers for their constructive comments that help improve the manuscript.
1) From both of observation and simulation, there is an obvious change in surface emissivity after the freezing of soil. The amplitude of change increases with initial soil moisture at both polarizations. 2) The intrinsic impact on changes in emissivity is caused by the phase change of soil moisture during freezing and thawing rather than total soil moisture. The relationship between differences in Quasi-emissivity and phase transition water content can be expressed as a linear function for v polarization under the configuration of AMSR-E. 3) Based on the conclusions above, an algorithm was developed to estimate the phase transition water content using Quasi-emissivity at 10.65 GHz and the retrieved results were compared with the ground-truth measurements. The good agreements confirmed the ability of microwave radiometer for phase transition water content estimation.
ZHANG et al.: ESTIMATE OF PHASE TRANSITION WATER CONTENT
R EFERENCES [1] U. Wegmüller, “The effect of freezing and thawing on the microwave signatures of bare soil,” Remote Sens. Environ., vol. 33, no. 2, pp. 123–135, Aug. 1990. [2] B. W. Zuerndorfer and A. W. England, “Radiobrightness decision criteria for freeze/thaw boundaries,” IEEE Trans. Geosci. Remote Sens., vol. 30, no. 1, pp. 89–102, Jan. 1992. [3] B. W. Zuerndorfer, A. W. England, C. M. Dobson, and F. T. Ulaby, “Mapping freeze/thaw boundaries with SMMR data,” J. Agric. For. Meteorol., vol. 52, no. 1/2, pp. 199–225, Aug. 1990. [4] B. W. Zuerndorfer, A. W. England, and G. H. Wakefield, “The radiobrightness of freezing terrain,” in Proc. IGARSS, 1989, vol. 5, pp. 2748–2751. [5] J. Judge, J. F. Galantowicz, A. W. England, and P. Dahl, “Freeze/thaw classification for prairie soils using SSM/I radiobrightness,” IEEE Trans. Geosci. Remote Sens., vol. 35, no. 4, pp. 827–832, Jul. 1997. [6] L. X. Zhang, J. C. Shi, Z. J. Zhang, and K. G. Zhao, “The estimation of dielectric constant of frozen soil-water mixture at microwave bands,” in Proc. IGARSS, 2003, vol. 4, pp. 2903–2905. [7] L. X. Zhang, K. G. Zhao, Y. Zhu, and B. Qin, “Simulated radiation characteristics of frozen soil surface at typical microwave bands,” in Proc. IGARSS, 2004, vol. 6, pp. 4310–4313. [8] M. T. Hallikainen, F. T. Ulaby, and M. C. Dobson, “Microwave dielectric behavior of wet soil—Part I: Empirical models and experimental observations,” IEEE Trans. Geosci. Remote Sens., vol. GRS-23, no. 1, pp. 25–34, Jan. 1985. [9] M. C. Dobson, F. T. Ulaby, M. T. Hallikainen, and M. A. EI-Rayes, “Microwave dielectric behavior of wet soil—Part II: Dielectric mixing models,” IEEE Trans. Geosci. Remote Sens., vol. GRS-23, no. 1, pp. 35– 46, Jan. 1985. [10] 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. Remote Sens., vol. 41, no. 1, pp. 90–101, Jan. 2003. [11] D. M. Anderson and A. R. Tice, “Predicting unfrozen water contents in frozen soils from surface area measurements,” Highway Res. Rec., vol. 373, pp. 12–18, 1972. [12] S. Ersahin, H. Gunal, T. Kutlu, B. Yetgin, and S. Coban, “Estimating specific surface area and cation exchange capacity in soils using fractal dimension of particle-size distribution,” Geoderma, vol. 136, no. 3/4, pp. 588–597, Dec. 2006. [13] L. Tsang, J. A. Kong, and R. T. Shin, Theory of Microwave Remote Sensing. New York: Wiley-Interscience, 1985. [14] A. K. Fung, Microwave Scattering and Emission Models and Their Applications. Norwood, MA: Artech House, 1994. [15] A. K. Fung, Z. Li, and K. S. Chen, “Backscattering from a randomly rough dielectric surface,” IEEE Trans. Geosci. Remote Sens., vol. 30, no. 2, pp. 356–369, Mar. 1992. [16] K. S. Chen and A. K. Fung, “A comparison of backscattering models for rough surfaces,” IEEE Trans. Geosci. Remote Sens., vol. 33, no. 1, pp. 195–200, Jan. 1995. [17] J. C. Shi, L. M. Jiang, L. X. Zhang, K. S. Chen, J.-P. Wigneron, and A. Chanzy, “A parameterized multifrequency-polarization surface emission model,” IEEE Trans. Geosci. Remote Sens., vol. 43, no. 12, pp. 2831– 2840, Dec. 2005. [18] J. T. Pulliainen, J. Grandell, and M. T. Hallikainen, “HUT snow emission model and its applicability to snow water equivalent retrieval,” IEEE Trans. Geosci. Remote Sens., vol. 37, no. 3, pp. 1378–1390, May 1999. [19] M. J. McFarland, R. L. Miller, and C. M. U. Neale, “Land surface temperature derived from the SSM/I passive microwave brightness temperatures,” IEEE Trans. Geosci. Remote Sens., vol. 28, no. 5, pp. 839–845, Sep. 1990. [20] T. Zhang, R. L. Armstrong, and J. Smith, “Investigation of the nearsurface soil freeze–thaw cycle in the contiguous United States: Algorithm development and validation,” J. Geophys. Res., vol. 108, no. D22, p. 8860, Nov. 2003.
4255
[21] L. Li, E. G. Njoku, E. Im, P. S. Chang, and K. S. Germain, “A preliminary survey of radio-frequency interference over the U.S. in Aqua AMSR-E data,” IEEE Trans. Geosci. Remote Sens., vol. 42, no. 2, pp. 380–390, Feb. 2004. [22] Y. H. Kerr, P. Waldteufel, J.-P. Wigneron, J.-M. Martinuzzi, J. Font, and M. Berger, “Soil moisture retrieval from space: The Soil Moisture and Ocean Salinity (SMOS) mission,” IEEE Trans. Geosci. Remote Sens., vol. 39, no. 8, pp. 1729–1735, Aug. 2001.
Lixin Zhang (M’08) received the B.A. degree from the University of Lanzhou, Lanzhou, China, in 1988, and the M.A. and Ph.D. degrees in geography from the Institute of Glaciology and Geocryology, Lanzhou, in 1991, and 2000, respectively. He then joined the Research Center for Remote Sensing and GIS, Beijing Normal University, Beijing, China, as a Professor. His research interests are physical properties of frozen soil and inversion algorithms or criteria for monitoring physical parameters of soil from remote sensing data.
Tianjie Zhao (S’09) received the B.S. degree from Beijing Normal University, Beijing, China, in 2007, where he is currently working toward the Ph.D. degree in cartography and geography information systems. He is also currently a Visiting Scientist in the Agricultural Research Service, Hydrology and Remote Sensing Laboratory, U.S. Department of Agriculture, Beltsville, MD. His research interests are microwave remote sensing of soil moisture and physical parameters of frozen soil.
Lingmei Jiang (AM’09) was born in Zhejiang, China, on October 31, 1978. She received the B.S. degree in agricultural meteorology from the Nanjing Institute of Meteorology, Nanjing, China, in 2000, and the Ph.D. degree in geography from Beijing Normal University, Beijing, China, in 2005. Her research interests include microwave remote sensing and modeling and retrieval of snow and soil properties.
Shaojie Zhao (S’08) received the B.S. degree in geography from Beijing Normal University, Beijing, China, in 2006, where he is currently working toward the Ph.D. degree, major in microwave remote sensing. His current interests include passive microwave remote sensing, modeling, and retrieval of frozen soil properties.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 12, DECEMBER 2010
Estimate of Phase Transition Water Content in Freeze–Thaw Process Using Microwave Radiometer Lixin Zhang, Member, IEEE, Tianjie Zhao, Student Member, IEEE, Lingmei Jiang, Associate Member, IEEE, and Shaojie Zhao, Student Member, IEEE
Abstract—Ground surface freeze–thaw cycles caused by changes in solar radiation have a great impact on soil–air water heat exchanges due to the phase transition of pore water. This influence should not be ignored in the land surface process and global environment change studies because of its large extent and the rapid changes in daily and seasonal frozen ground. The key index for evaluating the influence intensity is the content of water–ice phase transition in soil pores at the ground surface. In this paper, a data set was generated by observing field experiments and physical model simulations based on the configuration of the Advanced Microwave Scanning Radiometer–EOS (AMSR-E). The results showed that microwave radiation from freezing/thawing soil has an obvious correlation to the phase transition process of soil water. A large change in soil surface emissivity was shown after the freezing of soil. The magnitude of the difference in emissivity change is strongly related to the amount of water–ice phase transition. It can be shown that the higher the phase transition water content (PTWC), the greater the emissivity difference, and the higher the frequency, the smaller the emissivity difference. Based on an analysis of a large amount of random simulation data, an interesting characteristic was found, in that the emissivity difference in vertical polarization at each frequency is nearly proportional to the phase transition water content. Thus, a ratio index called Quasi-emissivity (Qe) was developed to eliminate temperature effects during retrieval. Using these clear rules, a physical statistical algorithm was put forth to estimate the phase transition water content. Finally, the inferred results by ground-based radiometer observation were compared with the ground truth. A satisfying agreement was achieved with a root mean square error of 0.0265 (v/v). This indicated that the microwave radiometer has a great potential in the measurement of PTWC. Index Terms—Advanced Microwave Scanning Radiometer– EOS (AMSR-E), freeze–thaw process, microwave radiometer, phase transition water content (PTWC).
I. I NTRODUCTION
F
REQUENT changes in the state of the soil surface (frozen or thawed) caused by changes in air temperature and solar radiation have a large impact on water and energy exchange Manuscript received September 3, 2009; revised February 23, 2010. Date of publication July 8, 2010; date of current version November 24, 2010. This work was supported in part by the Public Sector (Meteorology) Special Research Project GYHY200706044 and in part by the Chinese State Key Basic Research Project 2007CB714403. The work presented in this paper was performed at the State Key Laboratory of Remote Sensing Science, jointly sponsored by Beijing Normal University and Institute of Remote Sensing Applications, CAS, Beijing, China. This study was mainly supported by the special funds from China Meteorological Administration. The authors are with the School of Geography and Remote Sensing Science, Beijing Normal University, Beijing 100875, China (e-mail: lxzhang@ bnu.edu.cn). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TGRS.2010.2051158
processes between the soil and atmosphere. On one hand, it can be shown that the physical conditions of frozen soil and unfrozen soil are different. For example, compared to unfrozen soil, the heat transfer coefficient of frozen soil is high while the water transfer coefficient is low due to pore ice formation. On the other hand, phase transition of soil pore water during the freeze–thaw process causes the emission and absorption of a large amount of latent heat which delays the freeze–thaw process. These factors cannot be ignored in land surface process and global environment change studies when considering their large extents and the intensely dynamic changes in daily, seasonal, and permanent frozen ground. Phase transition water content (PTWC) is the amount of water that freezes or melts during soil freeze/thaw processes. This is an important indicator of the intensity of soil freezing/thawing and is the crucial parameter that influences soil freeze–thaw erosion and land surface energy balancing. Microwave remote sensing has the potential to monitor the soil freeze–thaw process and the accompanying water content of the phase transition at the upper layer. Ground experiments and model simulations of the characteristics of microwave radiation from the soil surface during freezing and thawing present a basis for monitoring the soil freeze–thaw process with a satelliteborne microwave radiometer. Wegmüller [1] measured the brightness temperature (Tb) change during several freeze–thaw cycles using a ground-based microwave radiometer and established a semiempirical model for frozen soil microwave emissions. Zuerndorfer et al. [2]–[4] developed a freeze/thaw status classification algorithm based on radiobrightness data from the Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR). Judge et al. [5] adjusted the Zuerndorfer’s algorithm using meteorological data from the American prairie corresponding to Special Sensor Microwave/Imager (SSM/I) data. Most of the related research work discusses qualitative monitoring of the freeze/thaw status of soil rather than a quantitative estimation of its physical parameters. The soil-freezing process is essentially a phase transition process of liquid water in soil pores. The microwave emissions of soil are sensitive to soil liquid water content mainly due to the large differences in the dielectric constants of liquid water, dry soil, and ice. Zhang et al. [6], [7] developed a dielectric constant model for frozen soil based on the semiempirical dielectric constant model for a soil–water mixture established by Dobson et al. [8], [9]. This dielectric constant model of frozen soil was integrated with the AIEM model that was improved by Chen et al. [10] to simulate the microwave emission of frozen soil. This paper further contributes an algorithm to obtain the PTWC at exposed soil surface during freeze–thaw processes.
0196-2892/$26.00 © 2010 IEEE
ZHANG et al.: ESTIMATE OF PHASE TRANSITION WATER CONTENT
For this purpose, we carried out field experiments using a new Truck-mounted Multifrequency Microwave Radiometer (TMMR). An irrigated agriculture land was measured during one winter season including three major freeze–thaw processes. It is the first time to present high temporal resolution radiometric multichannel (X, Ku, and Ka bands with dual polarization) data showing the daily freezing and thawing process of seasonal frozen soil. Several models were investigated for the microwave radiation simulation and phase transition water inversion. Its estimation with remote sensing technique has not yet been studied in detail before, although it is rather crucial in surface energy balance. The object of this paper is to demonstrate the sensitivity of microwave emission to the PTWC, to develop a retrieval method to estimate phase transition water in freeze–thaw processes using microwave radiometer, to promote the quantitative remote sensing of ground surface parameters in a permafrost region, and to provide information for the studies of freeze–thaw erosion and land surface processes. Section II introduces the field experiments using TMMR. Section III first briefly describes the theory used to investigate the microwave radiation from the freeze/thaw ground. Then, an algorithm used to estimate the amount of phase transition is shown. The measurement results are compared and discussed in Section IV. Finally, the conclusions are summarized in Section V.
4249
Fig. 1. TMMR and test field. TABLE I SPECIFICATIONS OF THE TMMR
II. F IELD E XPERIMENTS The main goal of the field experiments is to investigate the microwave radiation characteristics of soil during the freezing and thawing processes and to validate the method for phase transition water estimation. The most important instrument is the TMMR. The field experiments are introduced in detail below. A. Surface Preparation and Characterization The experiment area is located in Xiguzhuang, Hebei province, China (Latitude, 38◦ 35 ; Longitude, 115◦ 12 ), which is a typical seasonal frozen soil area in China. We choose a piece of farmland with a width of 30 m and a length of 50 m. It is large enough for the 3-dB beam width when scanning with the viewing angle from 20◦ to 70◦ . This experiment lasted intermittently from November 27, 2007 to Mar. 4, 2008 to cover nearly all of the freeze–thaw season. Before the experiment, the field was ploughed with the soil layer turned over about 30 cm deep. An adequate irrigation was conducted to keep the soil profile saturated. Then, the soil was allowed to dry by natural dehydration through evaporation and infiltration. These made the surface smooth and homogenous. The radiometer was mounted on the hydraulic elevator and was lifted to 8 m above the ground. It was set to work facing the south to avoid the shadow of the truck. Unfortunately, the C-band receiver failed to work before the experiment, and the X-band receiver encountered the same problem after a few days’ measurement. Finally, we succeeded to obtain continuous observation data of three complete freeze–thaw cycles including X, Ku, and Ka bands. We took soil cores for soil moisture measurements at 10:00, 14:00, 18:00, and 22:00 on every day. Soil cores were collected
randomly around the view field of the radiometer using a soil borer. Then, soil moisture was measured using the traditional drying method at 0–5 cm, 5–15 cm, 15–25 cm, 25–35 cm, and 35–50 cm. Four thermistor probes were placed below the soil surface. A DT-85 data logger was used to collect the physical temperature of soil profiles. Soil temperature was sampled approximately below the surface 2.5 cm, 12.5 cm, 22.5 cm, 32.5 cm, and 42.5 cm. A cutting ring is used to measure soil bulk density of the soil surface layer (0–5 cm). The mean bulk density was obtained as 1.02 g/cm3 . The soil texture is classified as silt loam (sand: 13.06%, silt: 61.84%, clay: 25.10%) according to the U.S. Department of Agriculture classification scheme. The roughness is confirmed using a grid plate and a digital camera. The pictures of the radiometer and test field are shown in Fig. 1. B. Microwave Radiometer System The microwave radiometer system used in this experiment is a four-band (6.925, 10.65, 18.7, and 36.5 GHz) dual-polarize radiometer designed and made by Radiometer Physics Gmbh and Beijing Normal University. The receiver modules are fixed on a positioner, by which the azimuth and elevation angle of the viewing direction could be changed at a precision of 0.1◦ . The whole system is controlled by a host computer. Table I shows some of the specifications of TMMR. The receivers of each frequency are calibrated through a four-point calibration procedure before the experiment. In each receiver, there is an internal hot load, of which the physical
4250
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 12, DECEMBER 2010
temperature is measured by internal thermometer. The cold load is gained by pointing the antennas to a liquid nitrogen cooled black body target. The physical temperature of liquid nitrogen is calculated according to the air pressure. There is also a stable noise diode in each receiver, which is used to correct detector diode nonlinearity. By select or not select this noise source, another two calibration points could be gained. Therefore, four formulas could be used to solve for the gain and system noise of each receiver as well as the system nonlinearity and the noise temperature of the noise diode. Then, the internal hot load and the noise diode are used to calibrate the fluctuation of system gain and noise automatically and continuously during measurements. The system nonlinearity and the noise temperature of the noise diode are considered constant during automatic calibration. Fig. 2. Simulated unfrozen water content (y-axis) of different soil types versus soil physical temperature (x-axis).
III. M ETHODOLOGY In this section, related models are used to simulate the various freeze–thaw processes. The analysis shows that the simulation models can accurately predict the characteristics of the change, thus providing a reliable foundation for the following algorithm. An algorithm used to estimate the PTWC is developed according to the Advanced Microwave Scanning Radiometer–EOS (AMSR-E) configuration and characteristics of microwave radiation of soil in freeze–thaw processes. A. Theory 1) Physical Properties of Frozen Soil: Even when cooled below 0 ◦ C, some water in soil remains unfrozen. This unfrozen water plays an important role in the freezing and thawing processes of the active layer in permafrost area. It directly affects the processes of water migration during freezing and thawing. It is also related to the thermal properties of the permafrost. The amount of unfrozen water is mainly dependent on the characteristics of soil and physical temperature. The adsorption forces and curvature at the soil particle surfaces is the main reason why water does not freeze at below freezing temperature. These processes can be correlated to the specific surface area (SSA) of the soil, which is the total surface area of a 1-g soil. The greater the SSA is, the greater the binding force on water is caused by soil particles. This makes the soil not prone to freeze. On the other side, with the temperature deceases, the water overcomes the adsorption forces from soil particles to further freeze, and the remaining unfrozen water content decreases. The relationship that the unfrozen water depends on SSA and temperature can be expressed using an empirical model [11] mu = a|Ts |−b
S = 0.042 + 4.23clay% + 1.12silt% − 1.16sand%.
(1)
where mu is the unfrozen gravimetric soil moisture (g/g), a and b are constants determined by soil properties (correlated to SSA here), Ts is soil temperature in degrees Celsius. S is soil specific area (SSA, m2 /g). The soil freezing and thawing processes occur at the interface between the liquid and solid phases. The adsorption forces are directly determined by the total surface area of soil particles.
(2)
In Fig. 2, we demonstrate three curves that the unfrozen water varies with temperature. At the same temperature, the unfrozen water content behaves clay soil > silt soil > sandy soil. In other words, the unfrozen water increases with the SSA of soil. As we are able to predict the liquid (unfrozen) water content of frozen soil, it is not adequate to simulate the microwave radiation from frozen soil. At microwave bands, electromagnetic signals from the ground surface with a certain roughness are mainly determined by the dielectric constant of soils. An important model should be used to calculate the dielectric constant of the frozen soil. Based on the previous studies, Dobson developed a semiempirical model to calculate the dielectric constant of soil [9]. It was concluded that the dielectric constant of wet soil is a function of its volumetric content and of the soil’s textural composition. The model has been widely used in the field of microwave remote sensing. The expression [9] is α β α εα m = 1 + (ρb /ρs ) (εs − 1) + mv εf w − mv
ln a = 0.5519 ln S + 0.2618 ln b = −0.264 ln S + 0.3711
However, SSA varies greatly among soils due to differences in particle-size distribution, mineralogy, and organic matter content. It is primarily associated with the clay fraction because of the high degree of subdivision. Ersahin [12] studied the soils with the texture varied from clay to sandy loam. It is concluded that clay% was the most successful variable in estimating SSA. To evaluate the combined effect of soil textural components, the sand, silt, and clay contents were used together in a multiple regression analysis. An equation for predicting SSA was established [12]
(3)
where εm , εs , and εf w are the relative permittivity of wet soil, soil solids, and free water, respectively. ρb and ρs are bulk density and specific density, respectively (g/cm3 ). mv is the volumetric soil moisture (v/v). α is a constant shape factor (optimized as 0.65), and β is a soil textural composition dependent coefficient. As described above, not all liquid water transforms into ice. Frozen soil is a mixture of soil, water, and ice. Zhang [7] added a new term to describe the contribution of ice fraction
ZHANG et al.: ESTIMATE OF PHASE TRANSITION WATER CONTENT
4251
Fig. 3. Simulated brightness temperatures (y-axis) versus soil physical temperature (x-axis) during freeze–thaw processes corresponding to increasing soil moisture, the dashed line represents the freezing point.
to the dielectric constant of frozen soil. It can be used for various soil types with inputs of soil texture, bulk density, soil moisture, and temperature. The simulated results were validated by experimental data obtained by an Agilent PNA Network Analyzer E8362B. The expression [7] can be written as α εα mf = 1 + (ρb /ρs ) (εs − 1) α + mβvu εα f w − mvu + mvi εi − mvi
(4)
where the subscripts s, i, f w, vu, and vi refer to solid soil, ice, free water, volumetric unfrozen water, and volumetric ice content, respectively. 2) Microwave Radiation from Frozen Soil: Understanding of the interaction between soil surface and electromagnetic waves has progressed significantly in recent decades. The simplest model of soil surface is the model of uniform halfspace with a flat surface, which is given by Fresnel formulas. However, natural soil surface always have a roughness. In general, we use root mean square height to measure the vertical roughness and correlation length to measure the horizontal roughness. In this case, the total reflectivity should be expressed in terms of the bistatic scattering coefficients. Several models have been developed to describe the microwave scattering behavior of random rough soil surfaces, including the Physical Optical Model (POM), Geometric Optical Model (GOM), Small Perturbation Model (SPM), and Integral Equa-
tion Model (IEM) [14]. Among the models, POM, GOM, and SPM are applicable for a limited range of roughness and frequency [13]–[16]. The IEM model can simulate the emissions of microwave with the information that considers the roughness correlation function of the soil surface [15]. It has been shown that the IEM with a transitional function can provide very good backscattering coefficients results for a wide range of surface roughness parameters, by comparing them with the Monte Carlo simulations and laboratory-controlled microwave experimental measurements [10], [15]. However, it is also shown that the original IEM produces larger errors in the order of tens of degrees Kelvin in brightness temperature, which is unacceptable for passive remote sensing [15]–[17]. A modified IEM, known as Advanced IEM (AIEM) [10] has been developed in recent years to calculate the emissivity. The results simulated at L-band with small root mean square height and correlation lengths agree with those simulated using 3-D method of moment Monte Carlo [10]. Shi [17] compared AIEM results with experimental data at AMSR-E frequencies and over a wider range of roughness. The results showed that the predicted surface emissivity is very similar to that of the experimental data over a large number of observations and the characteristics of the surface emission signals in their frequency and polarization dependences. Thus, the volumetric liquid water and ice content of soils with various textures can be calculated using the unfrozen water
4252
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 12, DECEMBER 2010
content predicting model. Then, the permittivity of frozen soil is estimated using the soil–water–ice mixture dielectric constant model. By assuming the frozen soil performs as a homogeneous semiinfinite space, we could use the AIEM model to simulate the brightness temperature observed by the ground radiometer. It can be approximately described by the following expression: Tb = Tgnd (1 − rgnd ) + Tatm,↓ rgnd + 2.7 K · tatm rgnd
TABLE II WIDE RANGES OF GROUND KEY PARAMETERS
(5)
where rgnd is the ground surface reflectivity calculated by the AIEM, Tgnd is the effective physical temperature of ground, tatm is the atmospheric transmissivity, and Tatm,↓ is downwelling atmospheric brightness temperatures. 2.7 K refers to the cosmic background radiation. These atmospheric parameters are calculated using formulas presented in the HUT snow emission model [18]. The brightness temperature of freeze/thaw soil was simulated with the in situ parameters of soil texture and roughness as basic inputs. Because the goal is to investigate the influence of soil moisture and physical temperature on the microwave radiation, the inputs of soil moisture and physical temperature are set to cover a wide range, with soil moisture changing from 0.1 to 0.4 (v/v), and soil temperature from −15 ◦ C to 15 ◦ C. The viewing angle was set as 55◦ according to the configuration of AMSR-E. The simulated results are shown in Fig. 3. When the initial soil moisture is as low as 0.1 (v/v), brightness temperatures changes are mainly determined by the physical temperature. When the soil moisture is large enough, there are obvious changes in brightness temperatures after soil freezing. This is caused by the decrease of unfrozen water during the soil freezing process. The changes of H polarization are larger than those of V polarization at the same frequency and with the same initial soil moisture. With the increase in initial moisture content, the changes at every channel also increase. In fact, the amplitude of changes is related to the PTWC. Before and after soil freezing, the absolute brightness temperature of V polarization is higher than that of H polarization at each frequency. When freezing occurs, the brightness temperature at V polarization of 36.5 GHz has a best stability. It has the best correlation with physical temperature. From the above simulations, we could see that the H polarization is more sensitive to phase transition water than V polarization, and that low frequencies are more sensitive than high frequencies. The phase transition of water has the smallest effects on changing the brightness temperature at 36.5 GHz for V polarization. B. Inversion The AMSR-E instrument measures radiation at six frequencies in the range 6.925–89 GHz, all with dual polarization. The antenna scans conically at a fixed incident angle of 55◦ , providing near-global coverage in two days or less. The spatial resolution varies from approximately 60 km at 6.925 GHz to 5 km at 89 GHz. The Aqua orbit is sun-synchronous with equator crossings at 13:30 (ascending) and 1:30 (descending) local solar time. This is advantageous when monitoring the freeze–thaw cycle and is the base for algorithm development to estimate the PTWC using two phase data that correspond to the ascending and descending data.
Fig. 4. Difference in quasi-emissivity (x-axis) versus PTWC (y-axis) at 10.65 GHz of V-polarization.
To estimate the phase transition water, the emissivity difference between frozen and thawed soils should be observed. However, this is very difficult when the surface physical temperature is uncertain. McFarland [19] carried out a number of studies of surface physical temperature inversion using microwave data and showed that Tb37V is the most suitable channel to estimate surface temperature. This is also approved by the above simulations. With the Tb36.5V indicating the change of physical temperature, a ratio of Tb at the other channels and Tb36.5V can be considered as a Quasi-emissivity (Qe). It can be used to establish the algorithm instead of actual emissivity Qef,p =
Tbf,p . Tb36.5,v
(6)
So, the Quasi-emissivity change in the freeze–thaw process can be calculated using AMSR-E data according to the equation ΔQef,p =
TbF,f,p TbT,f,p − . TbF,36.5,v TbT,36.5,v
(7)
Here, p = h or v polarization, f is frequency, F and T represent frozen soil and thawed soil, respectively. Tb is brightness temperature observed by the radiometer. The freeze/thaw status of soil can be obtained by the ground stations records, or estimated by MODIS land surface temperatures (LST) products. Using a freeze/thaw discrimination algorithm is also an effective method [5], [20].
ZHANG et al.: ESTIMATE OF PHASE TRANSITION WATER CONTENT
Fig. 5.
4253
Physical temperature measurements of soil in the field experiment at Xiguzhuang. The dashed line represents the freezing point.
Mass data is simulated randomly with the variation ranges of parameters in Table II, which contains the greatest number of soil conditions in China. The relationship between the Quasiemissivity change and PTWC is analyzed using the simulation database. The V-polarization data is found much suitable for retrieval due to uniform sensitivity. A good linear relationship exists between phase transition water and the difference in Quasi-emissivity. It is well known that 6.925 and 10.65 GHz (low frequency) are the channels with more sensitivity to soil moisture. However, a survey deployed by Li [21] showed that the radio frequency interference is widespread in the C-band (6.925 GHz) channels. So, in this paper, the retrieval algorithm is established at 10.65 GHz with V-polarization. Based on mass data simulated using physical models, a statistical retrieval model is presented through regression to describe the relationship between PTWC and differences in Quasi-emissivity as follows. The correlation coefficient can reach 0.9061 (see Fig. 4). mvpt = 3.0185ΔQe10.65,v + 0.0008
(8) Fig. 6. Soil moisture measurements in the field experiment at Xiguzhuang.
where mvpt stands for volumetric phase transition water during the freeze–thaw process, and ΔQe10.65,v is the Quasiemissivity difference at 10.65V. IV. R ESULTS AND D ISCUSSION The measured physical temperatures of soil profile are shown in Fig. 5. We can see the surface soil (0–5 cm) occurred freeze–thaw cycles which would determine the microwave radiation. The soil below the surface about 10 cm was always frozen, and the deeper soil was always thawed. The averaged soil moisture of soil profile was plotted in Fig. 6. It can be seen that the water was migrated to the surface soil due to the effect of the continuous freeze–thaw processes since the moisture of surface soil was bigger than the deeper soil. The continuous measurements from December 19, 2007 to 21 at Xiguzhuang are shown in Fig. 7. As the soil moisture content was as large as 0.28 (v/v). Thus, there must be a great deal of water phase changed to alter the dielectric properties of soil. It can be seen that the brightness temperatures of every channel went down rapidly when the physical temperature rose above the freezing point. This is because the thawing of soil causes the permittivity to increase and emissivity to decrease significantly.
When the ground surface was completely frozen or thawed, the brightness temperatures only varied with physical temperature. The reliability of the above retrieval method was confirmed based on the continuous radiometer measurements in the field experiments. The soil moisture used here is an average of the measurements at 0–5 cm. The soil physical temperature was processed corresponding to the step of Tb measurements. With the measured initial soil moisture and physical temperature, the PTWC can be calculated as the ground truth. And, the phase transition water can also be inferred from observation Tb using function (7) and (8). The measured and estimated PTWC are compared in Fig. 8 during three days. It shows that the PTWC can be detected clearly using the developed physical-statistical model with a root mean square error of 0.0265 (v/v). This validation results are satisfying, and confirm that the microwave radiometer have the ability to measure the PTWC in soil freeze–thaw process. A physically based statistical method for estimating the PTWC during freeze–thaw process is presented in this paper. Our objective is to develop a simple approach that could make use of two phase observation from radiometer, and to demonstrate its sensitivity to the phase transition water, which
4254
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 12, DECEMBER 2010
Fig. 7. Brightness temperature measurements in the field experiment at Xiguzhuang. The dashed line represents the freezing point.
Fig. 8. Comparison of measured and estimated PTWC. The dashed line represents the freezing point.
can help us further enhance our understanding of frozen soil. This method also has limitations: the input parameters for modeling are site-specific and requires auxiliary data or other methods to obtain the freeze–thaw state of soil. To further verify the method, the next step is to find a relatively large and flat exposed surface, and carry out simultaneous measurements with AMSR-E. So, we can see the applicability when using satellite data. Researches that should be promoted in the future include: 1) evaluating land cover (especially the vegetation in the winter and alpine meadow) effects on the detecting of phase transition water in soil; 2) using the ancillary data to consider the effective PTWC under a more complex ground surface situation over a pixel-scale; and 3) applying retrieved PTWC to soil freeze–thaw erosion and surface energy balance researches. On the other hand, the soil moisture and ocean salinity mission [22] is available for providing L-band microwave measurement. It is considered optimum for soil moisture monitoring due to greater sensitivity. Also, this will be more helpful for phase transition water monitoring that needs further work.
mounted radiometer. Several interesting conclusions related to the PTWC are found.
V. S UMMARY
ACKNOWLEDGMENT
A randomly simulated data set of microwave emission from soil during the freeze–thaw process was established by integrating the dielectric constant model of frozen soil and AIEM with confirmed reliability. Based on the regression analysis, a simple statistical method was developed for estimating the PTWC. The method was successfully applied over a farmland using a truck-
The authors would like to thank J. C. Shi and K. S. Chen for their support with the model codes. Special thanks to S. Chang, B. Li, W. P. Xing, Y. Zhen, Z. Y. Zhang, L. M. Zhang, and H. X. Qi for their contribution to the field experiments. The authors also wish to thank the three reviewers for their constructive comments that help improve the manuscript.
1) From both of observation and simulation, there is an obvious change in surface emissivity after the freezing of soil. The amplitude of change increases with initial soil moisture at both polarizations. 2) The intrinsic impact on changes in emissivity is caused by the phase change of soil moisture during freezing and thawing rather than total soil moisture. The relationship between differences in Quasi-emissivity and phase transition water content can be expressed as a linear function for v polarization under the configuration of AMSR-E. 3) Based on the conclusions above, an algorithm was developed to estimate the phase transition water content using Quasi-emissivity at 10.65 GHz and the retrieved results were compared with the ground-truth measurements. The good agreements confirmed the ability of microwave radiometer for phase transition water content estimation.
ZHANG et al.: ESTIMATE OF PHASE TRANSITION WATER CONTENT
R EFERENCES [1] U. Wegmüller, “The effect of freezing and thawing on the microwave signatures of bare soil,” Remote Sens. Environ., vol. 33, no. 2, pp. 123–135, Aug. 1990. [2] B. W. Zuerndorfer and A. W. England, “Radiobrightness decision criteria for freeze/thaw boundaries,” IEEE Trans. Geosci. Remote Sens., vol. 30, no. 1, pp. 89–102, Jan. 1992. [3] B. W. Zuerndorfer, A. W. England, C. M. Dobson, and F. T. Ulaby, “Mapping freeze/thaw boundaries with SMMR data,” J. Agric. For. Meteorol., vol. 52, no. 1/2, pp. 199–225, Aug. 1990. [4] B. W. Zuerndorfer, A. W. England, and G. H. Wakefield, “The radiobrightness of freezing terrain,” in Proc. IGARSS, 1989, vol. 5, pp. 2748–2751. [5] J. Judge, J. F. Galantowicz, A. W. England, and P. Dahl, “Freeze/thaw classification for prairie soils using SSM/I radiobrightness,” IEEE Trans. Geosci. Remote Sens., vol. 35, no. 4, pp. 827–832, Jul. 1997. [6] L. X. Zhang, J. C. Shi, Z. J. Zhang, and K. G. Zhao, “The estimation of dielectric constant of frozen soil-water mixture at microwave bands,” in Proc. IGARSS, 2003, vol. 4, pp. 2903–2905. [7] L. X. Zhang, K. G. Zhao, Y. Zhu, and B. Qin, “Simulated radiation characteristics of frozen soil surface at typical microwave bands,” in Proc. IGARSS, 2004, vol. 6, pp. 4310–4313. [8] M. T. Hallikainen, F. T. Ulaby, and M. C. Dobson, “Microwave dielectric behavior of wet soil—Part I: Empirical models and experimental observations,” IEEE Trans. Geosci. Remote Sens., vol. GRS-23, no. 1, pp. 25–34, Jan. 1985. [9] M. C. Dobson, F. T. Ulaby, M. T. Hallikainen, and M. A. EI-Rayes, “Microwave dielectric behavior of wet soil—Part II: Dielectric mixing models,” IEEE Trans. Geosci. Remote Sens., vol. GRS-23, no. 1, pp. 35– 46, Jan. 1985. [10] 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. Remote Sens., vol. 41, no. 1, pp. 90–101, Jan. 2003. [11] D. M. Anderson and A. R. Tice, “Predicting unfrozen water contents in frozen soils from surface area measurements,” Highway Res. Rec., vol. 373, pp. 12–18, 1972. [12] S. Ersahin, H. Gunal, T. Kutlu, B. Yetgin, and S. Coban, “Estimating specific surface area and cation exchange capacity in soils using fractal dimension of particle-size distribution,” Geoderma, vol. 136, no. 3/4, pp. 588–597, Dec. 2006. [13] L. Tsang, J. A. Kong, and R. T. Shin, Theory of Microwave Remote Sensing. New York: Wiley-Interscience, 1985. [14] A. K. Fung, Microwave Scattering and Emission Models and Their Applications. Norwood, MA: Artech House, 1994. [15] A. K. Fung, Z. Li, and K. S. Chen, “Backscattering from a randomly rough dielectric surface,” IEEE Trans. Geosci. Remote Sens., vol. 30, no. 2, pp. 356–369, Mar. 1992. [16] K. S. Chen and A. K. Fung, “A comparison of backscattering models for rough surfaces,” IEEE Trans. Geosci. Remote Sens., vol. 33, no. 1, pp. 195–200, Jan. 1995. [17] J. C. Shi, L. M. Jiang, L. X. Zhang, K. S. Chen, J.-P. Wigneron, and A. Chanzy, “A parameterized multifrequency-polarization surface emission model,” IEEE Trans. Geosci. Remote Sens., vol. 43, no. 12, pp. 2831– 2840, Dec. 2005. [18] J. T. Pulliainen, J. Grandell, and M. T. Hallikainen, “HUT snow emission model and its applicability to snow water equivalent retrieval,” IEEE Trans. Geosci. Remote Sens., vol. 37, no. 3, pp. 1378–1390, May 1999. [19] M. J. McFarland, R. L. Miller, and C. M. U. Neale, “Land surface temperature derived from the SSM/I passive microwave brightness temperatures,” IEEE Trans. Geosci. Remote Sens., vol. 28, no. 5, pp. 839–845, Sep. 1990. [20] T. Zhang, R. L. Armstrong, and J. Smith, “Investigation of the nearsurface soil freeze–thaw cycle in the contiguous United States: Algorithm development and validation,” J. Geophys. Res., vol. 108, no. D22, p. 8860, Nov. 2003.
4255
[21] L. Li, E. G. Njoku, E. Im, P. S. Chang, and K. S. Germain, “A preliminary survey of radio-frequency interference over the U.S. in Aqua AMSR-E data,” IEEE Trans. Geosci. Remote Sens., vol. 42, no. 2, pp. 380–390, Feb. 2004. [22] Y. H. Kerr, P. Waldteufel, J.-P. Wigneron, J.-M. Martinuzzi, J. Font, and M. Berger, “Soil moisture retrieval from space: The Soil Moisture and Ocean Salinity (SMOS) mission,” IEEE Trans. Geosci. Remote Sens., vol. 39, no. 8, pp. 1729–1735, Aug. 2001.
Lixin Zhang (M’08) received the B.A. degree from the University of Lanzhou, Lanzhou, China, in 1988, and the M.A. and Ph.D. degrees in geography from the Institute of Glaciology and Geocryology, Lanzhou, in 1991, and 2000, respectively. He then joined the Research Center for Remote Sensing and GIS, Beijing Normal University, Beijing, China, as a Professor. His research interests are physical properties of frozen soil and inversion algorithms or criteria for monitoring physical parameters of soil from remote sensing data.
Tianjie Zhao (S’09) received the B.S. degree from Beijing Normal University, Beijing, China, in 2007, where he is currently working toward the Ph.D. degree in cartography and geography information systems. He is also currently a Visiting Scientist in the Agricultural Research Service, Hydrology and Remote Sensing Laboratory, U.S. Department of Agriculture, Beltsville, MD. His research interests are microwave remote sensing of soil moisture and physical parameters of frozen soil.
Lingmei Jiang (AM’09) was born in Zhejiang, China, on October 31, 1978. She received the B.S. degree in agricultural meteorology from the Nanjing Institute of Meteorology, Nanjing, China, in 2000, and the Ph.D. degree in geography from Beijing Normal University, Beijing, China, in 2005. Her research interests include microwave remote sensing and modeling and retrieval of snow and soil properties.
Shaojie Zhao (S’08) received the B.S. degree in geography from Beijing Normal University, Beijing, China, in 2006, where he is currently working toward the Ph.D. degree, major in microwave remote sensing. His current interests include passive microwave remote sensing, modeling, and retrieval of frozen soil properties.
