L-Band Dielectric Properties of Different Soil Types Collected during ...
L-Band Dielectric Properties of Different Soil Types Collected during the MOUSE 2004 Field Experiment Mercè Vall-llossera, Miquel Cardona, Sebastián Blanch, Adriano Camps, Alessandra Monerris, Ignasi Corbella, Francesc Torres, Nuria Duffo Dept. Signal Theory and Communications, Universitat Politècnica de Catalunya (UPC), Campus Nord –D4, Jordi Girona 1-3, 08034 Barcelona, Spain Phone: +34 93 401 7261 FAX: +34 93 401 7232 E-mail: [email protected]
Abstract—The brightness temperature measured from land is related to the soil moisture through the dielectric constant, which depends, among other parameters, on soil composition and porosity. The aim of MOUSE campaign was to establish the brightness temperature dependence on the soil type and moisture content [1]. For this study and for the retrieval of the soil moisture content from radiometric measurements, it is necessary to use accurate models for the soil complex dielectric constant. This paper presents a technique to measure the complex dielectric constant using a microwave stripline setup, and its application to the soil samples measured during MOUSE 2004 campaign. The measured dielectric constants of each soil type versus volumetric soil moisture (0% to 40%) are compared with the values obtained using models already appeared in the literature. Keywords-Soil moisture; dielectric constant.
I.
INTRODUCTION
The objective of the Soil Moisture and Ocean Salinity (SMOS) Earth Explorer Opportunity Mission of the European Space Agency (ESA) over land is to provide a global and precise soil moisture estimation from multi-angular L-band radiometric measurements, which will be obtained with the MIRAS (Microwave Imaging Radiometer by Aperture Synthesis) instrument. It is known that the brightness temperature measured from land is related to the soil moisture through the complex dielectric constant (or equivalently, the dielectric permittivity and the electrical conductivity). For dry soils the dielectric permittivity (εr) varies between 2 and 4, depending on the soil type, and it is nearly constant vs. frequency and temperature. The imaginary part of the dielectric constant is usually smaller than 0.5. On the other hand, free water has a much high dielectric permittivity, close to 80, and then the dielectric constant of wet soils is highly dependent of their water content. Nevertheless, other factors, such as the temperature, frequency, bulk soil density and particle shapes influence the soil dielectric constant. Several dielectric constant models are reported in the literature [2, 3], but important differences on the predicted dielectric constants are appreciated (e.g. Wang: red lines, and Dobson models: green lines, in Fig. 5). Consequently, as the precision of the retrieved soil moisture from radiometric measurements is directly related to the precision of the dielectric constant model, for the MOUSE experiment, samples of the measured soil types were colleted and their dielectric properties were determined using the method developed by
0-7803-9050-4/05/$20.00 ©2005 IEEE.
Blanch and Aguasca (described in section III [5]). In section III the measured experimental data are presented and the behavior is compared with those obtained applying the models developed by Wang [2] and Dobson [3]. MOUSE experiment is briefly presented in section II and a complete description can be found in [1]. II.
MOUSE 2004 FIELD CAMPAIGN DESCRIPTION
The Monitoring Underground Soil Experiment (MOUSE) 2004 took place at the Joint Research Centre Outdoor Test Facility (JRC, 45º 48’N, 8º37’E), in Ispra (North Italy) from June 7th to July 1st, 2004. This field experiment was devoted to the study of the brightness temperature dependence with the soil type, moisture content and the incidence angle. The test lane is about 80 m long and 5.7 m wide and it is formed by a strip of seven plots of different soil types (grassy terrain, loamy terrain, sandy soil, pure sand, clay soil, organic soil and ferromagnetic soil) aligned and separated by concrete walls. All the soils were bare except for the first one, which was not measured in MOUSE. Three complete measurement cycles, from saturation to complete dry soil, were performed for the 6 types of soils under study. More details of this field campaign and the results from the data processing can be found in [1]. To overcome the uncertainties of the dielectric constant values for the different types of soil and moisture conditions measured in MOUSE, soil samples were acquired for their analysis in the laboratory. In table 1 the properties of the measured soils are presented. TABLE I. MEASURED SOIL TYPES [4] Field
Soil Type
1 2 3 4 5
Loamy Sandy Clay Organic Ferromagnetic
III.
Soil Texture Sand Clay (%) (%) 78 2 95 0 15 40 78 2 82.5 2.5
Organic matter (%) 2 0 0.3 5.7 0.1
Conductivity (S/cm) 66 0 0.3 5.7 0.1
MEASUREMENT SETUP AND PROCEDURE
The measurement setup used for determining the dielectric constant of soil samples is the one designed by Blanch and Aguasca [5]. The system is based on a stripline structure: a
1109
1109
• VNA calibration.
cylindrical conductor (270 mm of length and 3 mm of diameter) inside to two parallel plates (100 mm x 270 mm) separated 10 mm and the soil sample is introduced between them (see Fig. 1). The main advantages of this system are: its large bandwidth, the possibility of introducing a wide (?) soil sample and that it is an open structure. The strip-line is connected with two coaxial cables to the Vector Network Analyzer (VNA) (Fig. 2) to obtain the Sparameter values of the structure for each sample and from them its complex dielectric constant. This technique was already used for measuring sea water permittivity and detailed information can be found in [5].
• Measurement of sample’s ‘S parameters’ computation the dielectric constant.
and
• Introduction 27 ml of water (10% humidity) uniformly distributed in the cavity. • Weight of the new sample. • Measurement of sample’s ‘S parameters’ and computation of the dielectric constant for this new humidity of the same soil type. • Repetition of the last three steps by introducing controlled amounts of water to obtain increasing levels of moisture contents and for each one measurement of their dielectric constants. Fig. 3 shows the dielectric constant of the clay (a) real part and b) imaginary part) respect to the frequency for different volumetric moistures. Real and imaginary parts of the dielectric constant increase with soil moisture. Almost no dependence with frequency is appreciated for both the real and the imaginary parts, in this frequency range (1.2-2.5GHz). Same behavior was observed for the other soils.
Figure 1. Strip-line structure. Top outer conductor removed.
Table 1 shows the characteristics of 5 types of soils. They correspond to samples of the bare soils at JRC-Ispra test lane, except for the reference field (pure sand), because it has the same composition as the sandy soil.
TABLE II.
WATER QUANTITY VS VOLUMETRIC HUMIDITY
Soil volumetric humidity (%) 10 20 30 40
Water volume (ml) 27 54 81 107
Figure 2. Measurement Set-up
The objective of this study was to obtain the dielectric constant of these soils under different moisture conditions: 0%, 10%, 20%, 30% and 40%. In order to reduce the error on the obtained values, the same measurements were repeated several times. Considering the sizes presented in Fig. 2 the sample volume was 266.1 ml and the water volume for each studied humidity is presented on Table 2. For each soil type the measurement procedure was the following: • To dry completely the soil inside an oven at 100ºC to obtain a 0% reference humidity. • Introduce the soil sample inside the stripline structure and weight it.
0-7803-9050-4/05/$20.00 ©2005 IEEE.
1110
(a)
(b) Figure 3. Clay dielectric constant for different water contents (red 0%, green 10%, yellow 20%, blue 30% and black 40%) respect to frequency (from 1.2 2.5 Ghz): (a) Real part (b) Imaginary part.
1110
Sand 1(b)
1(a)
Loam
2(b)
2(a)
Organic
3(b)
3(a)
Ferromagnetic
4(b)
4(a)
Clay
5(b)
5(a)
Figure 4.
Comparison of the dielectric constant (a) real and b) imaginary parts): blue line: polinomial fitting from experimental data. Green line: Dobson model and red line: Wang model for 1) sandy, 2) loamy, 3) organic 4) ferromagnetic and 5) clay soils.
0-7803-9050-4/05/$20.00 ©2005 IEEE.
1111
1111
Fig. 4 compares the behavior of the regression line derived from the experimental data (in blue) for the real part (left column) and the imaginary part (right column) of the dielectric constant for the five types of measured soils with the values obtained using Wang (red line) and Dobson (green line) models. The agreement of the real part obtained from the experimental data is better with Wang model, except for the Clay soil (the three models are very alike). For loamy soil the agreement of the experimental data values and the ones of Wang model is excellent and for organic and ferromagnetic soils discrepancies increase with water content. Finally, for sandy soil the drier the soil is the regression line is more similar to Wang model, but when the wetness increase this line tends to be closer to the Dobson model and to separate from the Wang model. For the imaginary part, the values obtained with Blanch and Aguasca method tend to increase faster than the other two methods.
a)
Plots in Fig. 4 show the regression lines derived for each type of soil and Fig. 5 presents the comparison of the measured dielectric constant (a) real part and b) imaginary part) with the soil type and moisture. Differences increase a little bit with the water content. Finally in order to study the impact of the dielectric constant errors with the measured brightness temperature, which will translate into an error in the retrieved soil moisture, Fig. 6 plots the predicted brightness temperature at nadir for each soil and moisture for a temperature of 15ºC, using the dielectric constants of Fig. 5. Differences are as high as 28 K for dry soil. This differences reduce with humidity. IV.
CONCLUSIONS
In this paper we have applied the technique proposed by Blanch and Aguasca to measure the complex dielectric constant [5] to characterize soil samples. No variation of the permittivity and imaginary part of the dielectric constant with frequency (range 1.2-2.5 GHz) is appreciated for any kind of soil and moisture content. As expected, both real and imaginary parts of the dielectric constant increase with water content. Finally, for the real part of the dielectric constant measured using the Blanch and Aguasca technique is in better agreement with the results obtained with the Wang model than with the Dobson one. For the imaginary part, the behavior derived from our experimental data, always predicts higher values than the other two models. The errors in the estimation of the dielectric constant can lead to brightness temperature errors at nadir as high as 28 K for a soil surface temperature of 15ºC, which is an important uncertainty term in the retrieval of soil moisture values. ACKNOWLEDGEMENTS
b)
This work has been financed by the Spanish MCYT and FEDER funds (projects TIC 2002-04451-C02-01 and ESP 2004-00671). The authors want to thank the JRC personnel for their support during the MOUSE 2004 experiment.
Figure 5. Comparison of the dielectric constant (a) real part and b)imaginary parts) dependence respect to SM of the 5 different type of soils measured.
REFERENCES
Figure 6.
Dependence of the brightness temperature respect to soil type and moisture.
0-7803-9050-4/05/$20.00 ©2005 IEEE.
[1] A. Monerris, M. Cardona, M. Vall-llossera, A. Camps, R. Sabia, R. Villarino, E. Álvarez, and S. Sosa, “Soil Moisture Retrieval Errors, Using L-Band Radiometry Induced by the Soil Type Variability”, in Proc. IGARSS 2005 Symposium, Seul, Korea, July 2005 [2] J. R. Wang and T.J. Schmugge, “An empirical model for the complex dielectric permittivity of soils as a function of water content”, IEEE Trans. Geoscience and Remote Sensing, vol. 18, pp. 288-295, 1980. [3] M. C. Dobson, F. T. Ulaby, M. T. Hallikainen, and M. A. El-Rayes, “Microwave dielectric behaviour of wet soils, Part II: Dielectric mixing models”, IEEE Trans. Geoscience and Remote Sensing, vol. GRS-23, pp. 35-46, Jan. 1985. [4]http://www.itep.ws/facilities/EC/test_centre/Facilities/Facility2/facility2.ht m [5] S. Blanch and A. Aguasca, “Seawater dielectric permittivity model from measurements at L-band”, in Proc. IGARSS 2004 Symposium, Anchorage, USA, Sept. 2004.
1112
1112
Abstract—The brightness temperature measured from land is related to the soil moisture through the dielectric constant, which depends, among other parameters, on soil composition and porosity. The aim of MOUSE campaign was to establish the brightness temperature dependence on the soil type and moisture content [1]. For this study and for the retrieval of the soil moisture content from radiometric measurements, it is necessary to use accurate models for the soil complex dielectric constant. This paper presents a technique to measure the complex dielectric constant using a microwave stripline setup, and its application to the soil samples measured during MOUSE 2004 campaign. The measured dielectric constants of each soil type versus volumetric soil moisture (0% to 40%) are compared with the values obtained using models already appeared in the literature. Keywords-Soil moisture; dielectric constant.
I.
INTRODUCTION
The objective of the Soil Moisture and Ocean Salinity (SMOS) Earth Explorer Opportunity Mission of the European Space Agency (ESA) over land is to provide a global and precise soil moisture estimation from multi-angular L-band radiometric measurements, which will be obtained with the MIRAS (Microwave Imaging Radiometer by Aperture Synthesis) instrument. It is known that the brightness temperature measured from land is related to the soil moisture through the complex dielectric constant (or equivalently, the dielectric permittivity and the electrical conductivity). For dry soils the dielectric permittivity (εr) varies between 2 and 4, depending on the soil type, and it is nearly constant vs. frequency and temperature. The imaginary part of the dielectric constant is usually smaller than 0.5. On the other hand, free water has a much high dielectric permittivity, close to 80, and then the dielectric constant of wet soils is highly dependent of their water content. Nevertheless, other factors, such as the temperature, frequency, bulk soil density and particle shapes influence the soil dielectric constant. Several dielectric constant models are reported in the literature [2, 3], but important differences on the predicted dielectric constants are appreciated (e.g. Wang: red lines, and Dobson models: green lines, in Fig. 5). Consequently, as the precision of the retrieved soil moisture from radiometric measurements is directly related to the precision of the dielectric constant model, for the MOUSE experiment, samples of the measured soil types were colleted and their dielectric properties were determined using the method developed by
0-7803-9050-4/05/$20.00 ©2005 IEEE.
Blanch and Aguasca (described in section III [5]). In section III the measured experimental data are presented and the behavior is compared with those obtained applying the models developed by Wang [2] and Dobson [3]. MOUSE experiment is briefly presented in section II and a complete description can be found in [1]. II.
MOUSE 2004 FIELD CAMPAIGN DESCRIPTION
The Monitoring Underground Soil Experiment (MOUSE) 2004 took place at the Joint Research Centre Outdoor Test Facility (JRC, 45º 48’N, 8º37’E), in Ispra (North Italy) from June 7th to July 1st, 2004. This field experiment was devoted to the study of the brightness temperature dependence with the soil type, moisture content and the incidence angle. The test lane is about 80 m long and 5.7 m wide and it is formed by a strip of seven plots of different soil types (grassy terrain, loamy terrain, sandy soil, pure sand, clay soil, organic soil and ferromagnetic soil) aligned and separated by concrete walls. All the soils were bare except for the first one, which was not measured in MOUSE. Three complete measurement cycles, from saturation to complete dry soil, were performed for the 6 types of soils under study. More details of this field campaign and the results from the data processing can be found in [1]. To overcome the uncertainties of the dielectric constant values for the different types of soil and moisture conditions measured in MOUSE, soil samples were acquired for their analysis in the laboratory. In table 1 the properties of the measured soils are presented. TABLE I. MEASURED SOIL TYPES [4] Field
Soil Type
1 2 3 4 5
Loamy Sandy Clay Organic Ferromagnetic
III.
Soil Texture Sand Clay (%) (%) 78 2 95 0 15 40 78 2 82.5 2.5
Organic matter (%) 2 0 0.3 5.7 0.1
Conductivity (S/cm) 66 0 0.3 5.7 0.1
MEASUREMENT SETUP AND PROCEDURE
The measurement setup used for determining the dielectric constant of soil samples is the one designed by Blanch and Aguasca [5]. The system is based on a stripline structure: a
1109
1109
• VNA calibration.
cylindrical conductor (270 mm of length and 3 mm of diameter) inside to two parallel plates (100 mm x 270 mm) separated 10 mm and the soil sample is introduced between them (see Fig. 1). The main advantages of this system are: its large bandwidth, the possibility of introducing a wide (?) soil sample and that it is an open structure. The strip-line is connected with two coaxial cables to the Vector Network Analyzer (VNA) (Fig. 2) to obtain the Sparameter values of the structure for each sample and from them its complex dielectric constant. This technique was already used for measuring sea water permittivity and detailed information can be found in [5].
• Measurement of sample’s ‘S parameters’ computation the dielectric constant.
and
• Introduction 27 ml of water (10% humidity) uniformly distributed in the cavity. • Weight of the new sample. • Measurement of sample’s ‘S parameters’ and computation of the dielectric constant for this new humidity of the same soil type. • Repetition of the last three steps by introducing controlled amounts of water to obtain increasing levels of moisture contents and for each one measurement of their dielectric constants. Fig. 3 shows the dielectric constant of the clay (a) real part and b) imaginary part) respect to the frequency for different volumetric moistures. Real and imaginary parts of the dielectric constant increase with soil moisture. Almost no dependence with frequency is appreciated for both the real and the imaginary parts, in this frequency range (1.2-2.5GHz). Same behavior was observed for the other soils.
Figure 1. Strip-line structure. Top outer conductor removed.
Table 1 shows the characteristics of 5 types of soils. They correspond to samples of the bare soils at JRC-Ispra test lane, except for the reference field (pure sand), because it has the same composition as the sandy soil.
TABLE II.
WATER QUANTITY VS VOLUMETRIC HUMIDITY
Soil volumetric humidity (%) 10 20 30 40
Water volume (ml) 27 54 81 107
Figure 2. Measurement Set-up
The objective of this study was to obtain the dielectric constant of these soils under different moisture conditions: 0%, 10%, 20%, 30% and 40%. In order to reduce the error on the obtained values, the same measurements were repeated several times. Considering the sizes presented in Fig. 2 the sample volume was 266.1 ml and the water volume for each studied humidity is presented on Table 2. For each soil type the measurement procedure was the following: • To dry completely the soil inside an oven at 100ºC to obtain a 0% reference humidity. • Introduce the soil sample inside the stripline structure and weight it.
0-7803-9050-4/05/$20.00 ©2005 IEEE.
1110
(a)
(b) Figure 3. Clay dielectric constant for different water contents (red 0%, green 10%, yellow 20%, blue 30% and black 40%) respect to frequency (from 1.2 2.5 Ghz): (a) Real part (b) Imaginary part.
1110
Sand 1(b)
1(a)
Loam
2(b)
2(a)
Organic
3(b)
3(a)
Ferromagnetic
4(b)
4(a)
Clay
5(b)
5(a)
Figure 4.
Comparison of the dielectric constant (a) real and b) imaginary parts): blue line: polinomial fitting from experimental data. Green line: Dobson model and red line: Wang model for 1) sandy, 2) loamy, 3) organic 4) ferromagnetic and 5) clay soils.
0-7803-9050-4/05/$20.00 ©2005 IEEE.
1111
1111
Fig. 4 compares the behavior of the regression line derived from the experimental data (in blue) for the real part (left column) and the imaginary part (right column) of the dielectric constant for the five types of measured soils with the values obtained using Wang (red line) and Dobson (green line) models. The agreement of the real part obtained from the experimental data is better with Wang model, except for the Clay soil (the three models are very alike). For loamy soil the agreement of the experimental data values and the ones of Wang model is excellent and for organic and ferromagnetic soils discrepancies increase with water content. Finally, for sandy soil the drier the soil is the regression line is more similar to Wang model, but when the wetness increase this line tends to be closer to the Dobson model and to separate from the Wang model. For the imaginary part, the values obtained with Blanch and Aguasca method tend to increase faster than the other two methods.
a)
Plots in Fig. 4 show the regression lines derived for each type of soil and Fig. 5 presents the comparison of the measured dielectric constant (a) real part and b) imaginary part) with the soil type and moisture. Differences increase a little bit with the water content. Finally in order to study the impact of the dielectric constant errors with the measured brightness temperature, which will translate into an error in the retrieved soil moisture, Fig. 6 plots the predicted brightness temperature at nadir for each soil and moisture for a temperature of 15ºC, using the dielectric constants of Fig. 5. Differences are as high as 28 K for dry soil. This differences reduce with humidity. IV.
CONCLUSIONS
In this paper we have applied the technique proposed by Blanch and Aguasca to measure the complex dielectric constant [5] to characterize soil samples. No variation of the permittivity and imaginary part of the dielectric constant with frequency (range 1.2-2.5 GHz) is appreciated for any kind of soil and moisture content. As expected, both real and imaginary parts of the dielectric constant increase with water content. Finally, for the real part of the dielectric constant measured using the Blanch and Aguasca technique is in better agreement with the results obtained with the Wang model than with the Dobson one. For the imaginary part, the behavior derived from our experimental data, always predicts higher values than the other two models. The errors in the estimation of the dielectric constant can lead to brightness temperature errors at nadir as high as 28 K for a soil surface temperature of 15ºC, which is an important uncertainty term in the retrieval of soil moisture values. ACKNOWLEDGEMENTS
b)
This work has been financed by the Spanish MCYT and FEDER funds (projects TIC 2002-04451-C02-01 and ESP 2004-00671). The authors want to thank the JRC personnel for their support during the MOUSE 2004 experiment.
Figure 5. Comparison of the dielectric constant (a) real part and b)imaginary parts) dependence respect to SM of the 5 different type of soils measured.
REFERENCES
Figure 6.
Dependence of the brightness temperature respect to soil type and moisture.
0-7803-9050-4/05/$20.00 ©2005 IEEE.
[1] A. Monerris, M. Cardona, M. Vall-llossera, A. Camps, R. Sabia, R. Villarino, E. Álvarez, and S. Sosa, “Soil Moisture Retrieval Errors, Using L-Band Radiometry Induced by the Soil Type Variability”, in Proc. IGARSS 2005 Symposium, Seul, Korea, July 2005 [2] J. R. Wang and T.J. Schmugge, “An empirical model for the complex dielectric permittivity of soils as a function of water content”, IEEE Trans. Geoscience and Remote Sensing, vol. 18, pp. 288-295, 1980. [3] M. C. Dobson, F. T. Ulaby, M. T. Hallikainen, and M. A. El-Rayes, “Microwave dielectric behaviour of wet soils, Part II: Dielectric mixing models”, IEEE Trans. Geoscience and Remote Sensing, vol. GRS-23, pp. 35-46, Jan. 1985. [4]http://www.itep.ws/facilities/EC/test_centre/Facilities/Facility2/facility2.ht m [5] S. Blanch and A. Aguasca, “Seawater dielectric permittivity model from measurements at L-band”, in Proc. IGARSS 2004 Symposium, Anchorage, USA, Sept. 2004.
1112
1112
