1. INTRODUCTION

The theory behind microwave remote sensing of soil moisture is based on the large contrast between the dielectric properties of liquid water (~ 80) and dry soil (<4). Generally, low frequencies (1–6 GHz) are used for soil moisture retrievals. Currently, the soil moisture retrieval algorithm developed by Jackson (1993) is used for the Advanced Earth Observing Satellite (ADEOS) II Advanced Microwave Scanning Radiometer (AMSR). For operational implementation, each AMSR pixel is used as a cross reference to the land cover and NDVI ancillary data. Surface roughness effects are fixed at 0.1 (Koike et al. 2000).

In this study, we propose a unique inversion algorithm of soil moisture on the basis of the difference in the polarizations of microwave radiation over land surfaces.

2. METHODS

We derive here the formula for estimating the volumetric soil water content. First, the rough surface reflectivity is estimated from the ratio between the observed brightness temperature  and the surface temperature. In this research, the AMSR-E data correspond to a single day (Apr. 1, 2009) at each frequency and at an incidence angle of 55.0¢®¨°. Brewster¢®¯s angles for the soil-water mixed dielectric models are 54.14¢®¨° for 6.9 GHz. The surface temperature can be estimated independently using the National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) or the empirical linear relationship between the V-polarized T_B at 37 GHz and 1.25 cm soil temperature (De Jeu, 2003). Next, we estimate the vegetation optical depth (De Jeu, 2003).

An approximate relationship between and is used (hereafter, referring to as the Hong approximation). In addition, the roughness effects are different near Brewster¢®¯s angle (i.e.,  and). Thus, the quasi-specular reflectivity for H polarization,, is estimated using.

The roughness parametercan be estimated as

              (1)

Finally, we estimate the volumetric soil water content  using a relationship between the absolute value of the dielectric constant and the volumetric soil water content for multiple soils (Topp et al., 1980).

3. RESULTS

Figure 1 shows the comparison results using the refractive index of water (Sadiku, 1985) for comparing the Hong approximation and the Fresnel equations in the AMSR-E frequencies. In this case, the view angle ranges from 0¢®¨° to 55¢®¨°. The difference between the two equations for V and H polarizations is less than 0.001.

Figure 1. Validation of the Hong approximation:  (Fresnel)-(Simulation). The term ¢®°Æsimulation¢®± implies the reflectivities simulated using the Hong approximation. The curves cover the view angles 0¢®¨°, 10¢®¨°, 20¢®¨°, 30¢®¨°, 40¢®¨°, 50¢®¨°, and 55¢®¨°.

The rough and specular reflectivities are calculated for the soil-water mixed surface to simulate the realistic land surface. The roughness and the volumetric soil moisture are assumed to be 0.3 cm and 0.1 g/cm©©ª, respectively, at the AMSR-E 6.9 and 10 GHz channels, respectively. The dielectric constant of soil,, is obtained using the complex dielectric constant model of soil (Dobson et al., 1985).  Figure 2 shows the results of roughness effects on the polarized reflectivities. The assumptions andare satisfied near Brewster¢®¯s angles.

Figure 2. Difference between the rough and specular reflectivities for V and H polarizations. The AMSR-E 6.9 and 10 GHz channels are tested for the soil-water mixed model.  and are assumed to be 0.3 cm and 0.1 g/cm©©ª, respectively.

Figure 3 shows the estimated roughness parameters over land for the AMSR-E 6.9 GHz channel. The values of is approximately within 0.3 cm over land at 6.9 GHz.

Figure 4 shows the estimated surface soil moisture over land for the AMSR-E 6.9 GHz channel. The desert, snow, ice, and highlands have low amounts (¢®Â0.1 g/cm©©ª) of soil moisture. The European, Central African, Amazonian, and Central Asian regions exhibit the characteristic of vegetation regions. Compared with the current AMSR-E daily product (level 3), the proposed algorithm shows good qualitative agreement with the current soil moisture algorithms when we consider the uncertainty in the soil moisture estimation over vegetation and snow-covered regions.

Figure 3. Estimated roughness at the AMSR-E 6.9 GHz channel.

Figure 4. Volumetric soil water content measured at the AMSR-E 6.9 GHz channel.

4. CONCLUSION

The result indicate relatively larger soil moisture levels over coastal and vegetation regions as compared with snow, ice, and desert regions. Comparing these results with the current AMSR-E products, we find that the general features of the retrieved soil moistures from the two algorithms are qualitatively similar. The unique soil moisture algorithm proposed in this research is closer to physical reality in that it does not require fixing the roughness for operational implementation. In addition, another advantage of our algorithm is that it does not require a priori knowledge of the dielectric constants of the surface.

REFERENCES

De Jeu, R. A. M. 2003. Retrieval of Land Surface Parameters Using Passive Microwave Observations. Ph.D. Dissertation, VU Amsterdam, 32-33, 2003.

Dobson, M. C., F. T. Ulaby, M. T. Hallikainen, and M. A. El-Rayes. 1985. Microwave dielectric behavior of wet soil, II, Dielectric mixing models. IEEE Trans. Geosci. Remote Sens. 23: 35-46.

Jackson, T. J. 1993. Measuring surface soil moisture using passive microwave remote sensing. Hydrol. Processes. 7:139-152.

Koike, T., E. G. Njoku, T. J. Jackson, and S. Paloscia. 2000. Soil moisture algorithm development and validation for the ADEOS-II/AMSR. IEEE special issue IGARSS:1253-1255.

Topp, G. C., J. L. Davis, and A. P. Annan. 1980. Electromagnetic determination of soil water content: measurements in coaxial transmission lines. Water Resour. Res. 16:574-582.

Sadiku, M. N. O. 1985. Refractive index of snow at microwave frequencies. Appl. Opt. 24:.571-575.