High-Resolution Soil Moisture Estimation Using ASCAT

by

Print Friendly, PDF & Email

David Lindell and David G. Long, Electrical Engineering

I. INTRODUCTION

NASA’s Decadal Survey emphasizes the importance of soil moisture measurements, citing their significance in predicting natural hazards and the role of soil moisture levels in the water and carbon cycles [1]. Soil moisture estimates are used for a variety of applications including drought detection, flood and landslide forecasts, crop yield monitoring, and rain precipitation models [1]–[4]. Orbiting sensors provide an effective way to measure soil moisture globally because measurements of each area can be taken at a relatively high revisit frequency (typically one to two days). These sensors are generally able to measure soil moisture parameters despite cloud coverage and, in some cases, vegetation coverage of the target area.

Both 12.5- and 25-km resolution global soil moisture products are presently being produced using a change detection algorithm developed at the Vienna University of Technology (TU-Wien) [5]. The products are processed using data from the C-band Advanced Scatterometer (ASCAT) sensors aboard the Metop-A and Metop-B satellites [6]. With the aid of the Scatterometer Image Reconstruction (SIR) algorithm, higher resolution images (4.45- by 4.45 km/pixel) of the radar backscatter coefficient (σ0) can be produced [7]. Exploiting ASCAT’s high-resolution image capability enables higher resolution images of soil moisture than are currently available to be developed. Higher resolution soil moisture images could potentially help to improve studies in such areas as crop yield prediction, drought prediction and localization, and modeling of the water table and water cycle. This report describes how the TU-Wien model can be adapted to create high resolution soil moisture images with ASCAT.

II. METHODOLOGY

Data collected from both ASCAT and the European Remote-sensing Satellite (ERS) C-band scatterometers have been used to estimate relative soil moisture levels by applying a change detection algorithm to reported σ0 values [5], [8]. This change detection model (hereafter the TU-Wien model) exploits the dependence of σ0 on soil moisture levels: in general, σ0 increases as the amount of moisture in the soil increases [5]. The relative soil moisture in a particular area can be estimated by first normalizing all σ0 values to a nominal incidence angle. Then, the maximum and minimum backscatter measurements are found from a long time series of data collected from the area under study. An assumption is made that the highest backscatter value in the time series corresponds to soil which is saturated with water, and the lowest backscatter value corresponds to dry soil. For a given backscatter measurement, a relative topsoil moisture value (between 0% and 100%) is found by examining where the measurement falls between the measurements corresponding to dry and water-saturated soil. The volume content of moisture in the soil can also be extracted by multiplying the relative soil moisture value by the porosity of the soil [5]. In this way, a quantitative estimate of soil moisture can be produced using σ0 measurements.

The method for retrieving high-resolution soil estimates closely follows the TU-Wien model. High resolution (4.45- by 4.45 km/pix) C-band images of σ0 are developed by applying the SIR algorithm to ASCAT data. Using these images, a time series of σ0 measurements for each pixel are evaluated to obtain the soil moisture estimation parameters. A correction is applied to the parameters to compensate for an overestimation of soil wetness in arid regions, and the soil moisture is estimated on the 4.45- by 4.45 km/pixel grid.

III. RESULTS

The relative topsoil moisture estimates (ms) are produced at high resolution for land regions for every two days from 2009 to 2014. Each image uses 5 days of data to reduce coverage gaps, so consecutive images are produced using some of the same raw data. The high-resolution images are compared to the lower-resolution ASCAT soil moisture product produced with the Water Retrieval Package (WARP) version 5.5 [6] for validation purposes. Data from the WARP 5.5 ASCAT soil moisture product is used to produce images at the same temporal resolution, but using a courser 12.5-km grid. Low-resolution measurements are placed on the grid by averaging measurement values falling within each pixel.

fig 1
Fig. 1. Relative topsoil moisture and high resolution for day 251, 2011.

fig 2 Fig. 2. Average difference (left) and standard deviation (right) between relative topsoil moisture estimates at low and high resolution for the year 2011.

IV. DISCUSSION

An images series comparing the high-resolution and low-resolution retrieved soil moisture parameters is shown in Fig. 1. A closer view of areas near Lake Superior reveals the finer details present in the high-resolution image. Main differences in ms between the high- and low-resolution images occur primarily on the western side of the United States and are largely caused by differences in backscatter values chosen to indicate water saturation.

Fig. 2 demonstrates that the high-resolution ms estimates show general agreement with those reported by the low-resolution product, except in a few areas throughout the western United States. Overall, the high-resolution product appears to be slightly biased towards greater values of ms compared to the low-resolution product. The observed bias likely results from the high-resolution product’s comparatively lower values of the wet soil estimation parameters, but could also partially result from increased noise present in the high-resolution images.

V. CONCLUSION

Using high-resolution estimates of σ0 produced with ASCAT and the SIR algorithm, high-resolution soil moisture estimates can successfully be retrieved. Estimates of ms are produced using a modified version of the TU-Wien model. Most areas in the high-resolution estimates show good agreement with the lower-resolution product while revealing soil moisture features at a finer scale. I am preparing a paper to publish these results.

REFERENCES

    1. Earth Science and Applications from Space: National Imperatives for the Next Decade and Beyond. Washington, DC: The National Academies Press, 2007.
    2. W. Crow, G. Huffman, R. Bindlish, and T. Jackson, “Improving satellite-based rainfall accumulation estimates using spaceborne surface soil moisture retrievals,” ”J. Hydrometeorol.”, vol. 10, pp. 199–212, February 2009.
    3. C. Hauck, C. Barthlott, L. Krauss, and N. Kalthoff, “Soil moisture variability and its influence on convective precipitation over complex terrain,” Quarterly Journal of the Royal Meteorological Society, vol. 137, no. S1, pp. 42–56, 2011.
    4. A. K. Prasad, L. Chai, R. P. Singh, and M. Kafatos, “Crop yield estimation model for Iowa using remote sensing and surface parameters,” International Journal of Applied Earth Observation and Geoinformation, vol. 8, no. 1, pp. 26 – 33, 2006.
    5. W. Wagner, G. Lemoine, and H. Rott, “A method for estimating soil moisture from ERS scatterometer and soil data,” Remote Sensing of Environment, vol. 70, no. 2, pp. 191–207, 1999.
    6. V. Naeimi, K. Scipal, Z. Bartalis, S. Hasenauer, and W. Wagner, “An improved soil moisture retrieval algorithm for ERS and METOP scatterometer observations,” IEEE Trans. Geosci. Remote Sens., vol. 47, no. 7, pp. 1999–2013, July 2009.
    7. R. Lindsley and D. Long, “Adapting the SIR algorithm to ASCAT,” in Geoscience and Remote Sensing Symposium (IGARSS), 2010 IEEE International, July 2010, pp. 3402–3405.
    8. S. Hahn, T. Melzer, and W. Wagner, “Error assessment of the initial near real-time METOP ASCAT surface soil moisture product,” Geoscience and Remote Sensing, IEEE Transactions on, vol. 50, no. 7, pp. 2556–2565, July 2012.