Polrimetric Radarsat2 Imagery for Monitoring Spatial and Temporal Water Content Variability in Alpine Area
Notarnicola, Claudia¹; Pasolli, Luca¹; Bertoldi, Giacomo²; Lorenzo, Bruzzone³; Niedrist, Georg²; Tappeiner, Ulrike²; Zebisch, Marc^1
1EURAC-Institute for Applied Remote Sensing, ITALY; ²EURAC-Institute for Alpine Environment, ITALY; ³Dep. of Information Engineering and Computer Science, University of Trento, ITALY

The spatial and temporal monitoring of the topsoil moisture content has important implications in various application domains, such as water management as well as risk prevention (flood and landslide). To this aim, microwave space-borne remote sensing imagery could provide useful information. With this regard, in the last years the scientific community has shown a rising interest in the use of the new generation Synthetic Aperture Radars (SARs) systems, such as RADARSAT2, thanks to the good geometric resolution at the ground and polarimetric capability. In this framework, this paper aims at exploring the capability of fully polarimetric RADARSAT 2 SAR imagery to map topsoil moisture content in a mountain environment.
The retrieval of soil moisture is a challenging task, due to the complexity and non-linearity of the mapping between the input features (extracted from the microwave signal) and the desired target parameter (the soil moisture content). In addition to this, microwave signals are sensitive to other target properties, such as the surface roughness and the presence of vegetation coverage [1]. To overcome these effects, one may resort on multi-frequency or multi-angle acquisitions. However, such acquisition modalities are not commonly adopted by the current orbiting satellite SAR systems, thus making multi-frequency and multi-angle approaches not applicable for large scale or temporal monitoring. Another possibility is the polarimetric acquisition capability of the new generation SAR systems [2]. However, fully polarimetric estimation approaches have not been fully exploited yet, due to the only recent availability of this kind of data.
In mountainous environments, the estimation of soil parameters becomes even more complex, due to the effect of topography. SAR systems are strongly affected by the topography due to the side-looking acquisition geometry. Even though distortions (such as layover or shadowing) do not occur, the SAR signal is affected by the variability of the local incidence angle and of the distance of the target from the sensor. Moreover, mountain areas present typically high heterogeneity and variability of the land coverage which may significantly affect the soil moisture retrieval process, in particular when medium-high geometrical resolution imagery is considered.
The SOFIA project (ESA AO-6280) aimed ad investigating the effectiveness of new generation SAR imagery in combination with advanced retrieval methods for the estimation of topsoil moisture content. In this framework, 30 fully polarimetric RADARSAT2 SAR images have been acquired over the Alto Adige are (Northern Italy) during the summer 2010 and 2011. An advanced retrieval algorithm, based on the Support Vector regression method and proper feature extraction strategies [3], has been exploited for the estimation of topsoil water content from the polarimetric SAR imagery in combination with ancillary data. The capability of the estimation system to provide spatially and temporally distributed estimates of soil moisture over pastures and meadows in the Alto Adige area is addressed in this contribution. To this aim, comparisons with ground measurements of soil moisture as well as with data (both local and distributed) from meteorological stations are carried out indicating a RMSE value on soil moisture values of around 4.5%. The achieved results indicate that polarimetric RADARSAT 2 SAR imagery in combination with the proposed estimation system is effective for mapping the soil moisture status both in time and space in Alpine areas. More in details the high performances are reached when the combination of HH and HV polarization are used because the availability of HV pol allows compensating the effect of vegetation. The main found limitation is the reduced swath of fully polarimetric RADARSAT 2 SAR imagery, which implies the acquisition of numerous scenes to cover a large area (such as the Alto Adige region). This issue prevents also the possibility acquire multiple scenes over the same area with different geometries, in order to minimize the effects of geometric distortions, (very often in mountain areas where the sideslopes have different orientations) e.g. ascending and descending acquisitions. Another limitation is the influence of the vegetation coverage on the C-band SAR signal, with prevents the possibility to retrieve topsoil moisture content in presence of densely vegetated areas such as orchards which are quite widespread in the region. Such drawback could be overcome by using different acquisition modes of the RADARSAT 2 system (e.g., the dual polarization standard mode), which ensures a larger swath and by integrating available L band SAR data in the retrieval process.

[1] Ulaby, F.T., Bradley, G.A., and Dobson, M.C., (1979), "Microwave backscatter dependence on surface roughness, soil moisture and soil texture, Part-II: Vegetation covered soil," IEEE Trans. Geosci. Electron., vol. GE-17, no. 2, pp. 33-40.
[2] Kim, Y., and van Zyl, J.J., (2009), "A time series approach to estimate soil moisture using polarimetric radar data," IEEE Transaction on Geoscience and Remote Sensing, vol. 47, no. 8.
[3] L. Pasolli, C.Notarnicola, L. Bruzzone, G. Bertoldi, S. Della Chiesa, G. Niedrist, U. Tappeiner, M. Zebisch, Polarimetric Radarsat2 imagery for soil moisture retrieval in areas, Canadian Journal of Remote Sensing, 2011, 37:535-547, 10.5589/m11-065.