Courtesy of NASA/JPL

The activity coordinated by the IROE-CNR sees participation of research groups from the following institutions: ITIS-CNR, Matera; IESI-CNR Bar; Department of Physics, University of Florence; Massachusetts Institute of Technology (USA); Centre d'Etudes Spatiales de la Biosphère (France).

The research is aimed at assessing the effects of ground surface characteristics on the SAR signal and on the height profile obtained from the interferometric measurements. Interaction of the microwave signal with the ground is known to involve both surface and volumetric scattering, and both amplitude and phase of the back-diffused signal depend on the geometrical and dielectric characteristics of the surfaces investigated. Much work has been done on correlating signal amplitude with observed surface type, whereas there is only slight information on the influence of surface characteristics on phase and coherence of the SAR interferometric signal. On the other hand, on the basis of knowledge gained, we expect geometrical (e.g. surface roughness) and dielectric (induced by ground humidity) variations in surfaces observed by SAR to influence the interferometric signal. The research objective is accordingly to assess the contribution of interferometric coherence and phase measurements to classification of ground type, estimation of biophysical parameters of surfaces observed and assessment of penetration of vegetation-covered areas, to reduce error in height measurement.

The test areas

The test areas will be Matera (southern Italy), Montespertoli (central Italy) and Araude (southern France). In each of these areas experiments have already been done with various SAR systems (AIRSAR, SAR 580, ERS-1/2, JERS-1), so that a broad series of data is available, from both remote sensing and "ground truth".

The Matera site, the same one as used for the calibration experiment, is an agricultural area sown mainly to wheat, in winter with many bare fields in various conditions of surface roughness. The area of Montespertoli, near Florence, consists of the basin of the river Pesa with its small tributary Virginio. It is typical of the climatic, lithological and vegetation conditions of many agricultural areas in central Italy. Various types of cultivation are present, including vineyards and olive groves, and also natural woodlands. Finally, the French site Aurade, located some 30 km west of Toulouse, is again an agricultural region, characterized by undulating terrain.

In all these areas the following measurements will be taken in association with the mission:

• classification of cultivation and woodland types

• ground humidity and surface roughness

• wood biomass in the forests

• biometric parameters of cultivation types present.

Fig. 1 - SAR image of Montespertoli area made by combining P, L and C bands in HV polarization.

Electromagnetic modelling

To enable a physical interpretation of the SAR interferometric data, a coherent discrete-element model has been developed that predicts the backscattering, coherence and interferometric phase centre position for observations of vegetation-covered areas. The model coherently sums the fields diffused by the individual scatterers (disks to simulate leaves, cylinders for stalks and trunks) and calculates the attenuation using Foldy's approximation. All the individual contributions to scattering coming from the individual scatterers and reflected from the ground are considered (Fig. 2). The model has already been tested to simulate the interferometric response of wheat and sunflowers, taken as examples of two crops with very different characteristics from each other.

Fig. 2 - Contributions to total backscattering considered in the electromagnetic model.

Data analysis and results expected

The data obtained, both X-SAR and SIR-C, will primarily be used to estimate heights, which are then compared with GPS measurements and DEMs available for the test areas. Secondly, the interferometric data (amplitude, coherence and phase) obtained by X-SAR (and if available by SIR-C) will be used to identify the response type of various kinds of surface to the interferometric radar signal using statistical relationships, and to compare the simulations with the experimental data. In particular, penetration into forest and other vegetation types will be estimated using the interferometric measurement and the electromagnetic model.

Fig. 3 - Example of classification in Montespertoli area obtained from AIRSAR polarimetric multifrequency data (top) compared with the facts on the ground (bottom).