HydroSoil Campaign

Exploitation of the HydroSoil campaign to study the capabilities
of a geostationary SAR for continuous Earth observation

The aim of this task is to carry out the analysis and exploitation of the data acquired in the framework of the HydroSoil campaign.

The observation principle of a GEOSAR relies on the sensitivity of the measured radar echoes to the properties in the scene, and in particular the diurnal change in water content at soil (soil moisture) and vegetation (vegetation water content).
The basic observables correspond to the backscattering coefficient at the polarimetric channels in use. Nevertheless, there is evidence in previous works with LEO data of the sensitivity of other features, like the complex correlations between channels.
Additionally, all the possible polarimetric configurations (single-, compact- and quad-pol) need to be tested in this specific scenario (the HydroSoil test site), with shallow incidence angles and continuous observations, since the polarimetric mode has some impact on the satellite design and on its operation.
Therefore, an analysis of backscattering and polarimetry sensitivity to soil moisture and other vegetation properties is being performed using the data obtained by the installed GBSAR in the HydroSoil test site (in the roof of the EEABB building).

Not only the sensitivity to soil moisture and other vegetation characteristics of backscattering and polarimetry needs to be analysed, but also of interferometry. The continuous observation framework provided by a GEOSAR allows interferometrically combining pairs of images acquired at different times.
The coherence of the corresponding interferogram acts as a change detector and is strongly influenced by the scene properties and their evolution. The interferometric phase is also related to the vegetation dielectric structure and the soil conditions.  This sensitivity is strongly influenced by the images temporal baseline (being the time separation between the compared images) and the time of the day of the acquisitions.
Current satellite observations are limited by their orbital constraints and fixed observation frameworks. For this reason, the analysis of the sensitivity  of HydroSoil data to the field and vegetation characteristics is a huge opportunity to fully explore this topic.

Apart from the vegetation parameters, the radar data is also influenced by the meteorological conditions of the scene under observation.
The long integration times required by all GEOSAR systems to generate the radar images may affect the sensitivity of the data to the scene properties, or at least they may compromise the observation of some events with a duration shorter than the acquisition time.
The motion of vegetation elements (leaves, branches), mainly induced by wind, generates a loss in coherence which may severely affect the image focusing due to the long integration time.
The effect of the presence of wind can be analysed in the frequency domain, attending to the Doppler spectrum of the data. Atmospheric phase screen is one of the main error sources of interferometric phase and it needs to be cancelled in interferometric techniques. However, the strong dependence of the atmospheric phase screen on atmospheric variations could allow obtaining meteorological information of the scene from radar data in continuous observation modes.

The retrieval of soil moisture is one of the man challenges of this project. The current most successful retrieval approaches for soil moisture based on time series of radar data have been developed and tested under very different circumstances with respect to the characteristics of our project (i.e., using Sentinel-1 images which are acquired every 6 days). For this reason, the applicability of the existing methods need to be tested on GEOSAR data.
The data obained by the radar system developed by our team is characterised by the short acquisition time of 10 minutes. The evaluation of the existing soil moisture retrieval methods (based on the backscattering coefficient or on interferometry) with this data will show the potential and limitations for the GEOSAR case.

The availability of dense time series of radar images can be further exploited for improving the retrieval of vegetation biophysical parameters (including height, biomass or vegetation water content) using conventional approaches. With the available data, tests are being carried out to quantify the required sampling rate for the diverse applications (retrieval of SM and vegetation parameters). Moreover, the existing methods adapted to the observation configuration of a GEOSAR are expected to maximise its performance thanks to the amount and type of input data.

UPC is responsible for the coordination of this project task.

Main objectives

• Analysis of the backscattering and phase changes over time and of their sensitivity to dynamic water processes, including precipitation, interception and surface soil moisture evolution (especially in the events of rain, dew and evaporation)

• Correlation between polarimetric channels as a function of surface parameters (crop structure, vegetation water content, soil roughness...) assessment

• Analysis of the InSAR temporal decorrelation analysis, as a function of scene parameters, temporal baseline and daily time

• InSAR phase capability to observe crop physiology (movement of water, controlled by vegetation, in the soil and plant during the day)

• PolSAR system sensitivity comparison of the different polarimetric configurations (single-, dual-, compact- and quad-pol) to the scene characteristics

• Shallow incidence angles influence on the radar sensitivity to soil and vegetation properties

• C and X bands sensitivity comparison

• Design and evaluation of retrieval algorithms (for soil moisture and vegetation variables) for data with fast acquisition interval (adapted to the specific features of GEOSAR systems)

• Effect of long integration times of the synthetic aperture on the evaluation of the resulting SAR images

• Analysis of the possibility of retrieving meteorological information from radar data in continuous observation modes