SAR Polarimetric Interferometry (Polinsar)
POLinSAR Workshop 2003
Session Title: SAR Polarimetric Interferometry
Chair(s): E. Pottier and K.P. Papathanassiou
All five papers in the session were presented. The first three dealt
with forest height estimation from Pol-InSAR data (papers 1 and 2 reviewed
the main results that have been achieved in the frame of the two ESA studies
on “Applications of SAR Polarimetry”). It was shown that forest height
estimation over temperate forest test sites can be achieved with an accuracy
of about 10% - independently on topographic variations and without using
any a-priori information. This was one of the most important results presented.
Also height validation results were presented based on direct coherent
modeling of forest scattering using detailed 3-D tree architectural models.
In paper 4, for the first time, airborne Pol-InSAR data at X-band has
been demonstrated and a first order comparison between DEMs at different
polarizations has been performed. Finally, in paper 5 the estimation of
building heights from Pol-InSAR data has been addressed.
Questions for Round Table discussion.
Question 1. What is the value added by Quad-Pol Interferometry
compared to Dual-Pol Interferometry and (Pseudo) Dual-Pol Interferometry?
Question 2. What is the effect of temporal decorrelation on the applicability
and on the inversion performance of the actual models?
Do we need single pass Pol-InSAR systems in space? Is their realization
Question 3. Are the available data sets complete / sufficient for the
validation / development of new applications?
Question 4. What is the status of direct coherent (polarimetric / interferometric)
modeling for Forestry, Agriculture, Ice applications?
The discussion about questions 1-3 can be summarized in following points:
- Fully polarimetric (Quad-Pol) Single-pass interferometry is the optimum
and desired configuration for the developed (forest parameter) inversion
- Some of the individual parameters (e.g. underlying ground topography)
may be estimated from reduced dual-pol observation vectors.
- The scheduled Quad-pol SAR missions (ALOS/PalSAR, RadarSAT 2, TerraSAR-X)
will operate in a repeat pass interferometric mode. A quantitative study
of the effects of temporal decorrelation on parameter retrieval accuracy
is therefore urgently required.
- The implementation of single-pass Quad-Pol interferometry based on
passive micro-satellite technology is not likely to happen in the near
future due to technological / financial limitations.
- Multi-baseline scenarios at two polarizations are more probable to
be realized in the near future in terms of passive micro-satellite technology.
- The available data sets over temperate forests are sufficient for
the validation of tree height estimation accuracy.
- Pol-InSAR data over Boreal and Tropical Forest environments complemented
by good ground measurements are missing completely.
- There is a demand for Pol-InSAR data (and ground measurements) for
the development of agricultural and land ice applications.
In parallel with future airborne trials and data analysis there is a
need to further develop validated coherent vector scattering models for
forestry, ice and agriculture applications of Pol-InSAR
Conclusions and Recommendations:
In the time between 1999 (CEOS Workshop in Toulouse, France) and 2003
(POLinSAR Workshop in Frascatti, Italy) Pol-InSAR has clearly matured (scientifically
and technologically) from a “first-results” to a “science“ status.
A Pol-InSAR Airborne validation Experiment over Tropical Forest is
highly recommended in order to prove the performance of Pol-InSAR inversion
as well as to compare the inversion performance at L-band and P-band.
Forcing the communication with bio- / geo-chemical experts in order
to establish algorithm and sensor specific requirements and to develop
Exploration of potential Agriculture and Ice applications of
Exploration of potential X-band Pol-InSAR applications. Especially
the combination of the long term temporal stability of point scatterers
with their strong polarimetric behavior, can be a challenge for high resolution
systems as TerraSAR-X.
The long-term goal is a space-borne single-pass fully polarimetric InSAR
Until then the potential off
· limited polarimetric InSAR configurations
· repeat-pass Quad-Pol InSAR configurations
has to be investigated / validated.
In summary we consider that POLInSAR, the combination of (quad- / light-)
polarimetry with (single- multi-baseline) interferometry is a significant
break through in quantitative parameter estimation of combined surface
and volume scatterering.