<< Click to Display Table of Contents >> Navigation:  Help Content > Polarimetry and PolInSAR Module > Frequently Asked Questions - Pol&PolInSAR

Q. - What is the Faraday Rotation? Does SARscape take it into account??

 

A. - The Faraday Rotation, is an atmospheric effect which causes the polarization rotation of the scattering objects. SARscape does not foresee a specific functionality to correct these effects.

In case of ALOS PALSAR data generated by JAXA, these effect have been already compensated. A possible way to verify if a residual phase rotation is present in your data, it is to generate an interferogram between HV and VH, (for the mono-static sensors) polarizations: the average phase shall be different from 0 if a phase rotation exists.

 

Q. - Is it possible to first perform the Geocoding of the 4 SLC channels/polarizations and afterwards run the Polarimetric Decomposition?

 

A. - No, this is not possible since the decomposition process requires Single Look Complex data, while the result of the Geocoding is an amplitude (possibly multi looked) image. The geocoding process must be then executed after the Polarimetric Decomposition has been carried out.

 

Q. - How can I interpret the results obtained from the Polarimetric Decomposition?

 

A. - Any coherent polarimetric decomposition algorithm (e.g. Pauli Decomposition) tries to interpret the SAR measured signal contained in a full polarimetric dataset as linear coherent (i.e. including the phase) combination of elementary scatterings of some defined type. Three independent input channels (since the HV and VH are, for monostatic systems, fully correlated and thus identical) can be considered as a weighted sum of three types of pre-defined elementary scattering types. The difference among the existing decomposition methods stands in the elementary scattering mechanisms which are used as base to describe our data. For instance, according to Pauli, each pixel is considered as possible combination of some odd-bounce (e.g. single - planar or triple – corner reflector), even-bounce (e.g. dual – double bounce) and  even-45 degrees rotated / volume reflections. The goal of the decomposition algorithm is then to estimate for every pixel the weights of this combination.  

 

Q. - Is the Multitemporal Time Series De Grandi Filter suitable for filtering the Pauli Channels coming from different acquisition dates?

 

A. - Yes, this multitemporal filtering method can be adopted, given that the filter is executed by inputting homogeneous information (e.g. data with the same linear polarization or from the same component output of a decomposition, with same acquisition geometry, etc.). For instance you shall not mix in the same input list HH with VV polarizations or K1 and K2 decomposition channels.

 

Q. - The question concerns the Interferogram Generation and also the Coherence Optimization processes. Do I need to use different synthetic phase outputs: one calculated with the range and azimuth looks equal to 1 (Coherence Optimization step) and one with the range looks equal to 2 (PPD/Interferogram Generation step)?

 

A. - For what concerns the Interferogram Generation there are 3 mandatory inputs, which are the Reference and Secondary acquisitions (previously coregistered and thus overlooked 2 times in range) and the synthetic phase; this last file must have been previously generated using the coregistered Reference and Secondary data as inputs. Indeed, in order to have an output square pixel, it must be taken into account that the inputs (i.e. coregistered Reference and Secondary data) are oversampled in range direction, thus you have to multiply by 2 the range looks that you would normally adopt for those data. For instance, if you have PALSAR data that you normally multi-look with factors 7/1 (azimuth/range respectively), here you have to set the multi-look factors to 7/2. It is important to remember that the same factors (i.e. 7/2) have to be used both for the Synthetic Phase Generation and in the Interferogram Generation. A final useful note is that the interferogram can be generated, probably with less efforts, by means of the Interferometry module!

 

For what concerns the Coherence Optimization there are the same 3 mandatory inputs, generated in the same way as above, apart from the multi looking factors for the synthetic phase image; here indeed they must be set to 1/1 (azimuth/range). Afterwards, when the Coherence Optimisation process is executed, you must take into account the input data oversampling in range direction and consequently you must multiply by 2 the range looks that you would normally adopt for those data. The same example illustrated above for the PALSAR data, that you have to multi-look with factors 7/2 instead of 7/1, can be applied in this case. It must be finally noted that here the "Looks" button (for the automatic calculation of the multi looking factors) takes into account of the over sampling in the input data and, in the example above, it shall automatically set the values to 7/2.

 

Q. - How can I interpret the results obtained from the Coherence Optimisation?

 

A. - In terms of interpretation, the concept is similar to the "Polarimetric Decomposition" described above: we identify the most important scattering mechanisms (in this case we don’t even say which ones) that are contained in each cell; the mechanisms showing the highest coherence are separated from those characterized by respectively intermediate and lowest coherence levels. If different mechanisms exist in a cell (e.g. for penetration through a vegetation pixel, where some reflection comes from the crown and some from the trunk), they could have a different location in height (as the crown, the trunk and the underlying terrain), which is revealed by the interferometric phase difference of the different scattering mechanisms. Among others, possible use of this information are: i) tree height estimate; ii) classification based on the number (1, 2 or 3) of significant mechanisms that are contained in each pixel.

 

Q. - Is it possible to correlate the tree height with the output of the Coherence Optimisation?

 

A. - The result of this processing step consists of three interferograms and corresponding coherence images, one for each of the (possibly significant) optimized polarization pairs (one Reference and one Secondary); the differential interferograms between the first and the second one (in terms of coherence) is proportional to the trees height. In reality the coherence is strongly related to the temporal decorrelation, which in case of PALSAR pairs is quite notable and this factor must be taken into account.

 

Q. - How can I use the Polarimetric Phase Difference (PPD) and what are the information that I gather?

 

A. - The first consideration to do it is that, in a polarimetric acquisition, the (complex) correlation between one co-polar (HH or VV) and one cross-polar (HV or VH) channel is close to zero, unless we are dealing with a very low frequency acquisitions (lower than L band). Hence the major interest is for estimating the PPD of pairs of co-polar channels (HH and VV); this relates with the number of bounces of the scattering (90 degrees for each bounce): water will have a different phase respect to buildings, while volumes of vegetation will have a almost totally random phase, depending on their thickness. In the end PPD, and the corresponding coherence, can be used to help in discriminating objects of different kind and/or shape.