<< Click to Display Table of Contents >> Navigation:  Help Content > Interferometric Stacking > Post Processing Tools > Time Series Classification > Interferometric Stacking - Post Processing Tools - Time Series Classification - Classification

Purpose

 

This tool is used to generate a displacement classification based on the goodness of fitting according to the Akaike criterion. Each class of the classification identifies the areas that better fit the model.

 

Technical Notes

 

A classification map is created by selecting the best fitting model according to the Akaike Information Criterion which combines the fitting root mean square error (RMSE) with the number of free parameters in the model as following:

 

 

 

With:

 

 

 

the number of parameters

the number of measurements

the displacement obtained by the fitting model

the measured displacement

 

Input Files

 

Input File

Modeling meta files (*_meta) created the Phenomenological Modeling and/or the Analytical Modeling panels.

 

Parameters - Principal Parameters

 

Generate Classified TS

By setting this flag a composed fitted time series will be generated with the fitted time series of the best fitting model.

 

Majority filter

By setting this flag a majority filter will be performed with a filter window defined by MF window size.

 

MF window size

Dimension of the sliding window used to perform the majority filter.

 

L1 threshold

The L1 norm (normalized by the number of dates in the time series) displacement threshold to consider a pixel as a stable one. Pixels under this threshold will be considered as no displacement points in the classification map.

 

RMSE threshold

The maximum root mean square of allowed for a model to be considered. If all models yield an RMSE  over this threshold this pixel will be displayed as not classified on the classification map.

 

Output Files

 

A directory is created for each executed model combining the Output Root Name and the “_classification” string (I.e.: projectXXX_classification). Inside each of these directories you will find:

 

classification

A classification image showing the best fitting model for each pixel according to the lowest delta Akaike Information Criterion, non-classified areas, and stable areas.

 

classification_filtered

classification image filtered with a majority filter (if selected)

 

classification_daic

Delta Akaike Information Criterion value of the best fitting model for each pixel. *

 

classification_rms

RMS error of the best fitting model for each pixel. *

 

classification_input_TS

Best fitted time series images (if selected). This TS is a combination of the individual TS of each model using the TS of the best fitting model for each pixel. *

 

* if the classification majority filter is performed, these files correspond with this filtered classification

 

General Functions

 

Cancel        

The window will be closed.

 

Help        

Specific help document section.

 

 

 

Specific Function(s)

 

None.

 

See Also

 

Task, SARscapeBatch object, SARscapeBatch script example

 

References

 

De Grandi, A. (2019). PASTA - Phenomena Aware Spatial-Temporal Analysis, Bsc Thesis, Università degli Studi dell’Insubria, Italy.