AMPER
Dependence on the measurement interval of the coherence for HH and VV polarisations (red) and the Pauli base. It shows that the decorrelation is strongly dependent on the initial polarisation base, and that this is important for the correct extraction of information |
The AMPER project (Application of Multi-parameter Polarimetry in Environmental Remote Sensing) was funded by the European Commission over the period from January 2003 until December 2005. The Institute was responsible for the overall coordination of the project and research contributions in the field of modelling and understanding atmospheric effects and distortions on coherent polarimetric radar data, as well as in the field of polarimetric target decomposition theory.
The main aim of the project was the training and research of young scientists in the area of multi-parameter polarimetry at partner organisations spread all over Europe (i.e. DLR, TU Chemnitz and Definiens Imaging GmbH, Germany; University of Essex, UK; DDRE, Denmark; UPC, Spain; MOTHESIM and University of Rennes, France; JRC, Italy). The scientific activities of the network fall into three main areas dealing with sensor systems and measured data, the underlying physics and scattering models, parameter retrieval and product generation.
The focus of modelling atmospheric effects was on the scattering of hydrometeors, particularly rain drops. The main purpose is not only to have a coherent model for simulating weather radar data, but also to create a link between observations by weather radar systems and SAR systems. Polarimetric propagation effects play an important role for both systems, and, hence, the model can be used to predict or describe the influence of propagation through a medium containing rain drops for polarimetric radars, regardless of whether they are ground-based or airborne/spaceborne. Raindrops are oblate scatterers, where the oblateness increases with the size of the rain drop, and thus rainfall produces anisotropic scattering, which can be measured with polarimetric radars. One particular problem occurs, because the four elements of the scattering matrix are not all measured at the same time but with a certain time delay, which is usually around 1 ms. During this time, the observed scattering volume (containing falling rain drops) changes and, thus, decorrelation will be seen.
Using model results and experience drawn from the literature, phase delays and amplitude effects on radar data can be predicted. Special attention has been given to the impact of such effects on the processing of SAR images. Phase changes due to atmospheric distortions have a similar behaviour to motion errors of the sensor platform and thus might not be separated from such additional error sources.
With regard to polarimetric decomposition theorems, two signal processing approaches have been applied for the first time to fully polarimetric SAR and weather radar signatures. Considerable attention is paid to the eigenvectors of the covariance matrix.