26. January 2016 at 09:15 in Physicum's aud B103 Karlis Zalite will defend his doctoral theses "Radar Remote Sensing for Monitoring Forest Floods and Agricultural Grasslands".
Dr Mart Noorma, Tartu University, Institute of Physics
Dr Kaupo Voormansik, Tartu Observatory
Dr Anu Reinart, Tartu Observatory
Dr Juan Manuel Lopez-Sanchez, University of Alicante, Spain
Dr Rivo Uiboupin, Tallinn University of Technology, Estonia
This thesis presents research about the application of radar remote sensing for monitoring of complex natural environments, such as flooded forests and agricultural grasslands. The study was carried out in Tartu Observatory, University of Tartu, Ventspils University College, and Aalto University. The research consists of two distinctive parts devoted to polarimetric analysis of images from a seasonal flooding of wetlands, and to polarimetric and interferometric analysis of a summer-long campaign covering eleven agricultural grasslands. TerraSAR-X data from 2012 were used to assess the use of the double-bounce scattering mechanism for improving the mapping of flooded forest areas. The study confirmed that the HH–VV polarimetric channel that is sensitive to double-bounce scattering provides increased separation between flooded and unflooded forest areas when compared to the conventional HH channel. The increase in separation increases with decreasing forest height, and it is more pronounced for deciduous forests due to the leaf-off conditions during the study. The phase difference information provided by the HH–VV channel may provide additional information for delineating flooded and unflooded forest areas. Time series of X-band (TanDEM-X and COSMO-SkyMed) and C-band (RADARSAT-2) data from 2013 were analyzed in respect to vegetation parameters collected during a field survey. The one-day repeat-pass X-band interferometric coherence was shown to be correlated to the grassland vegetation height. The coherence was also found to be potentially useful for detecting mowing events. The polarimetric analysis of TanDEM-X and RADARSAT-2 data identified two parameters sensitive to mowing events - the HH/VV polarimetric coherence magnitude and the H2α entropy. Mowing of vegetation consistently caused the coherence magnitude to decrease and the entropy to increase. The effect was more pronounced in case of X-band data. Additionally, the effect was stronger with more vegetation left on the ground after mowing. The effect was explained using a vegetation particle scattering model. The changes in polarimetric variables was shown to be caused by the change of orientation and the randomness of the vegetation.