Master Thesis

Main content

The group of Earth Observation and Remote Sensing offers the following master topics at the moment. Students who are interested in these projects should contact Prof. Irena Hajnsek for further information.

General guidelines can be found here, while more information regarding citation etiquette can be found here.

Monitoring of agricultural crops using Satellite SAR data

Monitoring of agricultural crops using Satellite SAR data
X-band Pauli RGB of an agricultural testsite acquired by F-SAR

In agricultural resource management, soil parameters as well phenological information about agricultural vegetation play a key role. However, agricultural vegetation is very versatile and has a fast changing character during the phenological cycle. In most studies, data is only available on few dates, not fully representing the phenological cycle and therefore making conclusions difficult. Hence, it is crucial to have a time consistent dataset over a full growing season from sawing to harvesting. A time-series of TanDEM-X (German aerospace centre) and Radarsat-2 (Canadian space agency) data was collected over an agricultural testsite located in Germany accompanied by in-situ measurements for validation. This thesis work aims at an assessment of the polarimetric signatures of agricultural vegetation using the abovementioned satellite polarimetric SAR data. Special interest lies on the following points: 1. Variation of the polarimetric variables with time. 2. Comparison of the signatures for the different species of agricultural vegetation. 3. Differences regarding the sensors TanDEM-X (X-band, dual-pol) and Radarsat-2 (C-band, full-pol) and evaluation of their added value for monitoring of agricultural crops. The research work will be based in the German Aerospace Center (DLR), Wessling (Munich), Germany.

Requirements: The working language for this project will be English. Some experiences on programming (Matlab, C++, etc), remote sensing and image processing are very welcomed, but not compulsory.

Contact persons: Prof. Irena Hajnsek and Hannah Jeorg (DLR)

Observation of Paddy Rice Growth with Satellite Radar Data

Observation of Paddy Rice Growth with Satellite Radar Data  
Dual Polarimetric TerraSAR-X data of Ipsala, Turkey: Polarimetric and interferometric analysis of a satellite time series.

In the last years, temporal observation of agricultural fields has gained importance under the name of precision farming. While conducting in-field measurements of crop parameters are highly accurate, it increases the costs of analyses to a great extend. At this point, use of remote sensing techniques make precision farming applicable to larger scaled areas with much less cost and comparable accuracy.

In the literature, there are several remote sensing based researches that focus on the estimation of phenological parameters such as crop height, vegetative volume, yield and leaf area index (LAI). At this point, BBCH-scale (Biologische Bundesanstalt, Bundessortenamt und CHemische Industrie) has been defined to have a generalized perspective over the growth. It explains the phenological changes through growth of crops, by emphasizing the quantitative measures, such as: Number of leaves, number of tillers, grain size etc. The common property of all these parameters is that, they are all related to the physical structure of the plants. This relevance makes one able to detect and understand the changes in time-scale by means of polarimetric and interferometric properties of SAR (Synthetic Aperture Radar).

Paddy rice is of the most important crop in international agricultural market and recently it has been included in precision agriculture research. In terms of SAR, due to the complex nature of rice canopies, plants tend to have similar response in different growth stages. However, by combining polarimetric and interferometric SAR data, these complexity can be simplified. The objective of this work is to explain the physical reasons of the similarity and empirically distinguish the different growth stages with minimum number of descriptors. Polarimetric and interferometric methods will be used beside the field measurements.

Requirements: Programming knowledge in MATLAB or C/C++, interest in environmental systems and advanced English are prerequisites for this project. Project will be incorporation with Istanbul Technical University / Turkey.

Contact persons: Onur Yüzügüllü and Prof. Irena Hajnsek

Observation of the dry Lop Nur Lake in China

Observation of the dry Lop Nur Lake in China  
Lop Nur Lake observed by ALOS-PALSAR in HH polarization on 15 January 2011 [Remote Sens. 2014, 6(5), 4546-4562; doi:10.3390/rs6054546].

Lop Nur Lake is a large area located at the east of the Tarim Basin, northwest China. In the past, it was covered by water that went completely lost before 1972. The vanishing of Lop Nur Lake is still under investigation by the scientific community, raising questions regarding the reasons for its vanishing. Interestingly, it is still possible to find very wet soil under a dry soil layer. Understanding the distribution and evolution of such wet layer can provide indications regarding the motivations for the water vanishing. Synthetic aperture radar plays an important role in this context due to the capability of microwave to largely penetrate very dry soil, which allows to observe the underneath wet layer. Aim of the student is to analyze TanDEM-X dual polarimetric data and ALOS-PALSAR quad-polarimetric data acquired on the area, finding relationships of the radar observables with ground measurements carried out during the acquisitions. Such relationships may allow the estimation of soil conditions over the rest of the dry lake, providing a global view that cannot be achieved by ground measurements due to the largeness of the region.

Requirements: The working language for this project will be English. Some experiences on programming (Matlab, C++, etc), remote sensing and image processing are very welcomed, but not compulsory.

Contact persons: Armando Marino and Prof. Irena Hajnsek

Flexible and fast SAR imaging using factorization techniques

Flexible and fast SAR imaging using factorization techniques
SAR amplitude image of 90-degree curved sensor trajectory. Data: E-SAR, L-band. Processing: time-domain back-projection. The data set was processed directly to map coordinates using a DEM. The SAR amplitude image is shown on top of a 1:25'000-scale digital map of the area. Map reproduced by permission of swisstopo (BA081196).

SAR image generation from airborne high-resolution radar data is a demanding task in terms of finding an optimal trade-off between final image quality, and thus the choice of a suitable algorithm, on the one hand, as well as a reasonable processing speed in view of the large amount of data, on the other hand.

In recent years, time-domain based processing techniques that are able to cope with the difficult acquisition geometry occurring in airborne SAR imaging have gained increasing attention. Considerable speedup of these time-domain based approaches is obtained by applying so-called factorization techniques.

Within this master project various factorization techniques described in literature shall be explored in the context of non-standard airborne data acquisition geometries.

Requirements: Programming knowledge in either Matlab or C/C++ is a prerequisite for this project.

Contact persons: Othmar Frey and Prof. Irena Hajnsek

Parallelization processing of synthetic aperture radar (SAR) data using graphical processing units (GPU)


In the context of SAR imaging (see also Topic 1) general purpose computing on graphics processing units (GPGPU) is a potential alternative to CPU-based data processing: In SAR imaging the same instructions are typically executed on arrays of data. Therefore, the "single instruction multiple data" architecture of GPUs is an attractive means to speed up the SAR imaging process.

Within this project, different strategies shall be explored how to make use of the parallelization potential found in GPU hardware in the context of selected SAR image processing algorithms.

Requirements: Programming knowledge in either Matlab or C/C++ is a prerequisite for this project.

Contact persons: Othmar Frey and Prof. Irena Hajnsek

3D vegetation volume structure estimation using SAR tomography

Forests are semitransparent for microwaves at wavelengths of the order of decimeter to meter length. By using multiple parallel radar acquisitions three-dimensional images of vegetation volumes can be obtained.

Within this project, approaches to extract parameters describing the vegetation structure from three-dimensional polarimetric SAR data sets shall be investigated.

Requirements: Programming knowledge in either Matlab or C/C++ is a prerequisite for this project.

Contact persons: Othmar Frey and Prof. Irena Hajnsek

 
Page URL: http://www.eo.ifu.ethz.ch/studium/master-thesis.html
25.06.2017
© 2017 Eidgenössische Technische Hochschule Zürich