Adugna Mullissa

Adugna Mullissa
Wageningen University & Research | WUR · Laboratory of Geo-Information Science and Remote Sensing

Ph.D.

About

18
Publications
10,249
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
304
Citations
Citations since 2016
17 Research Items
302 Citations
2016201720182019202020212022020406080100120140
2016201720182019202020212022020406080100120140
2016201720182019202020212022020406080100120140
2016201720182019202020212022020406080100120140
Introduction
Adugna Mullissa is a researcher in 'Radar and Deep Learning Methods for near-real time forest change monitoring' at the Laboratory of Geo-information Science and Remote Sensing, Wageningen University (2019-Present). He obtained his PhD (2017) in radar remote sensing from the University of Twente, Enschede, The Netherlands. https://www.wur.nl/en/Persons/Adugna-dr.-AG-Adugna-Mullissa.htm.

Publications

Publications (18)
Article
Deep learning has successfully improved the classification accuracy of optical remote sensing images. Recent works attempted to transfer the success of these techniques to the microwave domain to classify Polarimetric SAR data. So far, most deep learning networks separate amplitude and phase as separate input images. In this article, we present a d...
Article
Deep learning (DL) has proven to be a suitable approach for despeckling synthetic aperture radar (SAR) images. So far, most DL models are trained to reduce speckle that follows a particular distribution, either using simulated noise or a specific set of real SAR images, limiting the applicability of these methods for real SAR images with unknown no...
Article
A polarimetric synthetic aperture radar (PolSAR) sensor is able to collect images in different polarization states, making it a rich source of information for target characterization. PolSAR images are inherently affected by speckle. Therefore, before deriving ad hoc products from the data, the polarimetric covariance matrix needs to be estimated b...
Article
Full-text available
Sentinel-1 satellites provide temporally dense and high spatial resolution synthetic aperture radar (SAR) imagery. The open data policy and global coverage of Sentinel-1 make it a valuable data source for a wide range of SAR-based applications. In this regard, the Google Earth Engine is a key platform for large area analysis with preprocessed Senti...
Article
Full-text available
Analysis of forest disturbance patterns in relation to precipitation seasonality is important for understanding African tropical forest dynamics under changing climate conditions and different levels of human activities. Newly available radar-based forest disturbance information now enables an investigation of the intra-annual relationship between...
Article
Full-text available
The paper evaluates Deep neural network architectures that account for either (a) spatial-temporal information, i.e., Hybrid Recurrent convolutional neural network, 3D-convolutions, ConvLSTM, and the novel CNN + Multi Head Self-Attention model, or (b) only spatial information, i.e., 2D-convolutions, (c) only temporal information, i.e.,Long short te...
Preprint
Full-text available
A Polarimetric Synthetic Aperture Radar (PolSAR) sensor is able to collect images in different polarization states, making it a rich source of information for target characterization. PolSAR images are inherently affected by speckle. Therefore, before deriving ad- hoc products from the data, the polarimetric covariance matrix needs to be estimated...
Article
Full-text available
A humid tropical forest disturbance alert using Sentinel-1 radar data is presented for the Congo Basin. Radar satellite signals can penetrate through clouds, allowing Sentinel-1 to provide gap-free observations for the tropics consistently every 6-12 days at 10 m spatial scale. In the densely cloud covered Congo Basin, this represents a major advan...
Preprint
Full-text available
Deep learning (DL) has proven to be a suitable approach for despeckling synthetic aperture radar (SAR) images. So far, most DL models are trained to reduce speckle that follows a particular distribution, either using simulated noise or a specific set of real SAR images, limiting the applicability of these methods for real SAR images with unknown no...
Article
Full-text available
This article provides an angular-based radiometric slope correction routine for Sentinel-1 SAR imagery on the Google Earth Engine platform. Two established physical reference models are implemented. The first model is optimised for vegetation applications by assuming volume scattering on the ground. The second model is optimised for surface scatter...
Conference Paper
Full-text available
Classification of crop types from multi-temporal SAR data is a complex task because of the need to extract spatial and temporal features from images affected by speckle. Previous methods applied speckle filtering and then classification in two separate processing steps. This paper introduces fully convolutional networks (FCN) for pixel-wise classif...
Thesis
Full-text available
Ground deformation measurements deliver valuable information for the understanding of natural hazards such as landslides, ground subsidence, earthquakes and volcanism. Satellite based SAR interferometry provides detailed, cost effective measurements for this purpose. The quality of these measurements, however, is often degraded by decorrelation. He...
Article
Reliability of the scattering model based polarimetric SAR (PolSAR) speckle filter depends upon the accurate decomposition and classification of the scattering mechanisms. This paper presents an improved scattering property based contextual speckle filter based upon an iterative classification of the scattering mechanisms. It applies a Cloude-Potti...
Presentation
Full-text available
PS-InSAR interferometry is a well-established technique to estimate linear and non-linear ground displacements as well as the atmospheric phase screen (APS) in an InSAR data. It achieves the highest accuracy in measuring deformation. To reach this accuracy a high density of high quality points (PS) is required for model fitting. In rural regions, w...
Article
Ground deformation measurements have contributed to a better understanding of the processes and mechanisms involved in natural hazards. Those include landslides, subsidence, earthquakes and volcanic eruptions. Spaceborne Differential Interferometric Synthetic Aperture RADAR (DInSAR) is a well studied technique for measuring ground deformation. Qual...
Book
In developing countries where agriculture serves as a backbone of the economy the availability of groundwater resources is crucial. However, the high cost associated with conventional field based groundwater exploration techniques has made the assessment of these resources difficult. This book, therefore, provides an integrated remote sensing and G...

Network

Cited By

Projects

Project (1)
Archived project
Ground deformation measurements deliver valuable information for the understanding of natural hazards such as landslides, ground subsidence, earthquakes and volcanism. Satellite based SAR interferometry provides detailed, cost effective measurements for this purpose. The quality of these measurements, however, is often degraded by decorrelation. Here quality is defined as the number of measurement points and accuracy of deformation measurements from these points. The degree of decorrelation is in particular severe in natural environments where the number of coherent targets is limited. Hence, improving the quality of deformation measurements in natural environments is important to improve their usability. This research exploits the polarimetric diversity provided by fully and partially polarized SAR images to increase the number of measurement points and deformation measurement accuracy. It is composed of three major topics. First, the performance of three polarimetric optimization methods has been analyzed and evaluated to select the polarization state that is least affected by decorrelation. This was done in a single and double phase center scenario. These scenarios are applied on dual and quad polarized SAR images acquired from a completely natural environment in Ethiopia. The purpose was to quantify the coherence improvement in different types of distributed scatterers. Bias in the coherence estimation of the different optimizers was investigated from simulated PolInSAR image. Deformations observed from three different polarimetric optimization methods were compared with the deformation values obtained from an in-situ GPS. The applied polarimetric optimization routines substantially improved coherence estimation. These routines have a lower estimation bias as compared to the traditionally used single polarization channel. Second, we developed a spatial filter using statistical homogeneous pixels (SHP) based upon the scattering mechanism to improve the signal to noise ratio of interferometric phase over distributed scatterers. This method was first developed to estimate polarimetric coherency matrix for a single image. It improved the estimation of coherency matrix by avoiding the indiscriminate use of moving averages. Improvement was achieved by iteratively classifying and refining the estimate of the coherency matrix. It derived the scattering mechanism by applying the Cloude-Pottier eigenvalue-eigenvector decomposition technique whereas the Wishart distance measure was used to classify the scattering mechanism. Classification of scattering mechanisms was used as a basis to group similar pixels for the spatial filter. To reduce bias caused by pixel selection and that is generated by an incorrect decomposition and classification of scattering mechanisms, an iterative refinement of decomposition and classification was used. The method was implemented on both simulated and real PolSAR images acquired from the San-Francisco area, USA and Flevoland, The Netherlands. The results were compared with other state of the art spatial filters. Results indicated that the proposed method compares favorably with other state of the art spatial filters in preserving polarimetric information, spatial details and point scatterers. Third, we adapted the scattering mechanism based spatial filter to improve the signal to noise ratio of interferometric phase over distributed scatterers and estimate interferometric coherence matrix. Selection of the most coherent scattering mechanisms within a distributed scatterer candidate is done by applying an eigenvalue decomposition of the interferometric coherence matrix. To identify the distributed scatterer candidate for optimization we derived a phase entropy measure to be applied as a threshold. Coherent scatterers were selected by identifying scattering mechanisms in the resolution cell that interfere with the dominant scattering mechanism. Performance was evaluated on full and dual polarized SAR images acquired over Los Angeles area, USA and Groningen, The Netherlands. Number of measurement points and deformation estimates were compared with those of traditional PSI methods. Results indicate that the proposed method substantially improved the number of measurement points and the deformation estimate. To summarize, this project contributes to improving the quality of deformation measurement in natural environments from both fully polarized and partially polarized SAR images.