In this paper, an in-depth analysis is presented on the role of the phase in ship classification using single-polarisation synthetic aperture radar (SAR) images. To conduct the analysis, a novel lightweight convolutional neural network (CNN) model, LeNet-2AP, is introduced. Based on the LeNet-5 architecture, the proposed model processes a two-channel amplitude and phase input through two distinct and independent branches, which are merged only at the final stage before the softmax layer. Several modern enhancements, including weight standardisation and global average pooling, are incorporated to improve performance and efficiency. The model is evaluated on the OpenSARShip dataset using single-look complex (SLC) samples and tested against amplitude-phase, in-phase and quadrature-phase (I-Q), and complex-valued inputs, achieving superior performance compared to existing models in the literature. Consequently, Shapley additive explanations (SHAP) is applied to analyse the model’s decisions, focusing on the contribution of the phase to the classification process. It is demonstrated that phase information is distributed across an extensive area beyond the ship’s amplitude signature in the SAR image. Furthermore, the phase is found to provide a comparable level of discriminative information to the amplitude, challenging the traditional approach of using the amplitude exclusively. These findings highlight the significance of the phase as a complementary source of information in single-polarisation SAR images and demonstrate that shallow, optimised CNNs are better suited to efficiently extract this information with limited data samples.

The importance of mapping the forest height (FH) is increasing due to the more frequent impacts of climate change in the society (wildfires, droughts, and extreme weather events). Remote Sensing is often used for mapping this variable; however, it usually relies in costly and extensive field or airborne campaigns. In addition, when using synthetic aperture radar (SAR), most approaches do not use freely available data. Considering this, in this work a model is proposed that resorts to Advanced Land Observing Satellite 2 (ALOS-2), Sentinel-1 (S1), and ancillary data. Airborne laser scanning (ALS) data are used for local calibration but, with the aim of developing a more scalable model, the latter is optimized to work with small calibration datasets (representative of just 25% of the study area to be mapped). With this purpose, the model combines a featuring generation and a features’ processing stage with a stacking regressor to produce estimates at the pixel level. Their impact was assessed, and an improvement of 8.11 and 2.01 pp in the relative root mean square error (rRMSE) was achieved by including the features’ generation and features’ processing stages, respectively. In addition, when the multifrequency dataset was used, the model achieved an rRMSE better than when using only a C-band dataset (S1) or only an L-band dataset (ALOS-2), respectively, by 4.21 and 3.05 pp. Finally, the model achieved an average${R} ^{2}$/rRMSE of 0.6240%/24.30% and 0.5901%/22.64% for the validation and test study areas, respectively. The proposed approach revealed to be effective on mapping the FH resorting to multifrequency SAR and small calibration datasets acquired by ALS.

Illegal Unreported and Unregulated (IUU) fishing is responsible for considerable economic losses in developing countries due to poor governance and limited resources for surveillance, especially in presence of vast Exclusive Economic Zones (EEZs). Recent studies conducted on behalf of the Food and Agriculture Organization (FAO) show that global figures of economic losses are inappropriate due to the varying availability of resources and regulations from country to country while methodologies and indicators applied locally, informing about the magnitude and impact of IUU fishing activities in a prescribed region, would be more robust. Following FAO guidelines, a novel methodology based on anomalies found in Synthetic Aperture Radar (SAR) and Automatic Identification System (AIS) data is here proposed to estimate losses to IUU fishing in Mauritius EEZ. Results from an exercise, conducted in Saya de Malha (Mauritius) from June to November 2023, show a worst-case scenario of potentially more than four-thousand tonnes of IUU caught which would translate in economic losses of more than 21 millions of US dollars.

This study introduces Bayesian Convolutional Neural Networks (BCNNs) for SAR ship classification, comparing their performance with traditional CNNs using the LeNet-5 architecture and VGG16 as a benchmark. It highlights BCNNs’ capability in uncertainty decomposition and improved model accuracy, demonstrated using the OpenSARShip dataset. Though further investigations using more complex datasets and additional network designs are necessary, the presented work highlights the potential of BCNNs in producing more reliable predictions in SAR ship classification.

Illegal, Unreported and Unregulated (IUU) fishing is a major global issue, costing up to £23 billion in the global economy per year [1]. Although the detection of vessels using Synthetic Aperture Radar (SAR) and Automatic Identification System (AIS) data has gained a lot of attention in recent years, the identification of illegal activities still remains largely unaddressed. This paper presents a model for the detection and identification of anomalous AIS shut-offs and transhipments around the Exclusive Economic Zone (EEZ) at Mauritius as part of the NEREUS project of the University of Surrey. The model combines, in a novel way, various techniques established in literature that consist of target detection in SAR images, vessel classification, construction and extrapolation of vessel tracklines using AIS data and AIS-SAR data association. The proposed model is preliminarily validated using NovaSAR data, as proof of concept, with varied results, that recommend for a change in the detection block.

The impacts of the climate change in the society make forest monitoring increasingly important. Consequently, there is a growing interest in mapping variables as the Forest Height (FH). The direct measurement of the FH through field campaigns is expensive and difficult to scale. Alternatively, Airborne Laser Scanning (ALS) campaigns can be used to map it, however they share the same disadvantages of the previous approach. Therefore, Remote Sensing (RS) data have been used for local and large-scale mapping of the FH. In this paper a Regression Methodology (RM) that combines GEDI, ALOS-2, Sentinel-1/2, and ancillary data is proposed for mapping the FH in Mediterranean forests. The proposed RM, tested for the 15 regions of interest, achieves a RMSE/rRMSE of 4.95m/33.93%, when evaluated with GEDI data, and 5.11m/41.70%, when evaluated with ALS data.

This research proposes an anomaly detection workflow intended primarily for integration into satellite-based tip and cue services for maritime surveillance. The workflow is centred around a ship trajectory prediction model which is designed to respond to anomalous events such as AIS "shut-off" events. This is important for accurately predicting the trajectories of potentially suspicious vessels that are moving, and subsequently scheduling the tasking of satellite acquisitions to monitor these vessels. The research implements a ship trajectory prediction model based on AIS data using Recurrent Neural Networks (RNNs). The efficacy of the model is evaluated based on the mean great circle distance between predicted and actual vessel trajectories, demonstrating satisfactory performance for typical motion patterns while acknowledging certain limitations in prediction accuracy. Overall, this study represents a significant step forward in the integration of imaging satellites and AIS data for maritime surveillance, offering a promising approach for anomaly detection and improving the efficiency of satellite-based monitoring systems. A GitHub repository containing the source code and related materials for this work is made available.

Illegal Unreported and Unregulated (IUU) fishing is a major threat to ocean biodiversity and preservation. A UK-Mauritius team is joining forces to develop a satellite-based monitoring solution that can improve maritime domain awareness in Mauritius Exclusive Economic Zone (EEZ) since official records of authorized fishing vessels are outdated or incomplete. This paper applies a previously developed methodology for Automatic Identification System (AIS) and Synthetic Aperture Radar (SAR) data matching through Artificial Intelligence (AI) to the case study area showing the potential of the technologies and techniques in detecting anomalies.

In this paper, a novel algorithm for ship classification in Sentinel-1 synthetic aperture radar (SAR) images is presented. The algorithm utilises layover as the main classification feature, which is based on the different relative heights of superstructures in oil tankers, container ships, geared and gearless bulk carriers. The algorithm has been tested using 20 ship samples from Sentinel-1 stripmap images over the port of Santos, divided equally amongst the four ship classes. An overall classification accuracy of 75% has been achieved.

Canopy Height (CH) is an important variable in any forest inventory, not only by its own information, but also as a proxy variable to estimate other parameters as the above-ground biomass. The CH information can also be helpful to understand the climate change trends, for forest management, and in decision support systems related to wildfires. The growing availability of Remote Sensing observations acquired from different sensors, create an alternative for the CH mapping to field campaigns and Airborne Laser Scanning (ALS) missions. Here a comparison between using Multispectral and Synthetic Aperture Radar sensors for CH estimation is presented. Both used the same Regression Methodology, being achieved a R2/RMSE between 43.71%-72.85%/0.85-4.03m for Multispectral and 42.12%-62.62%/0.96m-4.49m for SAR, for a total of 17 regions of interest. It is concluded that Multispectral data revealed to be more suitable for the CH mapping.

The knowledge of the Forest Height (FH) is important for monitoring the forests, and it can be used as a proxy variable of other forest parameters as the aboveground biomass. It is also important for understanding the climate change and prepare the wildfire seasons. The most effective way to map the FH is through field campaigns or airborne laser scanning, but both are expensive and not scalable. Alternatively, spaceborne Synthetic Aperture Radar (SAR) data may be used. However, it often relies on the acquisition of large ground truth datasets. In this paper, a new Regression Methodology (RM) that makes use of SAR data and a Stacking Regressor that minimises the amount of data needed to map the FH of a region is presented. Tested on a total of 16 regions between Portugal and Spain, plus one in California, the RM achieved a R2between 42.12%-62.62%, and a RMSE between 0.96m-4.49m.

The United Nations Development Programme (UNDP) launched the Ocean Innovators program to combat illegal fishing and destructive fishing practices, benefiting Small Island Developing States (SIDS) and Least Developed Countries (LDCs). One of the selected projects, ‘Nereus’, currently being developed by Surrey Space Centre (SSC) and Mauritius Research and Innovation Council (MRIC), utilises AI and satellite data fusion to monitor fishing vessel activity in Mauritius’ Exclusive Economic Zone (EEZ) and Marine Protected Areas (MPAs). The project combines various satellite technologies, including Synthetic Aperture Radar (SAR), Automatic Identification System (AIS) and Vessel Monitoring System (VMS). This paper analyses multi-temporal SAR and AIS data to identify "dark" ships that are not transmitting AIS signals. The methodology is applied to the Mauritius EEZ and MPAs, providing authorities with valuable information for informed decision-making and effective Maritime Domain Awareness (MDA).

The use of multi-temporal Synthetic Aperture Radar (SAR) imagery to assess damage caused by directed attacks in conflict zones is explored. This paper focuses on the Russia-Ukraine war as an example and emphasises the need for a reliable method to measure damage to urban infrastructure. The study presents a methodology that utilises a technique called Coherent Change Detection (CCD) using SAR imagery from Sentinel-1 to assess damaged areas in the city of Mariupol, Ukraine. The authors acquired SAR images before and after an artillery shelling event and measured the change in coherence between these images to assess the damage. They also compared the SAR results with contextual information from media reports and community-based projects to validate the findings. The paper provides specific examples of damage level classification maps for various types of infrastructure, such as a metallurgical factory, shopping mall and a maternity hospital. The results show good visual correlation between the bomb impact locations and the severity of damage. The authors conclude that multi-temporal SAR can complement other sensors in damage assessment mapping, especially in adverse weather conditions. Future work will focus on improving the damage assessment index and validating the damage level thresholds.

The monitoring of ships which do not report their Au-tomatic Identification System (AIS) information is important for Maritime Domain Awareness (MDA). In this paper, an improved maritime picture is generated by presenting a new methodology to map these so-called ‘dark’ ships over time. Firstly, a robust and accurate data association between Syn-thetic Aperture Radar (SAR) ship detections and AIS data is carried out on multi-temporal SAR imagery and AIS data. Subsequently, Kernel Density Estimation (KDE) is applied to unassigned SAR ship detections to reveal the spatial distri-bution of ‘dark’ zones (i.e. areas where repeated unassign-ments occur). This analysis helps identify areas where ships frequently do not report, which can help guide authorities in the best way to respond. The methodology is validated using Sentinel-l Interferometric Wide (IW) swath mode products and AIS data acquired from the English Channel, UK.

A novel alteration of the Water Cloud Model (WCM) and its inversion is proposed in this research work to improve the accuracy of Soil Moisture Content (SMC) mapping. This paper suggests using the Optical Trapezoid Model as the sole vegetation descriptor in the WCM, as well as the use of a specific configuration of parameters derived from a function of radar frequency, polarization, dielectric particle size, and orientation distribution. The proposed inversion scheme is applied on Sentinel-1 data along with the Landsat-8 images component. The proposed approach achieves higher SMC estimation accuracies compared to those produced by tested methods. Indeed, the designed approach achieved improvements in terms of accuracy as demonstrated by a decrease of Root Mean Square Error values in the order of 0.3% and 0.55% in the Blackwell farms and Sidi Rached study areas respectively.

GNSS reflectometry, whilst originally envisaged for ocean wind speed sensing, has recently been shown to be sensitive to land parameters such as soil moisture. Soil moisture is an important variable for many use cases including climate change monitoring, and as such there is a need to reduce gaps in datasets of this variable collected by satellites. By implementation on small platforms, GNSS-R missions can address this need, but current instrumentation must be updated to allow prediction of reflection points over the land surface. This paper presents an algorithm for achieving this along with results from both software testing and initial on-board implementation on DoT-1. These show that when Delay-Doppler maps are generated using the new algorithm the peak reflected power is successfully captured (in line with platform constraints) in 55% of software tests, compared with just 10% for the current method. Telemetry from DoT-1 shows that the algorithm has been successfully incorporated into the flight software. Future tasks to verify the on-board performance and improve the algorithm further are also discussed.

The use of airborne Mid-Wave InfraRed (MWIR) imagery for maritime applications such as ship detection is evaluated for future spaceborne missions. A flight campaign was carried out on 30 October 2020 over the Solent (UK) to provide a dataset of airborne MWIR that was acquired at the same time as a spaceborne Synthetic Aperture Radar (SAR) product. The thermal infrared imagery is compared to SAR to determine its suitability for ship detection. Automatic Identification System (AIS) data was also acquired and used to both identify and validate the ship detections where available. The results indicate that high spatial resolution satellite thermal infrared imagery can be a useful data source for ship detection.

This paper introduces a new multilevel fusion approach for Soil Moisture Content (SMC) estimation. This approach includes the following indices and models; the Normalized Difference Water Index (NDWI), the Modified Normalized Difference Water Index (MNDWI), the Perpendicular Drought Index (PDI), and the Temperature Vegetation Dryness Index (TVDI), extracted from Landsat-8 data, and the inversion of the Integral Equation Model (IEM) from Sentinel-1 data with the support of surface roughness measurements. The proposed fusion occurs on the feature and decision levels. At the feature level, features extracted from each of the above indices/models are combined to obtain three feature vectors, those vectors are later used in the decision level via the Fully Constrained Least Squares (FCLS) technique. The areas of interest of this study are; Blackwell Farms, Guildford, United Kingdom, and Sidi Rached, Tipasa, Algeria. The proposed system yielded lower Root Mean Square Errors (RMSE) (1.09% on average) than that of IEM inversion.

Change detection is one of the most addressed topics in the remote sensing community. When performed on synthetic aperture radar images, the most critical issues are as follows: 1) the labeling of the identified changing patterns and 2) the scarce robustness of classic pixel-based approaches based on threshold segmentation of an appropriate change index, which tend to fail when multiple changes are present in the study area. In this work, a new methodology for unsupervised change detection in vegetation canopy is presented. It overcomes these limitations by exploiting multitemporal geographical object-based image analysis with the aim to make the intrinsic semantic of data emerge and direct the processing toward the identification of precise classes of changes through dictionary-based preclassification and fuzzy combination of class-specific information layers. The proposed methodology has been tested in ten different experiments covering agriculture and clear-cut deforestation applications. The results, validated against literature methods, highlighted the superiority of the proposed approach, which was quantitatively assessed in terms of standard classification quality parameters. On agriculture experiments, it allowed for an average increase in the detection accuracy of about 11% with respect to the best performing literature method, with an increment of the false alarm rate in the order of 0.5%. In case of deforestation, the registered detection accuracy was comparable to that achieved by the literature, while the most significant benefit was the reduction, of more than one-third, of the number of detected false deforestation patterns. Overall, the main characteristics of the proposed architecture are the robustness and the lack of any supervision, which makes it very well-suited for operational scenarios.

In the above article[1], the author affiliations were incorrectly listed. The correct affiliations are as follows:

The paper presents an automatic algorithm tackling the problem of ship detection in situations where the ship signature itself is not visible. At this purpose, the algorithm proposed uses a combination of image processing techniques in order to identify ships exclusively by detecting the wakes they leave behind. Once a ship is identified, its position in the image, its heading and its speed are determined. The algorithm was developed and tested on SAR imagery from the European Space Agency (ESA) mission Sentinel-1. Here the algorithm's design and testing results are presented and its performance discussed with regard to the overall ship detection and the wake's components recognition. The algorithm resulted in a correct ship detection in 95% of the cases tested and in a correct recognition of the wakes' components in more than 80% of the cases analysed.

In the event of an accidental oil spillage from buried pipelines, the leakage may be detected fairly late, sometimes months after it started, with economic and environmental impacts that are difficult to recover. An early detection is desirable but difficult to achieve when the pipelines are buried, hidden by thick vegetation or in quite isolated areas. If the extent of buried pipelines network is also considered the problem may appear cumbersome. Satellites may support an early detection and, a feasibility study proving so, is described in this paper. A controlled spillage exercise has been organized in the UK and satellite data acquisitions tasked. Synthetic Aperture Radar datasets have been acquired over the site, before and after the controlled spillage, in three different tests with different quantities of red diesel spilled. The design of the whole exercise is described in the paper in addition to the SAR data processing. Preliminary results show that a significant change in the mean backscattering coefficient (a decrease of about 1dB) is appreciated in X-band SAR data when the volume of oil, spilled or poured, reaches values in the order of 160 gallons.

Research efforts have been turning in recent times to the use of Global Navigation Satellite System Reflectometry (GNSS-R) for sensing land parameters such as soil moisture and above ground biomass, which are essential for climate modelling. GNSS-R instrumentation to date has been designed with the purpose of sensing ocean parameters (e.g. wind speed). In order to enable operational, spaceborne GNSS-R missions for land-sensing, upgrades to instrumentation are required. One aspect is the prediction of reflection points, which over the land is affected by the presence of topography; a problem not encountered when predicting reflection points over the ocean. This paper presents an algorithm which enables accurate prediction of reflection points in areas of topography to enable real-time, on-board production of Delay-Doppler Maps of land reflected signals, a development which is critical for enabling operational land-sensing GNSS-R missions.

A novel SAR-AIS data association technique is proposed consistent with being used in dense shipping environments, where association of SAR and AIS datasets is non-trivial and SAR false alarm rates are typically high. A ship classification model based on transfer learning classifies ship types in SAR imagery. The classification results are subsequently used in the SAR-AIS data association, which uses a rank-ordered assignment technique. The methodology is validated using a Sentinel-1 SAR product and terrestrial-based AIS product acquired from the Gulf Coast, USA. Results show optimal data association which is improved using class (i.e. ship type) information.

Land cover mapping is one of the classic applications of synthetic aperture radar remote sensing. However, despite of the algorithmic progress in classification techniques, the semantic content of available maps does remain unchanged, with only a few macro-classes (like water, forest, urban, and bare soil) being discriminated in the majority of the works from past years. In this paper, a methodology to extract a higher level semantics from synthetic aperture radar images is presented. It is based on coupling pixel-based clustering with object-based image analysis and contextual information. Preliminary results have been produced from multitemporal SAR datasets over a forest area in Colombia. They demonstrate that the synergic exploitation of pixel and object information can provide higher quality land cover results and more information to map users.

The Urban Heat Island (UHI) effect is defined as an increase of the air and surface temperature inside a city compared to surrounding rural areas. This increment can be of several degrees, thus exposing populations to serious health risks, especially in hot developing countries, where the majority of the world's megacities are located. The UHI effect has been widely studied in the past with local methods employing field sensors. The use of satellites moved the analysis from local to city scale, but long-term investigations have been so far limited by storage and computational capacities. In this work, both ESA and NASA heritage data are used to study temporal evolution of the UHI of the city of Chennai, India over a 14-year period. The Google Earth Engine is exploited to process the available large dataset in a reasonable time. Results show that the UHI of Chennai has grown of 450% over time and that its main drivers are average temperature and city expansion.

This paper aims at developing a new enhanced algorithm for mapping semi-arid areas based on fusion techniques of Synthetic Aperture Radar (SAR) and Light Detection And Ranging (LIDAR) datasets. Firstly, both datasets are preprocessed to remove geometric and radiometric errors; then features of interest are extracted from SAR and LiDAR products to build masks and identify meaningful classes. Finally, classification results are refined with morphological filters. The new algorithm has been tested on data acquired by TerraSAR-X and an airborne LiDAR sensor over the Natural Reserve of Maspalomas in Canary Islands. Results show an overall classification accuracy of 85% with an absolute increment of more than 14% compared to a classification in which only LiDAR data are used.

In the Synthetic Aperture Radar (SAR) framework many detection algorithms and techniques have been published in the recent literature; however the detection of vessels whose dimensions are in the order of the image spatial resolution is still challenging in rough sea state scenarios. This issue is addressed in the paper presented here by comparing rationale and performance of two detectors developed by the same authors: the Generalized Likelihood Ratio Test (GLRT) and the Intensity Dual-Polarization Ratio Anomaly Detector (iDPolRAD). Both detectors are tested on a dual-polarization VV/VH Interferometric Wide Swath Sentinel-1 image acquired over the Suruga Bay on the Pacific Coast of Japan. The theory is presented here and the two detectors are compared against the Cell Average-Constant False Alarm Algorithm (CA-CFAR) showing both better performance than CFAR in terms of false alarms rejection.

In this research the best techniques of fusion for near-contemporaneous Synthetic Aperture Radar (SAR) and Automatic Identification System (AIS) datasets are studied to simulate the expected performance from NovaSAR-1. Specifi-cally, data association techniques are quantitatively compared by performing a series of Monte Carlo tests. The evaluation has been carried out using a satellite-based AIS dataset acquired from the English Channel on 07 June 2016, and SAR ship detections are simulated by reckoning the AIS dataset forward in time along a geodesic on a WGS84 reference ellipsoid. Accurate data association is achieved using an m-best multidimensional assignment technique, which is consistent with being used in an operational environment, especially in high-density shipping areas.

A new methodology, based on the International Organization for Standardization (ISO) Guide to the expression of Uncertainty in Measurement (GUM), for the analysis of uncertainties in Synthetic Aperture Radar (SAR) remote sensing data is presented. The principal idea is to assess the sources of uncertainty, developing computational approaches to propagate uncertainties through the whole SAR signal processing chain. The final goal is to construct uncertainties budgets to quantify the total uncertainty of SAR products, permitting their `trace-ability' to international reference standards.

A novel decision level data fusion algorithm for soil moisture content estimation is proposed in this paper. Firstly, individual estimations are determined, respectively, from the inversion of the Integral Equation Model (IEM) for Sentinel-l and from the Temperature Vegetation Dryness Index (TVDI) for LANDSAT-8. Then, a feature level fusion of these methods is performed using an Artificial Neural Network (ANN). Finally, all estimations including the feature level fusion estimation are fused at the decision level using a novel weights based estimation. The area of interest for this study is Blackwell Farms, Guildford, United Kingdom and datasets were taken on 17/11/2017 for both Landsat-8 and Sentinel-1. Estimation from the proposed decision level fusion method produces a Root Mean Square Error RMSE (1.090%) which is lower than RMSE of the individual estimations of each sensor as well as that of the feature level fusion estimation.

This paper presents a ship-detection study with Synthetic Aperture Radar (SAR) images acquired at two different frequencies: X- and C-band. The detection procedure relies on a novel algorithm based on the likelihood functions of both canonical ship target and sea clutter. Spaceborne images were acquired over the same area in the Solent Channel in UK at approximately the same time on the 7thJune 2016. Here, datasets are compared in terms of probability of detection (PD), probability of false alarm (PFA) and Target-to-Clutter Ratio (TCR). Detection maps are validated with Automatic Identification System (AIS) data when available and preliminary results show a higher TCR for the X-band SAR image.

In June 2014 an airborne Synthetic Aperture Radar (SAR) flight campaign was run in the south UK in order to acquire S-band datasets on several scenarios and for different applications in preparation for the launch of the first UK SAR mission NovaSAR-S. On request, X-band data could be acquired concurrently. This paper shows some of the results of the research project AS14-12 aimed at assessing S-band performance in retrieving soil moisture in bare or poorly vegetated areas. Comparison with X-band dataset was also among the objectives and, at this purpose, the Integral Equation Model (IEM) Inversion through an Artificial Neural Network (ANN) was considered a suitable technique at both frequencies and applied. Retrieval results at both frequencies are here presented against ground truth collected on the site concurrently with the airborne datasets. They show that, for the application considered, the retrieval from S-band data is more accurate than that from X-band with an average error smaller than 4%.

The singular characteristics of the Canarian archipelago (Spain) have allowed the development of a unique biological richness. Almost half of its territory is protected to preserve the natural environment. In this paper, different approaches to consider fusion of multi-sensor data are considered and corresponding methodologies described. The application to real datasets over Canarian islands is undergoing and fusion maps will be presented at the conference while preliminary classification results with multispectral data are described here.

This paper introduces a novel technique for ship-detection with Synthetic Aperture Radar (SAR) imagery based on the Generalized Likelihood Ratio Test (GLRT). Firstly, a suitable probability density function for a canonical ship is computed from the adoption of scattering models within the Geometrical Optic (GO) solution. Secondly, the GLRT is derived and the detector performance computed through Monte Carlo simulations. Finally, the GLRT technique is compared to the CFAR (Constant False Alarm Rate) algorithm in terms of ROC (Receiver Operating Characteristic) curves and computational load.

Multisensor data fusion is getting more importance with the increasing number of available satellite sensors. The aim of data fusion is to take advantages of combining different types of data to improve accuracies. However features extracted from different sensors will often have different statistical properties, and therefore combining data in an efficient way is not a trivial task. This paper proposes a new algorithm for data fusion between classification maps separately derived by application of clustering algorithms to PolSAR and Multi-spectral datasets. The expected new output is a map where all the classes identified with each single dataset will be present. The pixels assigned to the same class with both datasets will be characterized by a higher likelihood to belong to that class. The application for which the data fusion has been developed is that of hazardous areas reduction in land mines clearing operation. At this purpose a demonstration site has been set up in Poland within the FP7 D-BOX project and the fusion framework tested on it with acquisition of remote sensing imagery.

The retrieval of flooding levels with high-resolution (HR) synthetic aperture radar (SAR) images is presented in this paper. A new framework is proposed. It is based on the inversion of theoretical scattering models initially developed for nonflooded urban areas and here adapted to the flooding case. Starting from the theory, two possible retrieval approaches have been developed and are the main topic of this paper: two possible retrieval approaches have been developed and are the main topic of this paper: the local Single Image Objects Aware (SIObA) and the global Two Image Area Aware (TIArA). These two approaches are conceived to be applicable under different working conditions and consequently holding different properties and reliability. For each of them, a different algorithm is derived and tested, and the retrieval results are validated on a meaningful data set of HR TerraSAR-X images relevant to the Gloucestershire (U.K.) flooding that occurred in year 2007.

Some knowledge of sea state and conditions is input in ship detection algorithms based on inversion of scattering models for Synthetic Aperture Radar (SAR) images. This paper shows a novel technique for the estimation of roughness parameters of the sea surface from SAR images. The estimation procedure is based on the minimization of the absolute error between the Radar Cross Section (RCS) of the sea surface measured on the SAR image and the expected RCS computed using the Kirchhoff approach within the Geometrical Optics (GO) solution. The technique is tested on three different TerraSAR-X images acquired in November 2012 over the Portsmouth harbour in the UK.

This paper proposes an unsupervised clustering algorithm for multispectral images, which automatically determines the number of statistically distinct clusters in the image. It uses the multivariate student-t distribution as a more flexible underlying statistical model, with the Gaussian as only a special case. The algorithm shows better data modeling flexibility than the Gaussian case. Excellent and reproducible clustering results are observed for both simulated data and real data from Worldview-2 multispectral sensor.

Many multivariate statistical distributions have been derived using the well-known product model to stochastically model polSAR data. One important factor in their utilization is the estimation of their texture parameters. Recently, it has been shown that the method of matrix log cumulants (MoMLC) for multilook PolSAR statistical distributions results in estimators with low bias and variance properties. This method is becoming increasingly popular and can be regarded as state of the art. However, some distributions (e.g., G distribution) do not have closed-form MLC expressions, making the application of MoMLC a challenge. It is therefore desirable to have alternative parameter estimation methods. In this paper, we propose a new estimation method based on fractional moments of the multilook polarimetric whitening filter (MPWF). This results in estimators with mean square error that is even lower than the MoMLC-based estimators. In addition, the mathematical expressions of the estimators are computationally less complicated than MoMLC-based estimators. The proposed estimators can be easily derived for all commonly occurring multilook PolSAR distributions but have been only given for G, K, and G0distributions in this paper. Comparisons are made with other known estimators for these distributions using simulated and real PolSAR data. For real data, formal goodness-of-fit testing, which is based on MLCs, has been used to assess the fitting accuracy of G, K, and G0models using different estimators.

The K distribution can arguably be regarded as one of the most successful and widely used models for radar data. However, in the last two decades, we have seen tremendous growth in even more accurate modeling of radar statistics. In this regard, the relatively recent G0distribution has filled some deficiencies that were left unaccounted for by the K model. The G0model, in fact, resulted as a special case of a more general model, the G distribution, which also has the K model as its special form. Single-look and multilook complex polarimetric extensions of these models (and many others) have also been proposed in this prolific era. Unfortunately, statistical analysis using the polarimetric G distribution remained limited, primarily because of more complicated parameter estimation. In this paper, the authors have analyzed the G model for its parameter estimation using state-of-the-art univariate and matrix-variate Mellin-kind statistics (MKS). The outcome is a class of estimators based on the method of log cumulants and the method of matrix log cumulants. These estimators show superior performance characteristics for product model distributions such as the G model. Diverse regions in TerraSAR-X polarimetric synthetic aperture radar data have also been statistically analyzed using the G model with its new and old estimators. Formal goodness-of-fit testing, based on the MKS theory, has been used to assess the fitting accuracy between different estimators and also between the G, K, G0, and Kummer- U models.

Collection of ground truth to validate remote sensing classification and/or detection algorithms is rarely accounted for due to the inaccessibility of the sites or the elevated costs of such operations. In this paper some of the opportunities behind crowd-sourcing are explored through the description of a remote sensing project on water quality monitoring in Africa where the ground truth was collected involving and training people from local communities.

The paper shows a new algorithm for ship-detection from Synthetic Aperture Radar (SAR) images. The algorithm consists of three main stages: pre-processing, detection and discrimination. In the pre-processing a land mask is obtained considering the different statistics between the sea and the land's backscattered field; the detection stage isolates the bright points over the sea background employing a Constant False Alarm (CFAR) method; while the ships are retrieved, in the discrimination step, by evaluating the scattering contributions of the possible targets detected in the previous stage. The algorithm is tested on an airborne S-band SAR image of Portsmouth harbor, similar to those that will become available with the upcoming UK SAR mission NovaSAR-S.

This paper proposes a new method of estimating the shape parameters of polarimetric singlelook complex compound distributions which model synthetic aperture radar data. The estimators derived from this method utilize fractional moments of polarimetric whitening filter, and can be derived for all commonly occurring distributions. They also exhibit low variance properties.

Bright curvilinear features arising from the geometry of man-made structures are characteristic of synthetic aperture radar (SAR) images of urban areas, particularly due to double-reflection mechanisms. An approach to urban earthquake damage detection using double-reflection line amplitude change in single-look images has been established in previous literature. Based on this method, this paper introduces an automated tool for fast, unsupervised damage detection in urban areas. Ridge-based curvilinear features are extracted from a preevent SAR image, and double-reflection candidates are selected using prior probability distributions derived from a simple geometrical building model. The candidate features are then used with the ratio of a pair of single preevent and postevent SAR single-look amplitude images to estimate damage levels. The algorithm is very efficient, with overall computational complexity of O(Nlogk) for an N-pixel image containing features of mean length k. The technique is demonstrated using COSMO-SkyMed data covering L'Aquila, Italy, and Port-au-Prince, Haiti.

New Synthetic Aperture Radar (SAR) missions in S-band are currently under design but the potential performance of this microwave frequency is still under discussion. This paper presents the outcomes of a study on canopy classification carried out with fully polarimetric S- and X-band datasets contemporaneously acquired by the Astrium airborne SAR demonstrator. Classical polarimetric decompositions have been applied to investigate the S-band capabilities in vegetation monitoring and preliminary results are here presented.

A new form of G distribution, called the dual-texture G distribution, is proposed for Single-Look Complex (SLC) PolSAR data. Unlike the scalar texture productmodel, the dual texture G distribution is derived considering different texture variables for co-pol and cross-pol (x-pol) channels. The co-pol and x-pol texture variables are modelled by the Generalized Inverse Gaussian (GIG) distribution, separately. The result is a more flexible multivariate distribution. Also Mellin Kind Statistics (MKS) are utilized to analyze GIG textures, observe evidence of dual texture and to estimate shape parameters.

This paper deals with a new teaching experience carried on at the Department of Electronic Engineering at University of Surrey. The experience is based on the enhancement of students learning in the Satellite Remote Sensing class with the adoption of new technologies for assessment and teaching purposes. More precisely, podcasts have been introduced in a 3-year long project as supporting teaching material or coursework. Here the experience of the first year, where podcasts are introduced as group coursework, is described in detailed and commented.

A new concept for change detection algorithm for urban areas affected by earthquake is here presented. It is characterized as applicable to just one post-event amplitude Synthetic Aperture Radars (SAR) image and employs the inversion of sound scattering models already introduced in literature by the same authors. Aim of the algorithm is to try obtaining fast mapping of damaged areas and provide a first, even rough, evaluation of damage reported. In particular in this paper the overall block diagram chain and the algorithm rationale behind that framework are introduced and discussed in details. Some preliminary results are presented and the performance analyzed. New possible applications based on similar rationale are also commented.

There is no doubt that retrieving observed scene features is one of the most interesting and challenging activities in all fields of remote sensing: The successful extraction of scene parameters may not only mean the success of the adopted procedure but also the success of a prediction model, an image product, a sensor project, or even an entire mission. This paper is partly concerned with this. The mission, the sensor, and the products at issue are the TerraSAR-X; the feature retrieval approach is the deterministic model-based approach already tested on E-SAR images and now in phase of improvement and testing on high-resolution TerraSAR-X images. Together with assessing the performances of TerraSAR-X products, this paper deals with a new application which, until now, has not received enough attention even if being worth of it: monitoring of big tanks in suburban or urban areas. Detailed discussion concerning the most suitable product for this kind of application is accompanied by retrieval results carried out on recently acquired TerraSAR-X images.

Detection of man-made structures in urban areas, in terms of both geometric and electromagnetic features, from a single, possibly high resolution (HR), synthetic aperture radar (SAR) image is a highly interesting open challenge. Within this framework, a possible approach for the extraction of some relevant parameters, describing the shape and materials of a generic building, is proposed here. The approach is based on sound electromagnetic models for the radar returns of each element of the urban scene. A fully analytical representation of electromagnetic returns from the scene constituents to an active microwave sensor is employed. Some possible applications of feature extractions from real SAR images, based on the aforementioned approach, have already been presented in the literature as first examples of potentiality of a model-based approach, but here, the overall theory is analyzed and discussed in depth, to move to general considerations about its soundness and applicability, and the efficiency of further applications may be derived. For the sake of conciseness, although the proposed approach is general and can be applied for the retrieval of different scene parameters (in principle, anyone contributing to the radar return), we focus here on the extraction of the building height, and we assume that the other parameters are eithera prioriknown (e.g., electromagnetic properties of the materials) or have been previously retrieved from the same SAR image (e.g., building length and width). An analysis of the sensitiveness of the height retrieval to both model inaccuracies and errors on the knowledge of the other parameters is performed. Some simulation examples accompany and validate the solution scheme that we propose.

Cylindrical-shape tanks are typical of any suburban area and often contain dangerous gases or fluids. In this paper, we suggest a way to monitor them by means of high resolution Synthetic Aperture Radar (SAR) images and a scattering model able to quantitatively consider how the radar signal interacts with this kind of structures and how they appear m the SAR images. Adopting the model, geometrical information as the tank height is retrieved from the SAR images m a non-conventional way that is exploiting the information content contained m the double reflection contribution to the radar cross section. Results are compared with more traditional methods and discussed.

Analysis, interpretation and feature extraction concerning High Resolution (HR) Synthetic Aperture Radar (SAR) images of urban areas urgently require support of sound and appropriate electromagnetic modelling. The modelling takes into consideration the radar geometry and the (geometric and electromagnetic) scene parameters but also the novelty brought by high resolution. In this paper, this way of developing suitable electromagnetic modelling for HR SAR images of urban areas is shown to be successful as able to interpret and retrieve, from these scenarios, new and interesting details that will certainly represent the main actor of next generation of applications for urban areas with SAR sensors.

Interpretation of high resolution SAR (synthetic aperture radar) images is still a hard task, especially when man-made objects crowd the scene under detection. This paper contributes to the analysis of this kind of data by adopting an approach, based on a scattering model, for the retrieval of buildings height from real SAR images and presenting first numerical results.

Recently, a new operating mode for Synthetic Aperture Radar (SAR) system, referred to as hybrid stripmap/spotlight mode, has been presented. In the hybrid acquisition mode the radar antenna beam is steered about a point farther away from the radar than the area being illuminated, thus generating microwave images with an azimuth resolution better than that achieved in the stripmap configuration, and a ground coverage better than the one of the spotlight configuration. The subject of design, processing and data interpretation for the hybrid SAR mode is gaining an increasing interest in the remote sensing scientific community. Consequently, a hybrid SAR raw signal simulator is strongly required, especially when real raw data are not available yet, to test processing algorithms and to help mission planning. In addition, to analyse the effects of processing errors and to verify the impact of different system design choices on the final image for different kinds of imaged scenes, an extended scene SAR raw signal simulator is very useful: it is what we present in This work. After showing that in this case a 2D Fourier domain approach is not viable, we demonstrate that a 1D range Fourier domain approach, followed by 1D azimuth time domain integration, is possible when some approximations, usually valid in the actual cases, are accepted.

The hybrid stripmap/spotlight mode for a synthetic aperture radar (SAR) system is able to generate microwave images with an azimuth resolution better than the one achieved in the stripmap mode and a ground coverage better than the one of the spotlight mode. In this paper, time- and frequency-domain-based procedures to simulate the raw signal in the hybrid stripmap/spotlight mode are presented and compared. We show that a two-dimensional Fourier domain approach, although highly desirable for its efficiency, is not viable. Accordingly, we propose a one-dimensional (1-D) range Fourier domain approach, followed by 1-D azimuth time-domain integration. This method is much more efficient than the time-domain one, so that extended scenes can be considered. In addition, it involves approximations usually acceptable in actual cases. Effectiveness of the simulation scheme is assessed by using numerical examples.