The thesis describes and tests a method to combine satellite measurements of the geoid height with shipboard and terrestrial measurements of the freeair gravity anomally using Fourier transform techniques. Fast Fourier transforms can process very large data sets quickly but require the data to be available on a complete and regular grid. New methods were devised that grid altimeter and marine gravity data, and which remove the long-wavelength errors associated with these measurements. Satellite altimeter data distributed along widely spaced groundtracks are differenced along track to reduce long-wavelength orbit errors, and any repeat track observations are averaged to reduce time-dependent oceanographic noise. The ascending and descending along-track slopes are then gridded separately, using a kernel designed to extrapolate the observations smoothly into the gaps between tracks, while retaining the high along-track resolution of the data. The spectral properties of the gridding operation, were analysed by calculation of a transfer function. In a development of Sandwell's method, the grids of along-track slopes are transformed using vector algebra to grid-easting and -northing deflections of the vertical. Vertical deflection grids from any number of satellite missions may now be combined, then converted using Fourier techniques to give a grid of the geoid height. This grid is easily converted to a grid of the freeair anomaly. The effect of datum errors in the shipboard gravity data is also reduced by processing not gravity but along-track gravity gradient. An incomplete grid of freeair gravity anomalies is generated from the observed along-track gravity slopes, constrained by, and improving the altimeter-derived freeair anomaly grid. Land gravity measurements are interpolated directly onto a separate grid as there is no similar long-wavelength error in the data. The combination algorithm performs a weighted superposition of the geoid height and freeair anomaly grids in an iterative process, until self-consistent. The algorithms were first tested on a simulated dataset, extracted from accurate, high-resolution and self-consistent geoid and freeair anomaly models of the British Isles and North Sea. The geometry of real altimeter tracks from an area of equal size in the North Atlantic were superimposed on the British geoid, and simulated readings interpolated.

This work aims to show the value and limitations of aerial photographs, Landsat and side-looking airborne radar (SLAB) images for forest-type recognition, tree-species identification, timber-volume estimation and area estimation of the Southeast Asian, the Amazonian and the African tropical forests. The studies comprise an investigation by a review of literature on the applications of the three remote sensing systems in the three tropical forest regions and an experimental investigation of the tropical African forests using Southwestern Nigeria as a typical example. The findings on the technical capabilities of the three systems and the economics of their use were interrelated for the evaluation. The investigations show that remote sensing techniques are easier, quicker and cheaper than ground techniques for tropical forest surveys. Only aerial photography can be successfully used for species identification and volume estimation. Despite the similar cost of SLAR imagery and the cheaper cost of Landsat imagery, they cannot compete successfully with aerial photography for forest-type recognition and area estimation. In combined use, they none-the-less add extra value to photo-information. The lack of funds and the unfavourable weather conditions as well as the complex forest structure and floristic composition are the main factors constraining the use of remote sensing in tropical forest surveys. Options for an economic use of remote sensing are recommended and suggestions for further investigation are made.

This thesis proposes and demonstrates that a low-cost commercial DSP platform can be used to perform real-time motion estimation for the Dervish landmine-clearance vehicle based on digital imaging of the ground. The Dervish landmine-clearance vehicle provides specific <i>application</i> requirements that are first examined in detail. The motion estimation <i>algorithm</i> requirements are then examined and, based on the tracking accuracy and theoretical computational performance, correlation-based feature tracking is selected. The individual parameters, including block size, motion model and interpolation technique, are carefully examined and suitable values selected. Finally, the <i>hardware</i> requirements are examined, a DSP platform chose and a trial implementation reported. A difficult problem within correlation-based motion estimation is how to update the reference block. This thesis describes novel methods for reference block updating. The reference block update strategies are examined in detail for parameter selection, tracking accuracy and computational performance. Particular attention is given to the tracking accuracy under increasing levels of image noise. Using the <i>application, algorithm</i> and <i>hardware</i> requirements and decisions, a system is proposed that meets the main requirements of the Dervish landmine-clearance vehicle. The issues of robustness within long-term tracking and a novel dissimilarity measure based on the reference block update strategies are also presented. Parts of the research presented in this thesis have also been reported in conference and journal papers, which are listed at the end of the document.

No description available.

Soil is an important non-renewable source. Its protection and allocation is critical to sustainable development goals. Urban development presents an important drive of soil loss due to sealing over by buildings, pavements and transport infrastructure. Monitoring sealed soil surfaces in urban environments is gaining increasing interest not only for scientific research studies but also for local planning and national authorities. The aim of this research was to investigate the extent to which automated classification methods can detect soil sealing in UK urban environments, by remote sensing. The objectives include development of object-based classification methods, using two types of earth observation data, and evaluation by comparison with manual aerial photo interpretation techniques. Four sample areas within the city of Cambridge were used for the development of an object-based classification model. The acquired data was a true-colour aerial photography (0.125 m resolution) and a QuickBird satellite imagery (2.8 multi-spectral resolution). The classification scheme included the following land cover classes: sealed surfaces, vegetated surfaces, trees, bare soil and rail tracks. Shadowed areas were also identified as an initial class and attempts were made to reclassify them into the actual land cover type. The accuracy of the thematic maps was determined by comparison with polygons derived from manual air-photo interpretation; the average overall accuracy was 84%. The creation of simple binary maps of sealed vs. vegetated surfaces resulted in a statistically significant accuracy increase to 92%. The integration of ancillary data (OS MasterMap) into the object-based model did not improve the performance of the model (overall accuracy of 91%). The use of satellite data in the object-based model gave an overall accuracy of 80%, a 7% decrease compared to the aerial photography. Future investigation will explore whether the integration of elevation data will aid to discriminate features such as trees from other vegetation types. The use of colour infrared aerial photography should also be tested. Finally, the application of the object- based classification model into a different study area would test its transferability.

This thesis investigates the use of a novel optical fibre sensor (OFS) for structural monitoring in remote or inaccessible areas. OFSs are desirable for this application because they are intrinsically safe, add little weight and are free from electrical interference. Fibre Bragg gratings (FBGs) in particular are attractive since they act as optical strain gauges, converting fibre strain to a wavelength-encoded signal. This project uses FBGs written into three or four cores of a novel four-core fibre at the same point along the fibre length. This configuration allows the local curv~ture of the fibre to be obtained by measuring the strain difference between cores ·at an applied bend. The multicore fibre (MCF) eliminates any temperature sensitivity and strain transfer issues ofthe sensor. Dynamic curvature measurement was achieved by interrogating each of the MCF cores with an arrayed waveguide grating (AWG). This was demonstrated at interrogation speeds >11 kHz for a stainless steel cantilever vibrating at -30 Hz to achieve a curvature resolution of 0.09 m-I. Quasi-static tests found a curvature resolution of 0.02 m-I. This dynamic curvature measurement technique was applied to create an accelerometer from a cantilever formed from a short length ofMCF vibrating at frequencies up to 3 kHz. The accuracy of the acceleration measurement was better than 5 % at frequencies below 300 Hz. A commercial interrogator based on a tunable laser technique was used for several quasi-static applications. The curvature of a 240.8 mm diameter aluminium and Perspex cylinder was found to be resolved by the MCF sensor to 0.01 m-I. Four MCF FBGs were spliced together to form a multiplexed array in order to investigate shape measurement using several curvature measurements. The array was attached to a 33 cm long deformable Perspex rod and the MCF FBGs were found to measure curvatures that matched very closely with values predicted by a mathematical model. This was demonstrated for both horizontal and vertical deformations. Finally, an MCF FBG sensor was embedded in a short strip of compliant material (Sylgard) in order to create a sensor that could easily be wrapped around a small test object without the need for permanent bonding. This was used to measure the change in radius of a 2 cm diameter sample as it was compressed. This was found to be capable of measuring a radius change of< 30 Jlm.

Technological progress in recent decades has led to sensor networks and robotic explorers becoming principal tools for investigation of remote or "hostile" environments where it is difficult, if not impossible for humans to intervene. These situations include deep ocean and space environments where the devices can be subject to extreme pressures, temperatures and radiation levels. It is a costly enterprise to deploy an instrument in such settings and therefore reliable operation and ease of use are requisite features to build into the basic fabric of the machine. This thesis describes the design and implementation of a modular machine system based on a peer-to-peer, decentralised network topology where the power supply and electronic hardware resources are distributed homogeneously throughout a network of nodes. Embedded within each node is a minimal, low-power single board computer on which a real-time operating system and MicroCANopen protocol stack are operating to realise a standard interface to the network. The network is based on a grid paradigm where nodes act as resource producers and consumers, sharing information so that the machine system as a whole can perform tasks. The resulting architecture supports "plug-and-play" flexibility, to allow users or system developers to reconfigure or expand its capabilities by adding/removing nodes at a later time. An immediate application of this instrument is in-situ sampling of microbes in extreme aqueous habitats. The microbial sampler is targeted at providing improved sampling capabilities when performing physical, chemical and biological investigations in deep- ocean hydrothermal vent environments. At these depths the instrument is subject to immense pressures of many thousand pounds per square inch, where superheated, corrosive, mineral-loaded vent fluids mix with near-freezing seawater. In the longer term, it is anticipated that this flexible, open interface architecture on which the microbial sampler instrument is based will be applicable more generally to other sectors, including commercial and scientific markets.

The aim of this research project, towards the automation of the Hybrid Pipeline Robot (HPR), is the development of a control architecture and strategy, based on reconfiguration of the control strategy for speed-controlled pipeline operations and self-recovering action, while performing energy and time management. The HPR is a turbine powered pipeline device where the flow energy is converted to mechanical energy for traction of the crawler vehicle. Thus, the device is flow dependent, compromising the autonomy, and the range of tasks it can perform. The control strategy proposes pipeline operations supervised by a speed control, while optimizing the energy, solved as a multi-objective optimization problem. The states of robot cruising and self recovering, are controlled by solving a neuro-dynamic programming algorithm for energy and time optimization, The robust operation of the robot includes a self-recovering state either after completion of the mission, or as a result of failures leading to the loss of the robot inside the pipeline, and to guaranteeing the HPR autonomy and operations even under adverse pipeline conditions Two of the proposed models, system identification and tracking system, based on Artificial Neural Networks, have been simulated with trial data. Despite the satisfactory results, it is necessary to measure a full set of robot’s parameters for simulating the complete control strategy. To solve the problem, an instrumentation system, consisting on a set of probes and a signal conditioning board, was designed and developed, customized for the HPR’s mechanical and environmental constraints. As a result, the contribution of this research project to the Hybrid Pipeline Robot is to add the capabilities of energy management, for improving the vehicle autonomy, increasing the distances the device can travel inside the pipelines; the speed control for broadening the range of operations; and the self-recovery capability for improving the reliability of the device in pipeline operations, lowering the risk of potential loss of the robot inside the pipeline, causing the degradation of pipeline performance. All that means the pipeline robot can target new market sectors that before were prohibitive.

Interferometric synthetic aperture radar (InSAR) has been increasingly used to extract information about the earth‟s surface by exploiting the phase difference between two complex radar signals. Some significant application fields that utilize InSAR techniques are digital elevation model (DEM) generation, land use classification and land subsidence. In this thesis, by using ERS-1/2 tandem SAR images pairs, the potential implementations of SAR interferometry in tropical peatland forests in Central Kalimantan, Indonesia are described. Coherence was found to be a good tool for rapid assessment for burned and deforested areas. The coherence of burned forest area was increased by 0.2; whilst the minimum coherence was found to be than 0.35. However, many critical factors affect the quality of InSAR data and limit its applications, such as methods of InSAR data processing. This study emphasizes the impact of different processing and phase unwrapping techniques on DEM accuracy. Analyses of InSAR DEM accuracy indicate that DEMs with relative errors of less than 3 m root mean square error (RMSE) are possible in some regions in the former Mega Rice Project (Ex-MRP) area and could meet many objectives of a global mapping mission. Applying adaptive filtering many times with a decreasing window size has a strong impact to reduce the number of residues, which can increase the phase unwrapping efficiency and the final DEM accuracy. Furthermore, the differential SAR Interferometry (DInSAR) was examined to see if it can detect peatland subsidence accurately from October 1997 to January 2000 using 4-pass and complex interferogram combination methods. The subsidence rate of 2 cm per year is considered to be the best possible prediction for subsidence in the project area and between 53 and 83 Mt of peat carbon was lost for the same period of the study. The contribution of the maximum subsidence to the emission of CO2 was estimated to be 52 tonnes per hectare per year. These results are not reliable enough for detailed planning purposes, but they provide a basis for further work by highlighting where methodological development is needed.

A technique to model the spatial distribution of the 'reliability' of satellite-derived land cover maps based on the spectral distance information derived during the classification process has been described. The utility of the resulting 'reliability map' for highlighting incorrectly classified areas on land cover maps has been demonstrated by field verification. A spectral separability analysis showed that Lansat TM data did not allow for the spectral separation of all vegetation types. However, the vegetation types of each of the three geo-ecological formations 'terra firme', 'whitewater floodplains', and 'blackwater floodplains' could be separated spectrally. Therefore, the boundary between these formations was visually interpreted from ERS-1 SAR and Landsat TM data. Each of the resulting image regions was subsequently classified with the corresponding subset of classes. This enabled the mapping of the project region at a high level of thematic resolution while avoiding the misclassification of pixels between spectrally similar vegetation types. In agreement with the results reported by other researchers, the utility of ERS-1 SAR data for general vegetation mapping was found to be rather limited. However, ERS-1 SAR data appeared to be suitable for the updating of existing vegetation maps for further forest clearing. The data were also found useful for the visual interpretation of the floodplain boundary from geomorphological features.