Curtin University of Technology
Skip to content
Link to Curtin homepage
Curtin Western Australian School of Mines

RESEARCH IN SPATIAL SCIENCES

Contact: Prof Will Featherstone W.Featherstone@curtin.edu.au

The major areas of research in the Western Australian Centre for Geodesy (based in the Department of Spatial Sciences and also part of The Institute for Geoscience Research) include:

The major areas in Geographic Information Science and Remote Sensing research within the Department of Spatial Sciences include:

Western Australian Centre for Geodesy
(part of The Institute for Geoscience Research)

Geoid and Gravity Field Determination

We have been active in Australian and international geoid determination (the equipotential surface of the Earth’s gravity field) for over 15 years, funded throughout by numerous Australian Research Council grants. Our research includes a wide range of theoretical and methodological issues, coupled with software development. In 1998, we provided the methods and software to produce the AUSGeoid98 quasigeoid model, which remains the national standard for the transformation of Global Positioning System (GPS) heights to the Australian Height Datum. A technical description of AUSGeoid98 is given at http://www.cage.curtin.edu.au/~will/10750313.pdf and the model can be downloaded free of charge from http://www.ga.gov.au/geodesy/ausgeoid/. We are currently producing a new Australian quasigeoid model that will be released in 2008.

Cryospheric Mass Transport in the Earth System

We are analysing the impact of changes in the mass balance of the cryosphere (the portions of the Earth’s surface where water is in solid form) on the whole Earth system. Different procedures and software have been developed to model changes in the Earth’s gravity field, centre of mass, rotation and global sea level change based on changes in ice cover. We have shown that the melt of ice-covered areas does not lead to a globally uniform sea level change, but it varies depending on location, which is important when analysing the impact of local, regional and global sea level change (e.g., extents of flooding). GRACE (Gravity Recovery and Climate Experiment) satellite mission data have been analysed and used to assist in both the determination of recent mass changes in the cryosphere and corresponding changes in the Earth’s gravity field and regional sea level.

Spatial variations in Sea Level

Different data sources such as satellite radar altimetry, tide-gauge records and the GRACE (Gravity Recovery and Climate Experiment) satellite mission are being used to monitor and analyse the most dominant spatial and temporal variations in global sea level and the Earth’s gravity field. In order to do this, different analytical and statistical modelling techniques, such as harmonic analysis and principal component analysis, have been employed. This includes the identification and quantification of spatial and/or temporal biases in global sea level change estimates caused by insufficient sampling of the sea level change signal. Furthermore, recent GRACE-derived gravity changes are being used to derive mass changes in the cryosphere (the portions of the Earth’s surface where water is in solid form) and all major river drainage basins.

Forward Gravity Field Modelling

We have developed two comprehensive software packages to estimate the gravitational effects of global geo-referenced mass distributions (e.g., topographic or crustal masses) using space- and spectral-domain techniques. The latter software provides approximate results, but is well suited to modelling global mass distributions in a time-efficient way. The former software directly evaluates Newton’s integral in the space domain using discretised numerical integration based on the superposition of the gravitational effect of regular shaped bodies (e.g., rectangular or spherical prisms). These approaches can be applied to large datasets on a standard PC. The software has recently been applied to compute spherical terrain corrections over Australia at a 9-arc-second (200m) spatial resolution.

Synthetic Earth Gravity Models

Two different synthetic/simulated Earth gravity models (SEGMs) have been developed so far at Curtin: a global gravity model (CurtinSEGM), and a regional model over Australia only (AusSEGM). They can be downloaded free of change from http://www.cage.curtin.edu.au/~kuhnm/CurtinSEGM/. Both synthetic models provide a realistic representation of the Earth’s gravity field, so are well suited to validating, and thus possibly improving, the procedures and software currently in place for gravity field determination and modelling. AusSEGM is now being used to test geoid determination techniques in Australia and Canada. A new global forward gravity model is under construction, which will allow modelling of the gravity field inside the topographic masses, and will be used to validate different height systems.

 

Back to top

 

Flexural Isostasy

Using the wavelet transform, we have developed a new method to compute the elastic strength of the Earth’s lithosphere (the outermost shell of the solid Earth), and directional variations in this elastic strength. The new method involves a correlation between gravity and topography data, and inversion against (an)isotropic loading models. The resulting maps of elastic thickness show a very good correspondence with seismic and thermal data, indicating that the strength of old, cold cratonic areas (stable parts of the lithosphere) is very high, with mass-loads here largely supported by the mechanical rigidity of the upper mantle. The anisotropic studies suggest that both seismic and strength anisotropy arise from the same source, most likely being the preferred orientation of olivine crystals in the upper mantle.

Variations in Australia’s Stored Water

Both Australia’s water resources and environment are now at serious risk from drought. Principally, this has led to a decline in accessible water, which not only threatens livelihood, but also impacts negatively on the environment. Urgent monitoring measures are therefore essential for enhancing sustainable water use, protection and conservation in Australia. The GRACE (Gravity Recovery And Climate Experiment) satellites are being applied to monitoring variations in Australia’s water resources to support decision making by managers, State/Territory agencies, farmers, politicians, etc. This will benefit water and environmental management, and contribute techniques for the efficient space-based monitoring of Australia’s scarce water resources.

Assessment and Unification of Heights

For several years, we have investigated the integrity of the Australian Height Datum, which forms the fundamental vertical spatial data infrastructure in Australia. Current work is redefining this height datum using more sophisticated modelling, processing and adjustment strategies. We have also devised a rigorous orthometric height system that can be embedded in this new vertical datum. In collaboration with Land Information New Zealand, we have implemented a geoid-based vertical datum in New Zealand that also unifies the 13 separate vertical datums in use there. We ultimately aim to make vertical datums fully compatible with regional geoid models, thus allowing direct height determination from GPS.

Least-squares Collocation

Least-squares collocation is an optimal interpolation and prediction tool for gravity field modelling, somewhat akin to Kriging in geostatistics. We have used least-squares collocation for merging of ship-track and satellite altimeter gravity data, which was used in AUSGeoid98, the national standard geoid model. We also devised and implemented a cross-validation technique that gives a more objective indication of the correlation length and noise of the covariance function when merging GPS-levelling and gravimetric geoid data. We are currently implementing anisotropy and non-stationarity into standard least-squares collocation, which will ultimately enable it to optimally interpolate and predict spatially variable gravity field data.

Ellipsoidal Physical Geodesy

For over 300 years, geodesists have known that the Earth’s figure is more like an oblate ellipsoid of revolution than a sphere. However, physical geodesists often still use spherical Earth models and approximations. Therefore, we are examining - from first principles - the complete treatment of the Earth’s gravity field in a purely ellipsoidal framework, ranging from satellite-based gravity field determination to regional geoid modelling. This will lead to a general theory of ellipsoidal physical geodesy, which will improve the accuracy of gravity field and geoid modelling and thus allow us to fully profit from data observed by the new dedicated satellite gravity missions. Recent work in this direction has generalised the Meissl spectral scheme for the geodetic ellipsoid.

Back to top

South West Seismic Zone

The southwest seismic zone (SWSZ) in Western Australia is one of few intra-plate tectonic regions in the world where earthquake activity is not associated with the plate boundaries.�� Importantly, the proximity of the SWSZ zone to Perth presents a significant seismic hazard. However, the magnitude, type and controls deformation of the Earth’s crust in the SWSZ are not fully understood at present. We have therefore used palaeosiesmology (the study of ancient earthquakes) to create a longer time-series of large earthquakes, which will lead to improved risk-mapping techniques in the SWSZ. We have also used a combination of GPS-geodetic and geophysical measurements to place additional controls on the current seismicity in the SWSZ, also with a view to improved risk-mapping.

Environmental Geodesy

Australia, as part of the global community, is ominously poised in a period of significant environmental change caused by warming of the Earth’s atmosphere. Changing sea-levels and variation in national water storage present significant challenges that are at the forefront of Australian consciousness. This research programme realises the first assembly of national geodetic intellect (Australian National University, University of Tasmania, University of New South Wales and Curtin) to tackle these complex problems through the development and extension of space-geodetic observational techniques, and drawing upon recent and significant cash injections into geospatial infrastructure in Australia. It will provide the first-ever comprehensive indication of the contemporary state of changes in sea-level, Antarctic ice cover and broad-scale water storage in the Australian context.

Long-range GPS Navigation

In close collaboration with AAMHatch Pty. Ltd., we have developed, implemented and tested long-range airborne Global Positioning System (GPS) using a sparse network of ground-based receivers, resulting in the SkyControl system, which allows for accurate coordination of airborne mapping sensors over long ranges (>200 km). In parallel, we developed Australia’s first precise point positioning (PPP) software suite, which can give centimetre-level positions with a single GPS receiver. As well as reducing the cost of airborne mapping surveys in Australia, the SkyControl system has the added benefit of increased accuracy and reliability. This system is now in routine operational use by AAMHatch Pty. Ltd.

Optimal CORS GNSS site selection

The National Collaborative Research Infrastructure Strategy (NCRIS), through AuScope Geospatial, will establish over 100 CORS (continuously operating GNSS - global navigation satellite system - reference stations) across Australia. Approximately one third of the $65 million AuScope budget is allocated to install and maintain a CORS network. Through the Cooperative Research Centre for Spatial Information, we are determining the optimal way to select the best location and distribution of CORS sites to support as-many-as-possible scientific and commercial users of the new geodetic network.

Automated Analysis of Terrestrial Laser Scanner Data

Road asset management hinges upon an accurate inventory of the spatial location of all assets within the road corridor. We have produced algorithms embedded in software for automated processing of terrestrial laser scanner (TLS) data of road corridors to support applications such as asset management, engineering design and precise, as-built measurement of road surface shape, etc. The new software automatically extracts features (e.g., kerb lines, power poles, signs, etc) from the TLS point-cloud data. It fills an industry-identified void of automated processing in commercially-available packages, which previously required significant user interaction.

 

Back to top

 

On-site Self-calibration for Terrestrial Laser Scanners

Terrestrial laser scanners (TLSs) provide a 3D sampled representation (i.e., point-cloud) of any object scanned. TLSs have great potential to improve the measurement and representation of remote and large objects in applications such as engineering metrology, cultural heritage recording, forestry, etc ��Prior to performing measurement tasks such as these, however, proper error estimation and modelling is essential in order to remove the inherent systematic effects in TLSs. We have shown that a rigorous, point-based self-calibration method is effective, but is very labour-intensive. As such, we have developed a new planar-feature-based “on-site” self-calibration method that can reduce the amount of manual labour needed in the point-based method.

Metric Performance Evaluation of 3D Range Cameras

Modern 3D range cameras measure point-clouds and reflection intensity information from objects using time-of-flight methods with a CCD (charged coupled device) array. Their emerging applications are security, surveillance, biomechanics, etc. Since they comprise a CCD, a nearest-neighbour search for individual points is not necessary, which increases the efficiency in estimating the geometric properties of objects. This led us to broaden the application of 3D range cameras to real-time and dynamic applications such as mobile mapping and vehicle navigation. However, we have identified several instrumental factors that have degraded their metric performance. We are thus working towards developing an automated calibration method for 3D cameras, which will significantly improve their measurement precision and accuracy.

Classification and Feature Extraction from Point Clouds

We are alleviating the current terrestrial laser scanner (TLS) data processing bottleneck by automating the extraction of low-level features. A new localised method is used to classify edge, surface and boundary points based on various derived metrics. These metrics are based on common curvature approximation, but are also extended to principle curvature directions, radius of curvature, surface cohesiveness and local surface normal direction, all being derived from first-order principal component analysis. Several techniques have been developed to refine the classification results, as well as remove errors for the metrics caused by overlapping information sources, erroneous points and noisy data.

Mathematical Geodesy

Most of the proper solutions to inverse problems in geodesy are of a nonlinear nature. However, most of the strategies currently used are linear, requiring mathematical linearisation followed by iterative solutions to implement them in practice, thus resulting in approximate or inexact results. We have developed nonlinear algebraic solution approaches to provide complete and exact nonlinear solutions to geodetic problems, such as coordinate transformations, mapping, GPS (Global Positioning System) positioning and intersection. Our current research is now focussing on further improvements to these methods, as well as developing new ones.

GPS Meteorology

Traditionally, GPS (Global Positioning System) has been used mostly for positioning and navigation. However, the last decade has seen GPS turned into meteorological and environmental monitoring tool in what is commonly called GPS meteorology. Our current research involves the analysis of CHAMP (Challenging Mini-satellite Payload) and GRACE (Gravity Recovery And Climate Experiment) satellites’ application to the quality of the estimated water vapour, a greenhouse gas, using space-borne occultation analysis of the tropopause (the boundary between the troposphere and stratosphere) to assess regional atmospheric warming.

Back to top


Geographic Information Science and Remote Sensing

Mapping of Invasive Weeds

Invasive weeds cost Australian agriculture millions of dollars each year in lost production and, in severe cases, lost livestock and contaminated products. Under an Australian Research Council grant, we are examining ways in which remotely sensed imagery and GIS (geographical information systems) can be harnessed to combat invasive weeds in Australia. This work is currently looking at two case studies: one involves the mapping of Mesquite in pastoral areas in the Pilbara region, whilst the other is concerned with Paterson’s Curse in the agricultural areas in the southwest of Western Australia.

Pastoral Lease Assessment by Geospatial Analysis

Funded by a recent Australian Research Council grant, we will investigate ways in which the methods of pastoral lease assessment, which is currently carried out in Western Australia by on-ground inspections every five years, may be improved and enhanced by the incorporation of remotely sensed data and advanced geospatial analysis techniques. This work is in close collaboration with the Western Australian Department of Agriculture and Food, Specterra Pty. Ltd. and Curtin’s Centre for Management of Arid Environments. This will lead to cost savings in primary production in Western Australia.

Error Propagation in GIS

Any spatial analysis that combines various data layers in a geographical information system (GIS) will propagate - and thus be affected by - errors in those data layers, but a key risk is magnifying these errors, thus seriously and adversely affecting the analysis. As part of our broader research in this area, a project funded by the Cooperative Research Centre for Spatial Information is currently investigating error propagation through a number of raster data processes used in agricultural and environmental modelling. Ultimately, this whole programme will lead to more informed decision-making from GISs.

Agent-Based Frameworks for Intelligent Geocoding

Geocoding is essential for translating a physical address (e.g., a house, business or landmark) into geo-referenced coordinates that can be used in applications such as health, business, emergency services, etc. However, new techniques are needed to deal with existing issues and to cater for increased expectations of future geocoding users. In response, we are designing and implementing an agent-based framework to oversee the geocoding process. The agent-based paradigm is used together with a knowledge-based system to provide the intelligence and knowledge required to contextualise and enhance geocoding, with a view to it becoming fully automated.

Spatio-temporal Assessment of Virus Host and Vector Dynamics

Current climate change is altering the dynamics of the host and vectors of a number of viral diseases that threaten both Australian agriculture and the human population. This major research project, funded by the Australian Biosecurity Cooperative Research Centre, is investigating methods of improving the spatiotemporal mapping of the host and vectors that are competent for the transmission of arboviruses. This will lead to more informed management of the propagation of arboviruses that are significantly threatening in the Australian context.

Archaeological Predictive Modelling

In areas where modern development is taking place, care must be taken to minimise disturbance to the cultural heritage. This work is investigating methods of predicting the areas in the landscape that are most likely to contain sites of cultural significance. It is supported by two other projects: one on palaeo-environmental modelling, and the other on mapping ancient cultural landscapes (described next).

Back to top

Mapping Ancient Cultural Landscapes

The earliest inhabitants of the Australian continent may have made many of their choices of favoured areas on the basis of environmental suitability, but they also wove into the physical landscape a rich cultural landscape that has a material expression. This project is searching for patterns within that cultural landscape that can lead to a deeper understanding of the early occupation of the continent.

Palaeo-environmental Modelling

Archaeological predictive modelling relies to a great extent on the data layers that describe the environment in which previous land-use would have occurred. Therefore, we are investigating the optimal ways in which the palaeo-environment, particularly of South Western Australia, can be modelled at a sufficiently high spatiotemporal resolution to improve the accuracy and validity of archaeological models that are based on it.

Optimising the Usability of Location Based Services

Location-based services (LBS) are a vastly growing field, with the purpose of providing easily-accessible location-specific information to the general public. However, its full potential is not yet fully realised, partially because the current level of usability of most spatial systems is not sufficiently well adapted to lay-users. We are therefore investigating how spatial systems can be made more intelligent and learn from the user’s context in order to facilitate the effective interaction between the end-users and LBS-systems. The principles of user modelling and human-computer interaction are now being integrated into an AJAX-based spatial web portal.

Artificial Neural Networks for Surface Soil Moisture Retrieval

An artificial neural network (ANN) model is being designed, implemented and evaluated regarding its performance for soil-surface moisture retrieval. The capabilities of this ANN model are being verified utilising satellite imagery of lower resolution relative to aircraft data for estimating surface soil moisture. The significance of this research is being able to capitalise on future satellite remote sensing missions, such as the European Soil Moisture and Ocean Salinity (SMOS) satellite scheduled for launch in 2008. The data obtained from this satellite will be implemented in this high-performance ANN model.

Modelling Spatial and Temporal Movement of Tourists

Tourism is one of the rapidly developing industries in the world. The study of spatiotemporal movement models of tourists is currently undertaken in a variety of disciplines such as tourism, geography, mathematics, economics and artificial intelligence. However, firm knowledge from these different fields has been difficult to integrate properly because tourist-movement research has been conducted at different spatial and temporal scales. This project will therefore establish a methodology for rigorous modelling the spatiotemporal movement of tourists.

 

Back to top

 

Landmarks in Wayfinding Processes Based on Location-based Services

A landmark is a salient object that is used as a reference to help people memorise and recognise routes, and locate themselves in terms of their ultimate destination. However, there are difficulties in communicating landmarks between virtual and physical environments. They are caused by representation of landmarks in the virtual environment, prior familiarity with the physical environment, spatial and social ability, etc. We are now exploring the roles and usage of landmarks in the wayfinding processes based on location-based services. This will give wayfinding systems/devices that can effectively assist people in communicating between the virtual and physical environments.

Geographic Semantic Data Retrieval

Major contemporary data mining issues in the geographic domain, coupled with current methods used to manage spatial information, now result in exceedingly large datasets. Therefore, our current research involves an initial proof-of-concept that information-retrieval from large datasets can be improved using an ontology (conception of reality) to manage aspects of the retrieval process from query expansion to probabilistic results ranking. Whilst a dedicated ontology does not exist for the geographic domain, terms and relations in the WordNet lexical ontology are being exploited and tested on a small subset of the gazetteer of Australia for Perth.

Improved Crop and Pasture Management

We aim to deliver an operational, near-real time, easy-to-access, cost-effective farm package for pastures and crops, containing information on growth rate, crop yield forecast, biomass and pasture quality that can be used by primary producers to make better tactical and strategic decisions. At the paddock level, we are developing a capacity to estimate pasture quality at monthly intervals in the austral winter. Another objective is to incorporate high temporal resolution remote sensing to the existing Stress Index Model (STIN) developed in Western Australia to forecast yield at the shire level. Such integration will enable a better ‘spatialisation’ of crop growth status and yield forecasting in a near-real time.

Participatory GIS for Rural Community Development

We are investigating the processes for a rural community to learn modern business practices via spatial information and ICT (information and communication technology). This involves soliciting input from government and the private sector, mobilising a rural community to prepare, and building teams that will be engaged in the process. The premise is that knowledge is required to enable individuals and groups in rural communities to bridge the gap between themselves and the urban centres, thus improving their standard of living. The outcome will be a participatory GIS (geographic information system) process and knowledge-base that bridges the gap between data and users.

Back to top