2014
Type of resources
Keywords
Publication year
Service types
Scale
Topics
-
Spatially continuous information of seabed sediments provides important information for seabed mapping and characterisation, prediction of marine biodiversity, and marine environmental planning. Seabed sediment data is available at sampled point locations, so spatially continuous data can only be predicted from the point data. The accuracy of the predicted data is crucial to evidence-based decision making in marine environmental management and conservation. Improving the predictive accuracy is essential, but also challenging, since the accuracy is often data-specific and affected by many factors (Li & Heap 2011 and 2014). Because of their high predictive accuracy, machine learning methods were introduced by Geoscience Australia into spatial statistics by combining them with existing spatial interpolation methods (SIMs). This resulted in new hybrid methods, with the hybrids of random forest (RF) with inverse distance weighting (IDW) or ordinary kriging (OK) (i.e. RFIDW or RFOK) displaying high predictive capacity (Li & Heap 2014, and Li et al. 2011). However, their applications to environmental variables are still rare. Model selection for RF and the hybrid methods has proven to be necessary and further testing is required. Moreover, model averaging was argued to be able to improve predictive accuracy, but no consistent findings have been observed in previous studies. In this study, we aim to identify the most accurate methods for spatial prediction of seabed sediments in the Petrel sub-basin, northwest Australia. We experimentally examined: 1) the effects of input secondary variables on the performance of RFOK and RFIDW; 2) whether the performances of RF, SIMs and their hybrid methods are data-specific; 3) the effects of model averaging on predictive accuracy of these methods; and 4) whether additional samples improve spatial predictions of seabed sediments. For RF and the hybrid methods, up to 21 variables were used as predictors. The predictive accuracy was assessed in terms of relative mean absolute error (RMAE) and relative root mean squared error (RRMSE) based on the results of 100 iterations of 10-fold cross-validation. This study found that: - the predictive errors changes with the input secondary variables; - the most accurate model could be missed during the model selection; - RFOK proved to be the most accurate method; - the methods are not data specific, but their models are, so the best model needs to be identified; - model averaging is clearly data specific; and - contribution of additional samples to the spatial predictions is not apparent. In summary, model selection is important for identifying the most optimal models for RF and the hybrid methods. Effects of model averaging should also be examined for individual studies. The hybrid methods displayed substantial potential for predicting environmental properties and are recommended for spatial prediction not only in environmental sciences but also in other relevant disciplines. Guidelines are provided in this study for improving spatial predictions of biophysical variables in both marine and terrestrial environments.
-
This paper presents the initial building vulnerability schema proposed for the Bushfire and Natural Hazards Collaborative Research Centre (BNHCRC) project entitled 'Cost-Effective Mitigation Strategy Development for Building Related Earthquake Risk'. The development of a building schema which categorises the Australian building stock into distinctive vulnerability classes is an integral part of the risk and impact assessment process. In undertaking this categorisation a review was undertaken of existing earthquake vulnerability schema found in the literature alongside a schema developed by an expert group for Australia. The schemas found in the literature were the HAZUS, the United Nations Global Assessment Report on Risk (UN-GAR), the RiskScape, the Global Earthquake Model (GEM), the US Geological Survey Prompt Assessment Global Earthquakes for Response (USGS PAGER), and the European Macroseismic Scale-98 (EMS-98). Also included was an Australian specific schema developed based on the recommendations made at a workshop in Melbourne in February 2001 (Stehle et al., 2001). Key building parameters from each of these were considered along with the building types found in the countries or regions where these schemas were developed. The proposed schema categorises buildings by the building attributes: Building Usage, Primary Lateral Load Resisting System, Height Range, Proximity to Coast, Wall Type, Wall Material, Roof Material, and Age. The draft schema has been developed in recognition of the current and projected ability to define national building exposure and of the parallel BNHCRC mitigation projects examining vulnerability to wind and riverine flooding. While vulnerability schemas are hazard specific, alignment has been sought with schemas for the other hazards where possible. The draft schema is considered to be a preliminary version, and is expected to evolve during the project as it develops new knowledge on vulnerability and mitigation options for key high risk Australian building types.
-
A review of mineral exploration trends, activities and discoveries in Australia in 2013-2014
-
Due to extensive cover by Mesozoic and younger sedimentary basins and regolith, the geology of the southern Thomson Orogen is poorly understood. Small outcrops of the Thomson Orogen are exposed along the Eulo Ridge (Qld) and in the southwest around Tibooburra (NSW). Between these regions the thickness of cover averages ~200 m, which is within exploration and mining depths. The southern Thomson Orogen is true 'greenfields' country. Although the mineral potential of the region is largely unknown, the northeastern Thomson Orogen (e.g., Thalanga, Charters Towers) and the similar-aged Lachlan Orogen to south are well mineralised (e.g., Cadia, Cobar). In order to attract investment (exploration) into the southern Thomson Orogen, and also to improve the geological understanding of the area, Geoscience Australia, the Geological Survey of Queensland and the Geological Survey of New South Wales have commenced a collaborative project to collect new (and synthesise existing) pre-competitive data. The project is well advanced in its first stage, which includes a synthesis of existing datasets across the state borders to create a revised solid geology map (contributions by Purdy et al. and Doublier et al.). This map will form the basis of a 3D model (map), which will utilise pre-existing government and industry seismic and drilling data. In support of the 3D map, a programme of geophysical datasets will be acquired, processed and interpreted. These include: airborne electromagnetic (AEM - data acquisition and processing underway), broad-band magnetotelluric (MT) and gravity data (both in advanced planning stage). In order to understand the nature of the cover rocks and their relationship to basement, the first sampling campaign of a surface geochemical survey has been carried out providing a higher resolution infill of the National Geochemical Survey of Australia (NGSA) by expanding an earlier CRC LEME survey northwards into southern Queensland. In addition, the potential mineral systems of the region - and their relationship to adjacent areas - are currently being assessed, with initial studies focussing on timing, isotopic signatures, and structural context. Interim products and datasets will be released throughout the project, with the final results delivered to industry in 2016-17.
-
Recent events in Queensland in 2011 and 2013 have highlighted the vulnerability of housing to flooding and have caused billions of dollars in losses. To reduce future losses there is a significant need for mitigating the risk posed by existing residential buildings in flood prone areas. Therefore, a project is underway within the new Bushfire and Natural Hazards CRC to provide an evidence base to inform decision making on the mitigation of flood risk by providing information on the cost-effectiveness of a range of mitigation strategies. As an initial step to assess mitigation options, a building schema is developed to categorise the Australian residential buildings into a limited number of typical building types for which vulnerability functions can be developed. The proposed schema divides each building into the sub-elements of foundations, bottom floor, upper floors (if any) and roof to describe its vulnerability. The schema classifies each building storey type based on the attributes of Construction Period, Fit-out Quality, Storey Height, Bottom Floor System, Internal Wall Material and External Wall Material. The schema defines 60 discrete building/vulnerability classes based on the above mentioned attributes. These exclude combinations that are invalid in an Australian context. Furthermore, the schema proposes 6 roof types based on material and pitch of the roof. Within the project, a literature review of publically available research on retrofitting of flood prone building components and the strength implications of immersion of key structural elements is also being conducted. The outputs of this research will be the provision of information that gives a clear evidence base for optimal retrofit of flood prone existing buildings in Australian communities. The outputs will also enable more cost-effective construction of flood resistant structures in the future.
-
Accurate measurement of thermal maturity of organic matter in sedimentary rocks that pre-date land plants is challenging. Chemical and petrographic examinations can provide reproducible thermal estimates but unless calibrated to absolute maturity scales (e.g. based on vitrinite reflectance and/or temperature) their application in burial histories will result in misleading petroleum prospectivity. Hence various published correlations exist between vitrinite reflectance (VR) and chemical maturity parameters (e.g. Tmax, MPI-1), the reflectance of other organic macerals (e.g. inertinite, alginite and bitumen) and faunal remnants (e.g. conodonts, graptolites and chitinozoans). Herein, we use geochemical (Rock Eval, PyGC, GCMS of saturated and aromatic hydrocarbons) and petrographic techniques on organic matter from the early-Middle Cambrian Arthur Creek ('hot shale') Formation, Georgina Basin to characterise, compare and determine the thermal maturity on a composite natural maturity sequence; immature to gas mature (Figure 1). Petrographic maturity assessment employed modified vitrinite inertinite reflectance fluorescence (VIRF) which utilises quantitative fluorescence to objectively classify reflectance populations on a bivariate chart [1]. For very high maturity samples with negligible fluorescence Ro random was measured to correlate zooclast versus VR relationships [2].
-
Optical, Geospatial, Radar, and Elevation Supplies and Services Panel (OGRE) 2012-13 Annual Report (version 2 suspersedes GeoCat 77426)
-
Landslides are a complex geological hazard triggered by a combination of factors depending on their magnitude and type (Figure 1). There are a number of methodologies employed for landslide susceptibility mapping around the world. The method adopted should vary according to the individual characteristics of the landslide being considered. The method of landslide susceptibility mapping adopted here was developed using an existing method, the InfoVal method (van Westen 1997), adapting it for use with the open source software QGIS. QGIS was chosen as the GIS system due to its use by other natural hazard scientists in Papua New Guinea and in the region, and because it is free and open source.
-
Towards a Global Discrete Nested Grid: A Framework to Enable Multi-Resolution and Multi-Domain Analyses Simon Oliver (Geoscience Australia), Matthew Purss (Geoscience Australia), Adam Lewis (Geoscience Australia), Robert Gibb (Landcare Research NZ), Brian Sloan (ANZLIC) Geoscientific data has already exceeded the petabyte-scale barrier and is rapidly heading toward the exabyte-scale barrier. Converting this massive amount of data into timely information and decision support products is dependent on our capacity to rapidly analyse this data in a transparent and repeatable fashion. This can only be achieved through the conversion of traditional data archives into standardised data architectures that support parallel processing in high performance compute environments. Already, the challenges of high velocity, high volume (> a terabyte per day) data requires us to rethink the way we store data in order to make best use of it. These challenges will only grow as the variety and complexity of datasets we wish to combine to produce near-real-time decision support information increases. One of the key elements to standardising data architectures is a globally consistent grid. A discrete global nested grid that supports global data models is required to provide a common framework to link very large multi-resolution and multi-domain datasets together and to enable the next generation of analytic processes to be applied. Such a grid framework must be capable of handling multiple data streams rather than being explicitly linked to a sensor or data type. Work has now commenced to development international standards which will inform the specification of the global discrete nested grid.
-
Modern estuaries are part of a continuum of coastal depositional environments within which the variation in geomorphology is closely related to the dominant one of three main processes affecting sedimentation, viz waves, tides or rivers. The present location of the coast is controlled by the sea level rise brought about by the release of water from continental ice sheets following the glacial maximum around 20,000 years ago. The current form of the coast is partly inherited from the shape of the precedent land surface flooded by the rising sea, which is then modified by a combination of ongoing local erosion and/or deposition of sediment transported by rivers from the adjacent land mass, and then redistributed by the locally dominant marine process. Once sea level stabilised around 6-7,000 years ago, sediment was able to progressively infill the topographically lower areas. In some cases, where the rate of sedimentation is relatively high, infill of coastal indentations may have been completed, and the coastal is now prograding seaward. Elsewhere, where sedimentation rates are lower, or waves and tides are able to effectively move sediment away from the point of river entry, infill may have only partially proceeded, and the coast has been modified into characteristic forms. Where waves dominate over tides, features made from coarse-grained sediments such as barriers, beaches and bars, form parallel to the general trend of the coast. These establish less-energetic environments isolated from the full force of the ocean, where fine-grained sediments can accumulate. Where tidal forces are relatively dominant, the coarser-grained bars tend to orient at right angles to the coast, and fine-grained sediments accumulate in the intertidal areas as mud flats, and marshes.