eResearch

You are here

ANDS OzTrack Project

NeCTAR Logo OzTrack Project Update

From June 2012, The University of Queensland is proud to be in partnership with the National eResearch Collaboration Tools and Resources (NeCTAR) project to further develop the OzTrack eResearch Tools. OzTrack was initially developed with support from the Australian Nataional Data Service (ANDS) during 2010-2011.

Visit the OzTrack upgrade (funded by NeCTAR) page

Overview

The ANDS OzTrack project produced data management and data analysis services for the Australian animal tracking community.

Funding
Australian National Data Service (ANDS) Data Capture funding
Period
2010-2011
Structure
OzTrack is a collaboration between the UQ eResearch Group and the UQ Ecology, Conservation and Organismal Biology (ECO)-Lab
Project team
Prof Jane Hunter, Project lead, UQ eResearch Group
Prof Craig Franklin, Project advisor, UQ ECO-Lab
Dr Hamish Campbell, Data analyst, UQ ECO-Lab
Matthew Watts, Data analyst, UQ Environmental Decisions Hub
Ross Dwyer, Data Analyst, UQ ECO-Lab
Peggy Newman, Software developer, UQ eResearch Group
Dr Nigel Ward, Project Manager, UQ eResearch Group

Significance

A large number of research projects, currently being undertaken within Australia tag animals with electronic sensors in order to track species movement. Marine animals (crocodiles, sharks, whales, turtles and rays), avian taxa (birds, bats), terrestrial species (lizards, cane toads, koalas, dingos) and production livestock (cattle, sheep) are all being monitored to capture quantitative data that can be used to improve our understanding of their movement and behaviour. Recent advancements in telemetry technologies are generating large, continuous, high-frequency datasets that precisely document animal behaviour. Electronic devices, including acoustic, radio, archival, pop-up archival (PAT) and satellite positioning (GPS) tags, are revealing when, where and how animals travel, and how these movements relate to their environment. The ability to predict the movement of animals, based upon an understanding of what drives their movement, has a key role to play in environmental and marine conservation and management. It is also of critical importance to addressing environmental challenges including invasive species, infectious diseases, climate and land-use change.

Despite the increasingly widespread adoption of animal tagging devices, there are significant challenges associated with the management and analysis of the datasets being generated. The sheer scale and complexity of the data sets make it difficult for many research groups to exploit the full potential of the data. The processing of telemetry data currently relies on a diversity of software and file formats. Data capture, data management and data analyses steps currently involve manual and time-consuming data manipulation and import/export to different software products. The integration of animal movement data with complementary environmental information (e.g., remote sensing data) presents additional technical and computing challenges.

To date the vast majority of data from animal tracking studies undertaken throughout Australia is stored in small personal databases and inaccessible to the broader scientific community. Collation of these studies within a central repository would greatly increase data transparency, reduce study reproduction and enable comparisons of results between study groups. The Oztrack project aims to accelerate scientific research in this field by creating a common approach to the management and analysis of these large and diverse datasets for modelling animal behaviour and ecology.

Goals

Throughout 2011 the project focussed on the development of a set of eResearch software tools to support the capture, storage and analysis of animal tracking data being generated through telemetry devices. The project aims to:

  • establish a species tracking data repository that can be scaled up to support the Australian species tracking community;
  • solve the complex problem of projecting geo-coordinates collected from animals over a wide geographical area for routine analysis;
  • link the underlying database, analysis (R) services and visualisation (map layers) components through an intuitive but flexible, interactive and secure Web interface;
  • provide a suite of state-of-the-art Web-based software tools to enable search, browse, visualization, analysis, and animation of the species movements over time and space;
  • enable fast, effective and accurate management and analysis of extremely large volumes of data derived from tracking devices;
  • support secure authenticated sharing of analysed tracking data within and between research teams, government departments and industry (through the adoption of AAF);
  • enable the development of models that can predict spatio-temporal movement of particular species under different potential scenarios;
  • share RIF-CS research data collection descriptions of animal movement via the ANDS Research Data Australia (RDA).

OzTrack portal

The project delivered the OzTrack portal, a website for storing, analysing and visualising animal location data.

Data storage

OzTrack users firstly describe their projects, identify team members and define access rights for data files.

Researchers collect data from tracking devices, convert it to raw CSV format, and then upload it via a web form to a project within OzTrack. The upload process handles a variety of different formats for times, geospatial coordinate and animal identifiers. The aim is that files need only have the correct headers.

After processing and validation, the data is stored in an object relational database (PostgreSQL) with spatial extensions (PostGIS).

Track visualisation

The spatial and temporal complexity of animal tracking data prohibits effective analysis without visualisation tools.

visualising animal tracks
The OzTrack portal uses OpenLayers to visualize location data for one or more animals. In the Analysis Tools section, users see the tracks of all the tagged animals contained in the project. The trajectory from each animal is automatically assigned a unique colour, also displayed next to the Animal ID. The first point in each track is displayed by a green icon and the final point in the series is displayed by a red icon. Animals can be added and removed from the display by ticking the check box next to each individual.

Clicking on an Animal ID displays details of the animal's trajectory with the start and end dates. Users can pan and zoom within the map as well as download KML to view their animal location data as points in Google earth.

Home range analysis

visualising home ranges
The OzTrack portal can calculate home range density estimators such as minimum convex polygons and kernel utilisation distributions using Adehabitat, an open source package written in R containing commonly used habitat analysis tools. This package is often used in the animal tracking research community. This ensures that the home range estimators are accurately calculated using a trusted and widely used array of statistical functions.

The web server sends the geographic location data (longitude/latitude) to the R-driven program located on the OzTrack server. These are then projected into the correct localised coordinate system and the home range calculations are performed using the adehabitatHR library. The projected estimates and placement of home range area are returned to OzTrack for viewing by the user.

Access to data and metadata

collection description in ANDS Research Data Australia
OzTrack provides public pages for each of the projects it manages. The pages describe the species being tracked, the geographic and temporal extent of the data, rights and access statements for the data, as well identifying a contact for the data (the project owner).

Project owners have the option to publish their collection descriptions to both the UQ Data Collections Registry, and to ANDS Research Data Australia. This is ideal for researchers who wish to promote their research and discover similar data, whilst still protecting the intellectual property contained within their datasets (the datasets themselves are not published to Research Data Australia).

Project team members can filter and view raw tracking data and download it as CSV or KML. Project owners can also choose to make raw tracking data available to anyone with an OzTrack account, or open it up for public access.

Source code and system architecture

The source code implementing the OzTrack portal is available in GitHub under a GNU General Public License version 2 (GPL-2.0). The Adehabitat R library, and other R libraries used by OzTrack, are available under GPL-2.0 licence from CRAN (Comprehensive R Archive Network) and R-Forge.

OzTrack high-level system architecture

Researchers interact with the system via a web interface served from a Tomcat hosted Java servlet. The web interfaces are implemented using the Spring Web MVC (model-view-controller) framework. Authentication and access control are implemented using Spring Security.

OzTrack dynamically presents animal detections, tracks and home ranges to the user via JavaScript. JQuery captures and displays user requests. The OpenLayers mapping tools display animal tracks and home ranges on a Google Map. Additionally, users can download a KML representation of detections for individual animals for external analysis and display in tools like Google Earth.

The application calls R routines to perform geospatial statistical analysis. The application communicates with R via the Rserve TCP/IP server. The system currently calls R routines from the adehabitatHR R library to perform home range estimation, but has been designed to allow use of any R analysis tool that can return KML.

Geospatial data and project metadata are stored in a PostgreSQL database, extended with the PostGIS support for geographic objects. The application uses JDBC and Hibernate as abstractions to isolate the application code from the underlying the database.

The system syndicates collection level metadata (OzTrack project metadata) to the UQ data collections registry (dataspace.uq.edu.au) using a profile of the Atom Syndication Format transmitted using AtomPub. Dataspace then syndicates the description to ANDS Research Data Australia as RIF-CS transmitted using OAI-PMH.

Presentations and posters

P. Newman, N. Ward, H. Campbell, M. Watts, C. Franklin, J. Hunter, "OzTrack: Data Management and Analytics Tools for Australian Animal Tracking", eResearch Australasia, 2011, Melbourne 6-10 Nov, 2011

P. Newman, N. Ward, H. Campbell, M. Watts, R. Dwyer, C. Franklin, J. Hunter, "Data Management and Analytics for Australian Animal Tracking", Poster session at eResearch Australasia, 2011, Melbourne 6-10 Nov, 2011

Future work

The University of Queensland is currently in negotiations with NeCTAR (National eResearch Collaboration Tools and Resources) regarding a short listed proposal that aims to refine and expand the existing infrastructure developed through the ANDS-funded OzTrack project, so it can be adopted by the wider species tracking community within Australia and so that it provides a broader range of statistical analysis and modelling services than is currently supported. In addition, the NeCTAR funding will enable the development of a set of robust and sustainable services (by leveraging the NECTAR Cloud for hosting, RDSI for data storage, ANDS for data promotion, AAF for authentication). It will also enable the integration of the resulting animal tracking data with related environmental data (such as IMOS, TERN, ALA, Geosciences Australia and BOM data) to understand the impact of environmental changes on species migration, health and behaviour.


ANDS Logo This project is supported by the Australian National Data Service (ANDS). ANDS is supported by the Australian Government through the National Collaborative Research Infrastructure Strategy Program and the Education Investment Fund (EIF) Super Science Initiative.