Degree Name

Master of Science (Hons.)


School of Geosciences


This study investigated the use of satellite remote sensing and a Geographical Information System (GIS) in defining the habitat of the Ground Parrot (Pezoporus wallicus) and predicting the habitat suitable for Ground Parrots in Barren Grounds Nature Reserve and Budderoo National Park in New South Wales.

The Ground Parrot Pezoporus wallicus is a terrestrial, seed-eating parrot endemic to the coastal heathlands of eastern and southern Australia. Ground ~arrots were once common throughout SE and SW Australia but are now only distributed patchily along the south-east coast of Australia and in SW Western Australia. Data from studies of Ground Parrots by other researchers in Tasmania, Queensland, Western Australia, Victoria and New South Wales were used to determine the habitat requirements of the Ground Parrot.

A combination of listening and direct observation was used to determinethe present location of the Ground Parrots. The listening and direct observation were carried out on a irregular basis. Observation data of Ground Parrots from other researchers at Barren Grounds was also used.

A four band Landsat Thematic Mapper image was rectified and registered to the Australian Map Grid. A Maximum Likelihood supervised classification of the image was used to allocate pixels to 50 classes. The 50 classes were merged into six landcover classes following examination of between-class and within-class variation.

Field sampling was carried out in each of the mapped classes. Quadrats were accurately located using a global positioning system or surveying equipment. Projective foliage cover, height and plant species type were recorded in each quadrat. The results were used to label the six classes as closed heathland, closed graminoid heathland, sedgeland, woodland, forest.

Two geographical information systems were used in this study, SPANS GIS and E-RMS. The landcover, fire age, slope, aspect and drainage map along with the location of the ground parrots were imported into the SPANS GIS and E-RMS and used in the modelling of the habitat.

Both deductive and inductive modelling techniques were used in this study. The deductive, or knowledge-driven, models use the knowledge of an expert regarding the subject to set up the model parameters. The inductive, or data-driven, models use the training data such as the location of a species in relation to a number variables.

The deductive models used were boolean logic, weighted index overlay and fuzzy logic models. The boolean model was simple to implement and visualise and may be used in both E-RMS and SPANS. Weighted Index Overlay and Fuzzy Logic models were more complex to set up and may only be used in SP ANS. They had the distinct advantage of allowing a more flexible combination of maps and providing more useful information.

The inductive models used were weights of evidence and decision tree models. The weights of evidence model was implemented in SPANS and the decision tree inductive models in E-RMS. The weights of evidence model wasfound to be the most effective in determining suitable habitat although the success of all the inductive models depend to a large extent on the quality and the quantity of the known locations of the dependent variable, the Ground Parrot.

The study illustrated that conventional image classification routines provided a proven method of mapping vegetation with Landsat Thematic Mapper data. The modelling indicated that there is a complex relationship between vegetation structure, drainage, slope and fire history and Ground Parrot populations. Habitat modelling provides a valuable tool for the assessment of suitable habitat for a variety of species. Each method has advantages and disadvantages and both the modellers and the end users need to understand the assumptions that have been made during the development and application of the models. The methods used in this study could also be applied to other areas where Ground Parrots are known but where distribution has not been mapped. The methodology could also be adapted to map habitat suitability for other rare and endangered species.