Date of Award


Degree Type


Degree Name

Bachelor of Science Honours


Faculty of Computing, Health and Science

First Advisor

Pierre Horwitz

Second Advisor

Dr Andrea Hinwood

Third Advisor

Tim Perkins


The Organic Rich Sediments (ORS) of the Swan Coastal Plain (SCP) are an important component of the ecosystem, an important source of habitat and refuge for many species of fauna and act as natural fire control mechanisms through edaphic measures within the landscape. There have been numerous impacts on ORS, with fire now rated as one of the imminent threats to the continued existence and functioning of these entities. Conversely fire is an important tool for land mangers in their efforts to promote biodiversity and manage fuel loads to prevent wildfires, at present many of these ORS are becoming victim to these management practices and other disturbance events due to the lack of adequate information relating to their location. The current inadequacies to the appropriate management of these systems lies in the apparent lack of detailed spatial information relating to the distribution of these sediments, with current practices failing to account for historical sediment deposits and deficiencies with identifying deposits at smaller scales. Geographical information Systems (GIS) have been widely used in numerous environmental disciplines to map and predict the spatial distribution of a diverse range of environmental entities from species through to landscapes. The predictive modelling capabilities of GIS were of particular interest to this study and in particular the ability to predict the spatial distribution of ORS based on existing data. Datasets on wetland type, soil type, slope, aspect and depth to groundwater were compiled to develop a model of ORS on the SCP utilising the weighted overlay model function in Spatial Analyst Arc View 3.2. The predictive capability of the model was ascertained by field sampling while the effectiveness of the model to predict ORS as determined by the calculated organic content of sampled soils was analysed by logistic regression. Field verification of the model confirmed the predictive capabilities of the model in being able to predict the occurrence of ORS, with correct prediction rates in excess of 50%. It also highlighted the spatial variation which exists between and within deposits which makes predicting the distribution of ORS a very complicated process. Resolution testing showed that the predictive capabilities of the model degenerated at a smaller scale, possibly as a result of limitations associated with the resolution of the model input data. Key findings of the logistic regression analysis of the predictive model showed that the developed model was robust at the landscape scale. The fit of the model to the data was very strong with SPSS percentage correct values in excess of80%. Logistic regression highlighted the impact of soil moisture and vegetation type in improving the fit of the model and highlighted the deficiencies associated with using the wetland type variable alone. The logistic regression results showed that the model performed similarly at both sites. A number of remote sensing technologies are available which are available to obtain empirical data relating to soil moisture contents and vegetation types at a landscape level. These include the use of thermal mapping, normalized difference vegetation index (NDVI) and the use of ground penetrating radar. This study has set the foundation for the development of a vital tool for mangers involved in preserving ORS and their ecological functioning on the SCP and throughout the southwest of Western Australia.

Included in

Soil Science Commons