Infrastructure Engineering - Theses
Permanent URI for this collection
Search Results
1 - 2 of 2
-
ItemEucalypt leaf-flush detection from remotely sensed (MODIS) data( 2011)The apiary industry in Australia is unique from the rest of the world in that it is dependent upon tree species to provide the nectar for honey production, whereas the rest of the world utilises meadow vegetation and shrub species for nectar supply. Consequently, the apiary industry in Australia relies upon the ecosystem-dominant eucalypt trees within closed-forest, forest, open-forest, and woodland environments that have more complex ecosystem interactions than the meadow and shrub ecosystems outside Australia. Therefore, the Australian apiarists are dependent upon acquiring knowledge of greater complexity pertaining to forest ecosystem dynamics, such as including successional-ecology, interactions with different understorey community types, herbivory defoliation-events, and fire. This study has found it is possible to detect changes in vegetation indices over successive images, and between seasons, to correlate with honey-yield data. The supplied honey-yield data is used as a covariate for flowering intensity of eucalypt communities. Due to the lack of numerical honey-yield data (the supplied data was categorical/ordinal), modelling of the remote sensing vegetation indices with honey yield could not be done, and the analyses rely upon interpretation of graphical and rate-of-change values. This Master-by-Coursework project is a preliminary component of a larger project attempting to find a method of predicting flowering intensity of eucalypt ecosystems, and therefore potential honey yield, using the freely-available MODIS data. The data used for this project was the NDVI, EVI, Pixel Reliability, and VI Quality information extracted from images accessed from the MOD13Q1 product. With access to numerical honey-yield data, and weather and ecosystem data (available from the Bureau of Meteorology and the Department of Sustainability and Environment respectively), along with the MODIS data, it should be possible to perform a detailed analyses using multiple-regression techniques to model the flowering intensity of eucalypt ecosystems based upon prior leaf-flushing events. Predicting potential honey yield from MODIS data coupled with weather and ecosystem data will allow apiarists to determine the most likely site and time to place their beehives in order to maximise production.
-
ItemA linear regression analysis of factors affecting pavement roughness for asphalt and sealed pavement roads in Victoria( 2010)The investment on pavements is the largest part of capital investment in any modern highway system. Maintaining and operating pavements draw great attention of administrators of Australian road system. In large systems of pavement, pavement management systems can be very complex and look at very small kinds of differences in asphalt mix design, loading types, pavement thicknesses. The study of these details may generate less than a percent increase in the effective use of maintenance funding. It is significant to save even a half percent on rehabilitation costs for hundreds of millions of dollars system. Many types of distress are assessed in order to investigate the most contributed factors, which affect the overall pavement condition, should be identified to limit the surveyed distress types, including roughness, rutting and cracking. The evaluation of relationships between specific distress types so that the data number could be minimized in collection. Therefore, the cost on pavement condition survey for various types of distress could be reduced greatly. This paper focuses on analysing the linear relationship between pavement roughness and rutting and the relationship between roughness and cracking. The ranges of R square for the relationship between rutting and roughness from 0.0460 to 0.497 for in rural area; urban area has R square ranges from 0.0003 to 0.5719. The ranges of R square for the relationship between cracking and roughness from 0.0043 to 0.6205 for in rural area; urban area has R square ranges from 0.0056 to 0.5769. From the assessing of P-value point of view, in the relationship between roughness and rutting, the average P-value is 0.1803 for rural area; it is 0.3641 for urban area. The relationship between roughness and cracking, the average P- value is 0.3727 for rural area, it is 0.2746 for urban area. In the result, slopes of correlation of rutting and roughness, which are indicated by correlation coefficients, are ranged from -1.2613 or 2.2298. The slopes of cracking are ranging from -0.7143 to 0.0682. The inference could be made that the change of dependent variable roughness is not sensitive to the effect of the independent variables cracking and rutting. It is very low for the linear relationship of roughness variance and AADT. The mask effect of maintenance may attribute the result due to the variance of roughness is decreased by maintenance treatments. The results of this research confirm that the linear relationships between roughness and cracking, between roughness and rutting are small for the statistical significance of influence the rate of progression. The extent of rutting influence roughness has more sensitivity than cracking.