Typescript. Includes bibliographical references (leaves [279]-287) This thesis examines the growth ring structure of Eucalyptus camaldulensis Dehnhardt and investigates links between ring features and the climatic conditions under which growth took place. Samples of E. camaldulensis from the Barmah Forest (near the River Murray in northern Victoria) were used in the study. E. camaldulensis growth is linked to periodic flooding, and the Barmah Forest contains sites that are frequently flooded. Wood samples were obtained from sites subject to different average flooding frequencies. Trees used in the study grew out of natural regeneration in the 1920's and 1930's and from regeneration trials in the early 1960's. Initial investigation of E. camaldulensis samples revealed ring-like features that were able to be traced on samples by eye. Microscopic investigation showed that there existed considerable variations in the properties of individual rings at different positions on the samples, and that the boundaries between rings were often indistinct. Further examination of E. camaldulensis microstructure was performed on samples from two trees that grew on sites with significantly different flooding regimes. These samples possessed features that formed rings, with rings being successfully matched between samples taken from different heights in the trees. As the complex microstructure of E. camaldulensis did not lend itself to standard dendroclimatological techniques, methods were developed to facilitate the comparison and matching of rings. These methods were also used in the successful matching of ring patterns with the output from a simple climate-based tree growth model. Two sets of E. camaldulensis samples (BS1, with 33 samples, and BS2, with 39 samples) were studied to assess the level of individual variability in ring patterns, and to provide a representative ring pattern for climate comparison. Following the development of methods for identifying samples with similar ring patterns, a subset of similar samples was selected from the BS1 set. A member of this subset was selected to provide a ring-width pattern upon which a representative pattern of ring features for BS1 was based. The rings of the BS2 samples possessed poorer ring definition than the BS1 samples and provided no new or different information. Consequently, the representative ring pattern for BS1 was used in a dendroclimatological investigation for the site. The ring pattern was matched with the output from two tree growth event models. Although rings were matched with growth events over a 27 year period, the high variability of individual ring features prevented matching of particular types of ring features with particular types of climatic events. An investigation of numerical methods for matching ring patterns with ring or growth event patterns, and for identifying samples with similar ring patterns, was performed using signal smoothing and filtering techniques and a dynamic time-warping procedure. Ring matching and identification of similar ring patterns was found to be most successful on samples where the ring patterns, expressed as continuous signals, had similar mean and amplitude values. The techniques were unsuccessful in the matching of signals of different form, such as continuous ring pattern signals and discontinuous growth event signals.
| Identifer | oai:union.ndltd.org:ADTP/55149 |
| Date | January 1995 |
| Creators | Argent, Robert Murray. |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Relation | VBX, VLU, VU |
Page generated in 0.0015 seconds