Data Collection

To address the primary research objectives and question, data from two different forest measurement pilot projects were utilised. The first project undertook data collection in 2000, and concentrated on a study area in the south central Queensland forest and woodlands. The project was funded by the Cooperative Research Centre Greenhouse Accounting (CRCGA) (Tickle, et al., 2006; CRCGA, 2006) and an Australian Research Council SPIRT grant (Lucas, et al., 2004). The Queensland study area was the primary site for this thesis, and was where all development of the models was undertaken. The second project collected data in 2003 as part of a National Forest Inventory Continental Forest Monitoring Framework pilot project in North East Victoria (described in the Introduction chapter). The NE Victorian site was a secondary site for this thesis, and provides validation for the models developed at the Queensland site, and highlights any regional modification that may be required if the LiDAR models were to be utilised in other regions.

3.3.2 Queensland study site

The Injune study focused on an area of mixed species open forests and woodlands near the town of Injune (Latitude 25° 36’ S, Longitude 147° 30’ E) in the Southern Brigalow Belt (SBB) Interim Biogeographical Regionalisation for Australia (IBRA) region (Thackway and Cresswell, 1995) (Figure 8). The SBB region, which was one of 16 IBRA regions within Queensland, constitutes 12.5 % of the state’s land area. Approximately 8,901,250 ha, or 41.2 %, of the SBB was presently covered by forests and woodlands. The main study site had an area of 221,120 ha, with approximately 192, 153 ha (86.9 %) identified as forests and woodlands. Within the study site elevation varies from 437 - 850 m and the mean annual rainfall was approximately 635 mm, although was variable, with most recorded between December and February (summer). The mean annual maximum temperature was 27o_{C (Bureau}

Figure 8: Location of the 37 x 60 km Injune study area, within south-central Queensland.

The area was first explored by Europeans in 1846, with Thomas Mitchell travelling north through the eastern portion of the Injune study area following the Maranoa river, and recording landscape and vegetation observations along the way (see Figure 9; Mitchell, 1846). Settlement by graziers mostly likely occurred in the 1860’s and railways were constructed in the 1870’s. Severe drought in the 1890’s forced many landholders off their properties. From the early 1900’s the Queensland forestry department began reserving land and timber for wood production, with commercial harvesting of Callitris and Ironbark (Eucalyptus) species occurring from the 1940’s onwards. Throughout the area, selective logging has only occurred once at any one site. The level of timber resource use reflects both the extent of the resource and the relatively slow growth in this area. Growth was influenced by relatively poor soils and low rainfall (B. Howard, Qld DPI, pers. comm.). This then indicates that major human induced structural change in the forest of the study area has not been widespread (as indicated in Figure 9).

The forests in the study area contain a diverse range of species, although several dominate, and both excurrent (i.e. conical shape common to many gymnosperms) and decurrent (i.e. round or spreading shape common to most angiosperm trees) structural forms were commonplace (Lucas et al., 2004). White cypress pine (Callitris glaucophylla) dominates many stands but this species was selectively harvested, so large individuals were typically absent. Most Callitris occur in dense stands comprised of a large number of smaller individuals (up to several trees per m2_).

Figure 9: Sketch of the Maranoa River by the explorer Thomas Mitchell in 1846 (upper) (Mitchell, 1846); and a photo of the Maranoa from a nearby location in 2004 (lower).

Eucalyptus species are widespread throughout the area, with Silver-leaved Ironbark (E. melanaphloia) and Poplar Box (E. populnea) dominating. Tumbledown Red Gum or Baradine Gum (E. dealbata var chlorodata) and Smooth Barked Apples (Angophora leiocarpa) occur along the creeks and at scattered locations throughout the landscape, and individual trees often contribute the greater proportion of the stand biomass. Larger individuals of both Eucalyptus and Angophora species form the upper canopy of many stands, below which several layers of sub-canopy trees of varying densities occur. Brigalow (Acacia harpophylla) and tall understorey shrubs such as Wilga (Gejeira parviflora) and Sandalwood Box (Eremophila mitchelli) are commonplace. Many stands contain a mix of Acacia, Callitris, Eucalyptus and Angophora species. A wide range of growth (and therefore structural forms) exist because of the varying impacts of different soil types, natural fires, droughts, clearing (e.g., pulling, poisoning) and grazing (Tickle et al., 2006).

Overview of Queensland multi-stage sampling

The following section presents an overview of the Injune study sampling strategy. It should be noted that the sampling strategy and data acquisition were carried out prior to the initiation of this thesis. The methods are described in more detail in Lucas et al., (2006a) and Tickle et al., (2006).

The acquisition of image and field data was undertaken in four main stages (Table 4). In Stage I, a systematic sampling scheme (Schreuder et al., 1993) was implemented to guide the acquisition of large scale aerial photography (LSP) (Stage II) and LiDAR data (Stage III). Following collection and initial interpretation of these data, a stratified random selection strategy was used to select 34 field plots, where forest inventory data were collected (Stage IV). The majority of the fieldwork was carried out during the period of LiDAR data acquisition and within one month of the LSP data acquisition, thereby minimising any seasonal effects and the likely impacts of anthropogenic land cover change at the field sites. Description of processing stages I, II, and VII can be found in Tickle et al., (2006) (Appendix C), whilst stages III, IV, V, VI and VIII are described in subsequent sections.

Table 4: Main stages in the acquisition, processing and analysis of field and remotely sensed data (Tickle, et al., 2006)

Stage Task Purpose

Sampling and data acquisition

I Sample design To select appropriate field sample locations II LSP capture and pre-stratification To allow description of the species/community _composition III LiDAR capture _{(height, crown, foliage and/or branch cover)} To facilitate retrieval of structural attributes IV Field sampling To provide ground truth for interpretation of _{LSP and LiDAR and validation of products.}

Post-processing

V Georeferencing of LSP to LiDAR To allow overlay of API vector information VI Generation of LiDAR height _surfaces Calculation of a bare earth DEM and _{vegetation height}

Data analysis VII forest communities based on LSP Classification of

interpretation

To determine spatial distributions of dominant, co-dominant and sub-dominant species VIII Tree height, FBC, FPC and canopy cover

retrieval from LiDAR

To provide individual tree and stand level estimates