ESTRATEGIA UTILIZADA

Field inventory data were collected during August 2000. The collection of field data over the same period as the remotely sensed data acquisition was considered necessary to limit the impact of changes in seasonal foliage cover or land cover (associated with disturbance by fire or clearing) on the subsequent development of relationships with remotely sensed data.

Prior to acquisition of the field data, a 100 x 100 m regular grid on a transparency sheet was overlain on the overlap area of each of the 150 hardcopy large scale photography stereo pairs. A count of dots (located at the intersection of the grid lines) were used to estimate the proportions of land use, land cover and forest types as well as forest height and cover, disturbance regimes and vehicular access (Jones, 2000).

After interpreting the photography, the API codes were used to stratify and identify suitable locations for field sampling. The stratification assumed that the vegetation types contained within the 150 primary sampling units were representative of the proportions across the entire study area. For the purposes of stratification, the different forest types were classified into four distinct structural categories: Acacia or sparse vegetation (containing species such as BGL, SWB, BLH, BOK); Callitris (e.g., CP-, FMP); Eucalypt Ironbark (e.g., SLI, GTI, BRI, NRI); and Eucalypt Other/Angophora (e.g., PBX, ECH, SBA, RBA) (see Table 55, Appendix A for the species name and API codes). The Eucalypt class was split as the various Ironbark species were seen to contribute a significant proportion of the mapped landscape. Each forest type was then ranked into three (low, medium and high) potential and relative biomass classes using a combination of API and biomass estimates from the Japanese Earth Resources Satellite (JERS-1) Synthetic Aperture Radar (SAR) data, resulting in 12 vegetation strata (Jones, 2000; Lucas et al., 2000).

It was determined that the field inventory would be limited to 2 - 4 secondary sampling units per day using 2 field crews of 5 staff. Therefore, out of a potential 150 primary sampling units, 13 were selected that contained the necessary strata and met access, travel times, and safety criteria. Within these 13 sampling units, 34 secondary sampling units were randomly selected across the 12 strata (in proportion to their area within the 150 primary sampling units).

Once located, a 50 x 50 m square plot, aligned in a north-south direction, was established using GPS survey and laser range-finding equipment. Tapes of 50 m length were then laid out to produce a 10 x 10 m grid to guide the subsequent location of trees for measurement. Within each plot, the location of all trees > 10 cm in diameter (at 130 cm above ground level) was recorded digitally by placing reflectors at each of the plot corners and then using either a GEOSCAN or CENTURION Laser Rangefinder to record the distance and angle from each tree to the nearest visible reflector. Trees 5 - 10 cm in diameter were located by reading the x and y distances (in cm) from 50 m tapes placed perpendicularly (at 10 m intervals) across the entire plot. The cover and height of trees and shrubs < 5 cm in diameter was estimated within five 10 x 10 m sub-plots, with the centres of four located at a distance of 10 m from each of the corners and a fifth located at the centre of the plot. Within each plot, each tree was identified to species level and key measurements recorded included trunk diameter (cm, at both 30cm and 130cm) and height (m) to the top of the tree, to the canopy base, and to the first green limb. Three additional secondary sampling units that were identified as non-forest but containing regenerating vegetation, species and structural measures were selected. Within these ‘regrowth’ plots five 10 x 10 m subplots were located.

Transects were established within the field plot to estimate vegetation cover. Transects consisted of three 50 metre tapes laid out in the north-south direction at 10, 25 and 40 m, moving eastward from the south-west corner. Along each transect the presence or absence of overstorey canopy material was recorded at 1 m intervals. The recording method, after (Specht, 1970), uses a plastic tube which was attached to a 2 m length rod and contains an internal cross- hair. A mirror situated at the base of the tube at an angle of 45o then enables the operator to record the presence or absence of green leaves or wood (trunk or branches) in the canopy vertically above. Foliage-branch projected cover (FBC) and foliage projected cover (FPC) was then calculated as the sum of foliage and/or branch records as a proportion of the total observations. For the purposes of this study, foliage-branch cover relates to the amount of light that would reach the ground, and was the percentage of the plot area occupied by the vertical projection of foliage and branches, while foliage projected cover only considers light

interception by green foliage. Crown cover (CC) was defined as the percentage of the site within the vertical projection of the periphery of crowns, with crowns considered opaque (Walker et al., 1988; McDonald et al., 1998). As part of a subsequent 2004 field trip, an azimuthally independent estimation of foliage projective cover was tested on selected Queensland field plots. This was an improved Queensland Statewide Landcover and Trees Study (SLATS; QDNRM, 2003) field transect collection method, and involved laying out a 50 m transect on a bearing due north from the plot centre, with two other transect at 60o _{and 120}o

degrees respectively from the first transect. Canopy measures were then conducted as described previously (J. Armston, pers com).

Hemispherical photography was used to gather information on canopy openness and cover for the Injune study site. During the second field campaign to the Injune study area in 2004 to collect additional data for this thesis, hemispherical photographs were taken within 31 field plots at 10 m intervals along the three 50 m north-south transects previously established within each plot. Photos were collected using a Nikon D70 digital SLR 6.1 mega-pixel camera with a 10.5 mm full frame hemispherical prime lens with a tripod. Two photos were taken at each 10 m interval along the transects: one oriented north-south and the other east-west, which were subsequently merged for a single photo. Estimates of foliage cover for the fields plots were generated using Gap Light Analyzer (Version 2) software (Frazer et al., 1999). For comparisons with LiDAR cover estimates, two estimates were made. First, the single photo closest to the plot centre was used, and secondly, three photos from the centre of each transect. Appendix D describes the photographic calibration undertaken to generate estimates of foliage- branch cover from the hemispherical photography.