There is growing interest in airborne lidar for forest carbon accounting and precision forestry purposes. Airborne lidar systems offer a cost-effective, versatile, operationally flexible and robust sampling tool for forest managers. The objective of this study was to develop and test lidar canopy surface enhancement and segmentation processes for estimating dominant above-ground biomass (DAB) in a harvested eucalypt forest on the Central Coast of New South Wales (Australia). The Crown Infill, Trim and Smooth (CITS) process, incorporating a series of filters, algorithms, and selective multi-stage smoothing, was used to enhance lidar canopy surfaces prior to segmentation. Canopy segmentation was achieved using a vertical crown template approach termed the Spatially and Morphologically Isolated Crest (SMIC) process. SMIC delineates dominant tree crowns by detecting elevated crown crests within a 3D lidar canopy surface. Consolidated crown units constitute the basic sampling, analysis and reporting units for wall-to-wall forest inventory. The performance, sensitivity and limitations of these procedures were evaluated using a combination of simulated forest models and actual lidar forest data. Automated crown polygons were used as a sampling template to extract dominant tree height values which were converted to DAB estimates via height-to-biomass relationships derived from field survey and on-site destructive sampling. Results were compared with field based tree height and biomass estimates. Compared against a manually derived crown map from a 2ha field plot, canopy segmentation results revealed a producer???s accuracy of 76% and overall accuracy of 67%. Results indicated a trend toward greater crown splitting (fragmentation) as trees increase in age, height, stem diameter and crown size. Extracted dominant tree height values were highly correlated with ground survey height estimates (r2 0.95 for precision survey and r2 0.69 for standard survey). There was also no significant difference between SMIC and manual crown height estimates. SMIC units overestimated ground-based DAB by 5%; this increased to 36% with the inclusion of segmentation errors. However, SMIC estimation of total plot above-ground biomass (AGB) was within 9% of the ground-based estimate. Results are encouraging considering the mixed-species, multi-aged composition of the forest, and the combined effects of SMIC segmentation and lidar height errors.

Bibliography: leaves 280-300. / xix, 338 leaves : ill. (chiefly col.) ; 30 cm. / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / Thesis (Ph.D.)--University of Adelaide, Dept. of Horticulture, Viticulture and Oenology, 2000

There is growing interest in airborne lidar for forest carbon accounting and precision forestry purposes. Airborne lidar systems offer a cost-effective, versatile, operationally flexible and robust sampling tool for forest managers. The objective of this study was to develop and test lidar canopy surface enhancement and segmentation processes for estimating dominant above-ground biomass (DAB) in a harvested eucalypt forest on the Central Coast of New South Wales (Australia). The Crown Infill, Trim and Smooth (CITS) process, incorporating a series of filters, algorithms, and selective multi-stage smoothing, was used to enhance lidar canopy surfaces prior to segmentation. Canopy segmentation was achieved using a vertical crown template approach termed the Spatially and Morphologically Isolated Crest (SMIC) process. SMIC delineates dominant tree crowns by detecting elevated crown crests within a 3D lidar canopy surface. Consolidated crown units constitute the basic sampling, analysis and reporting units for wall-to-wall forest inventory. The performance, sensitivity and limitations of these procedures were evaluated using a combination of simulated forest models and actual lidar forest data. Automated crown polygons were used as a sampling template to extract dominant tree height values which were converted to DAB estimates via height-to-biomass relationships derived from field survey and on-site destructive sampling. Results were compared with field based tree height and biomass estimates. Compared against a manually derived crown map from a 2ha field plot, canopy segmentation results revealed a producer???s accuracy of 76% and overall accuracy of 67%. Results indicated a trend toward greater crown splitting (fragmentation) as trees increase in age, height, stem diameter and crown size. Extracted dominant tree height values were highly correlated with ground survey height estimates (r2 0.95 for precision survey and r2 0.69 for standard survey). There was also no significant difference between SMIC and manual crown height estimates. SMIC units overestimated ground-based DAB by 5%; this increased to 36% with the inclusion of segmentation errors. However, SMIC estimation of total plot above-ground biomass (AGB) was within 9% of the ground-based estimate. Results are encouraging considering the mixed-species, multi-aged composition of the forest, and the combined effects of SMIC segmentation and lidar height errors.

This thesis explores issues relating to the application of physiological-process models (???process models???) of forest growth to mixed species, mixed age forests, in particular the coastal blackbutt forests of New South Wales. Using a dataset provided by State Forests of New South Wales (Carter 1994 unpubl.) a numeric description of the forest was developed and stand-level parameters of interest were derived, in particular the plot by plot stemwood volume growth from 1975 to 1999. The amounts of harvested volume, volume that died and volume that grew into the measurement population were identified separately, and several different means of accounting for volume change over time were investigated. A method for quantifying the impact of harvesting and other silvicultural practices on the growth of the forest was developed and programs were written to convert the stand-level summary of the harvest impact into a semi-random selection of trees that would be ???harvested??? from the database under the set of silvicultural assumptions (Dore et al. 1999). A number of process models were investigated and reviewed before selecting one particular model, SUSTAIN (Dewar 1997) for adaption to these forests. This model is a relatively simple process model with a small number of input parameters. The model was adapted so that it could be used to compare the SUSTAIN estimate of growth with the growth of an individual stand of trees in the Kendall Forest Management Area, between Wauchope and Taree on the mid-north coast of NSW. To improve the accuracy of the prediction of growth by SUSTAIN, a method of re-setting the state of the stand to the actual condition at the time of remeasurement was developed. In addition, the SUSTAIN model was extended to enable two separate levels of canopy to be described and grown separately. Ultimately the model was only partially successful in mirroring the growth predicted by the empirical data. Its partial success is attributed primarily to the difficulties associated with correctly determining the allocation parameters used by the model to assign net photosynthate to the roots, foliage and stemwood. The nature of the change in allocation parameters when the forest stand is disturbed by harvest or fire needs further investigation.

This thesis explores issues relating to the application of physiological-process models (???process models???) of forest growth to mixed species, mixed age forests, in particular the coastal blackbutt forests of New South Wales. Using a dataset provided by State Forests of New South Wales (Carter 1994 unpubl.) a numeric description of the forest was developed and stand-level parameters of interest were derived, in particular the plot by plot stemwood volume growth from 1975 to 1999. The amounts of harvested volume, volume that died and volume that grew into the measurement population were identified separately, and several different means of accounting for volume change over time were investigated. A method for quantifying the impact of harvesting and other silvicultural practices on the growth of the forest was developed and programs were written to convert the stand-level summary of the harvest impact into a semi-random selection of trees that would be ???harvested??? from the database under the set of silvicultural assumptions (Dore et al. 1999). A number of process models were investigated and reviewed before selecting one particular model, SUSTAIN (Dewar 1997) for adaption to these forests. This model is a relatively simple process model with a small number of input parameters. The model was adapted so that it could be used to compare the SUSTAIN estimate of growth with the growth of an individual stand of trees in the Kendall Forest Management Area, between Wauchope and Taree on the mid-north coast of NSW. To improve the accuracy of the prediction of growth by SUSTAIN, a method of re-setting the state of the stand to the actual condition at the time of remeasurement was developed. In addition, the SUSTAIN model was extended to enable two separate levels of canopy to be described and grown separately. Ultimately the model was only partially successful in mirroring the growth predicted by the empirical data. Its partial success is attributed primarily to the difficulties associated with correctly determining the allocation parameters used by the model to assign net photosynthate to the roots, foliage and stemwood. The nature of the change in allocation parameters when the forest stand is disturbed by harvest or fire needs further investigation.