Abstract

Whether the value of the urban forest is defined by its socioeconomic benefits or its monetary value, paramount to the understanding of the importance of trees within urban environments is a quantitative understanding of how many trees there are and their spatial distribution throughout the community. City-wide assessments of urban tree canopy cover, e.g., McPherson et al., (2011), Jacobs et al. (2014) and S.J. Holt (2020), have become common foundations for national and local governments to measure, assess and increase the number of trees and green spaces within their government areas (Ordóñez et al., 2019). Accurate benchmarks and ongoing assessments of tree canopy cover and green spaces allow local government agencies to continuously review the performance of ongoing greening initiatives as well as ensuring that future policies are not only adequate, but achievable (Amati et al., 2017; Hurley, 2020)

Light Detection and Ranging (LIDAR) (Fig. 1) has become an industry standard remote sensing tool to map vegetation ecosystems (Coops et al., 2021) and quantify plant attributes and urban environments (Wang et al., 2019; Wang et al., 2020; Holt, 2020). Over the past year, Aerometrex has worked alongside urban environmental experts at the state and local government level to develop city-wide, LIDAR-derived tree canopy assessments, designed to provide a holistic understanding of the urban forest across large areas of interest and at a range of resolutions (council, suburb, property, or unit area) on both public and private land. Aerometrex urban forest data suite gives environmental experts and policy makers at all levels of government a critical understanding of the state of the urban forest at a given snapshot in time, while also providing them with the spatial and temporal context needed to value their green assets effectively and accurately.

Presented here are the results of three case-studies undertaken by Aerometrex utilising LIDAR to measure the characteristics and spatial distribution of the urban forest across large areas of interest:

  1. Tree Canopy Coverage Benchmark, Metropolitan Adelaide, 2018/19 – Undertaken on behalf of The Department of Environment and Water (DEW) and metropolitan local government climate change adaptation groups with the aim of producing a city-wide benchmark for tree canopy coverage across Metropolitan Adelaide.
  2. City-wide plantable space assessment, Metropolitan Adelaide, 2018/19 – Commissioned by Green Adelaide and DEW to identify priority areas in greatest need of greening investment and inform future project design to provide the greatest benefit to the community.
  • High resolution tree canopy change detection and classification, City of Unley, 2018-2021 – Carried out in collaboration with City of Unley to develop methodologies to quantitatively measure and accurately classify the changes in tree canopy between two epochs.

In combination, these case studies showcase the ability of LIDAR to quantify the urban forest and green infrastructure at the city scale, and measure how it is changing with time – all derived from a single data capture to provide the most robust yet cost-effective foundation possible for the valuation of green assets and measure their benefit to the community.

1.      Introduction

Increasing urban greenness, vegetation and urban tree canopy cover has become one of the most critical considerations for strategic planning within state and local government organisations (Amati et al., 2017). Urban trees and community green spaces have been identified as an important tool that can be used by policy-makers to mitigate the many negative environmental effects of urbanisation (Roy et al., 2012).

Numerous studies have shown that the presence of urban trees can provide socioeconomic value to the community through  numerous positive effects on the urban environment, such as economic benefits (Payton et al., 2008; Elmqvist et al., 2015; Donovan et al., 2019), positive effects on community health, well-being and safety (Kuo and Sullivan, 2001; McPherson et al., 2011; Kardan et al., 2015), improved air quality and storm water attenuation (Berland, 2012; Nowak et al., 2014; Selmi et al., 2016; Grey et al., 2018), combatting climate change through carbon sequestration (Brilli et al., 2019) and increasing community resilience to the effects of climate change by reducing the effects of urban heat islands (Tan et al., 2016; Rashid et al., 2020; Cheela et al., 2021). Translating the socioeconomic benefits of the urban forest into economic terms, which are often central to decision making processes and policy development, can be challenging (Rogers et al., 2017; Wang et al., 2018). Despite this, it can be advantageous to quantify the monetary value of trees as it enables a quantitative understanding of the balance of costs and benefits associated with green assets, leading to the effective integration of economic assessments into decision making processes at state and local governments (Jones and Davies, 2017; Song et al., 2018).

Whether the value of the urban forest is defined by its socioeconomic benefits or its monetary value, paramount to the understanding of the importance of trees within urban environments is a quantitative understanding of how many trees there are and their spatial distribution throughout the community. City-wide assessments of urban tree canopy cover, e.g., McPherson et al., (2011), Jacobs et al. (2014) and S.J. Holt (2020), have become common foundations for national and local governments to measure, assess and increase the number of trees and green spaces within their government areas (Ordóñez et al., 2019). Accurate benchmarks and ongoing assessments of tree canopy cover and green spaces allow local government agencies to continuously review the performance of ongoing greening initiatives as well as ensuring that future policies are not only adequate, but achievable (Amati et al., 2017; Hurley, 2020). For a city-wide assessment of the urban forest to yield actionable results, it is critical that it accurately accounts for trees on private as well as public land, as private land can account for over 50% of both urban tree cover and open space, thus, making a large contribution to the urban forest and it’s positive benefits (McPherson, 1998; Fuller and Gaston, 2009; Schmitt-Harsh et al., 2013; Klobucar et al., 2020).

Light Detection and Ranging (LIDAR) (Fig. 1) has become an industry standard remote sensing tool to map vegetation ecosystems (Coops et al., 2021) and quantify plant attributes in both native vegetation (Van Leeuwen and Nieuwenhuis, 2010; Li et al., 2015; Hagar et al., 2020), managed forests (Silva et al., 2016; Dalla Corte et al., 2020; Leite et al., 2020) and urban environments (Wang et al., 2019; Wang et al., 2020; Holt, 2020). LIDAR is often preferred to other passive and active remote sensing technologies (e.g. Synthetic Aperture Radar, SAR) for the purposes of measuring vegetation as LIDAR provides greater sensitivity to changes in vegetation structure (Dong and Chen, 2018b). LIDAR-derived metrics can also provide a more accurate description of the canopy coverage and be more sensitive to small-scale changes than random point classification methods such as iTree (Parmehr et al., 2016). Furthermore, LIDAR can provide greater positional accuracy and three dimensional detail compared to AI-derived canopy maps (e.g. Abdollahi and Pradhan, (2021)) as LIDAR measures the vegetation explicitly in three dimensions, free of perspective effects (Dong and Chen, 2018a).

Presented here are the results of three case-studies undertaken by Aerometrex utilising LIDAR to measure the characteristics and spatial distribution of the urban forest across large areas of interest:
Over the past year, Aerometrex has worked alongside urban environmental experts at the state and local government level to develop city-wide, LIDAR-derived tree canopy assessments, designed to provide a holistic understanding of the urban forest across large areas of interest and at a range of resolutions (council, suburb, property, or unit area) on both public and private land. Aerometrex urban forest data suite gives environmental experts and policy makers at all levels of government a critical understanding of the state of the urban forest at a given snapshot in time, while also providing them with the spatial and temporal context needed to value their green assets effectively and accurately.

  1. Tree Canopy Coverage Benchmark, Metropolitan Adelaide, 2018/19 – Undertaken on behalf of The Department of Environment and Water (DEW) and metropolitan local government climate change adaptation groups with the aim of producing a city-wide benchmark for tree canopy coverage across Metropolitan Adelaide.
  2. City-wide plantable space assessment, Metropolitan Adelaide, 2018/19 – Commissioned by Green Adelaide and DEW to identify priority areas in greatest need of greening investment and inform future project design to provide the greatest benefit to the community.
  3. High resolution tree canopy change detection and classification, City of Unley, 2018-2021 – Carried out in collaboration with City of Unley to develop methodologies to quantitatively measure and accurately classify the changes in tree canopy between two epochs.

In combination, these case studies showcase the ability of LIDAR to quantify the urban forest and green infrastructure at the city scale, and measure how it is changing with time – all derived from a single data capture to provide the most robust yet cost-effective foundation possible for the valuation of green assets and measure their benefit to the community.

2.      Project Datasets

All three case studies presented here utilize LIDAR data captured using a RIEGL VQ-780i sensor, with a vertical accuracy of 10cm and classified to Aerometrex’s Type 2 standard (Fig. 2), which includes manual refinement of the ground, building and high vegetation classes (ASPRS, 2019). The 2018 LIDAR survey used in all three case studies was captured in April 2018 and was expanded to include full coverages of sixteen metropolitan councils in November 2019. In total, the dataset covers 1,706 km2 of metropolitan Adelaide (Fig. 3). The third case study incorporates a subsequent LIDAR capture in April 2021 across City of Unley (14.5km2) and is compared to the 2018 data across the same area to quantify change detection. Both datasets have a minimum point density of – 8-12 pts.m-2. For all projects, following consultation with council representatives, a tree was defined as any vegetation above three metres in height, as it is above this height that the tree begins to provide shading.

3.      Case Study 1 – Tree Canopy Coverage Benchmark, Metropolitan Adelaide, 2018/19

3.1       Project Summary

In 2019 Aerometrex was commissioned by DEW and four metropolitan local government climate change adaptation groups (representing sixteen individual councils) to produce a quantitative tree canopy coverage benchmark across metropolitan Adelaide. The core aim of this project was to provide all participating LGAs and the State government with a quantitative, spatially explicit benchmark for tree canopy coverage in 2018/19. This benchmark would then form the foundation for the assessment of the effectiveness of greening initiatives across the city in coming years. As part of this project Aerometrex developed a suite of LIDAR-derived urban forest datasets (Fig. 4) to provide each participating council with not only a tree canopy coverage benchmark, but also an understanding of the vertical distribution of tree canopy (Canopy Height Model & Stratification Maps), the distribution of tree canopy across land ownership and land use types as well as a measure of the spatial distribution of tree canopy coverage across the city, independent of biases associated with measuring at the per council or per suburb level (Tree Canopy by Unit Area). The full set of results and associated graphical reports can be viewed at DEW’s Urban Heat and Tree Mapping viewer (Enviro Data SA, 2021).

3.2       Methodology

The geoprocessing methodology used to extract all tree canopy metrics from the LIDAR point cloud were based on those presented in Dong and Chen, (2018b) and Holt (2019). Land Use and Land Ownership statistics were generated using parcel definitions provided by Department of Planning, Transport and Infrastructure.

3.3       Results

Tree Canopy Height

The city-wide Canopy Height Model (CHM) for trees above three metres in height revealed that the average tree canopy height across metropolitan Adelaide is 10.1 m with a standard deviation of 5.5 m (minimum height of three metres). The highest non-native tree canopy can be found in Bone Gully Forest, Kangarilla and the tallest native vegetation can be found in the Boulevard of Gums, Belair National Park, Mitcham (Fig. 5), which are 48.3m and 48.2m above ground respectively. Canopy Stratification Maps for all sixteen participating LGAs show that tree canopy between three and ten metres in height constitutes the majority of tree canopy cover across the city (Fig. 5).

Tree Canopy Coverage

The overall tree canopy cover for the metropolitan Adelaide area (as defined by (Holt, 2020)) for 2018/19 was 23.37% (Fig. 6). The tree canopy coverage for local government areas wholly within the LIDAR capture area ranges from 48.81% in City of Mitcham to 9.89% in the City of Port Adelaide Enfield. The average tree canopy cover across all participating LGAs was 23.16%. The tree canopy cover for suburbs wholly within the LIDAR capture area ranges from 71.8% in Waterfall Gully to 1.7% at Parafield. When measured by unit area, the average tree canopy coverage across metropolitan Adelaide is 22.7% within a standard deviation of 23.0%. For twelve out of the fourteen LGAs within the survey area, the majority of tree canopy covers residential land use areas and private land ownership areas. In some cases, as much as 50% of the tree canopy exists on residential or private, e.g. City of Unley, Town of Walkerville and City of Onkaparinga.

3.4       Key Project Outcomes

This project provided accurate, spatially explicit LIDAR-derived tree canopy coverage benchmarks for metropolitan Adelaide (23.37%) as well as individual benchmarks for all sixteen participating LGAs. The datasets provide both the State and local governments in South Australia with a nationally leading, spatially explicit benchmark that can be used to track the change in tree canopy coverage in years to come and assess the effectiveness of ongoing greening initiatives. When assessing the tree canopy cover by LGA extent or suburb extent, the true distribution of tree canopy can be hidden due to areas associated with coarse sampling domains (Dark and Bram, 2007). Added to this, the tree canopy cover of any given LGA extent is biased by the land uses within it. For example, the tree canopy coverage in the City of Mitcham is positively biased by abundant national parks and reserves and the City of Onkaparinga’s is negatively biased by arable agriculture. Canopy Cover by Unit Area provides an unbiased representation of the tree canopy cover and a robust description of the spatial distribution of tree canopy cover across the city, also allowing for a better understanding of the representative tree canopy coverage of denser urban areas, in this case 0-20%. The tree canopy by land use and land ownership results provide quantitative evidence of one of the key challenges facing LGAs: that despite being tasked with increasing tree canopy cover to meet greening targets, in many cases they do not control enough of the tree canopy to meet those targets. This highlights the critical role that private landowners play in increasing the tree canopy cover across the city and the importance of initiatives that target public awareness of urban greening initiatives (Ordóñez-Barona et al., 2021).

4.   Case Study 2 – City-wide plantable space assessment, Metropolitan      Adelaide, 2018/19

4.1       Project Summary

The primary objective of the project was to produce a 50cm resolution, fully semantic classification of permeable and impermeable surfaces across 1,305 km2 of metropolitan Adelaide (Holt, 2021). Green Adelaide and DEW funded the project to allow experts to use the permeable and impermeable surface data to form a detailed understanding of the amount and spatial distribution of plantable space (permeable ground surfaces), assess the progress and feasibility of current greening initiatives, and develop better informed targets and policies for the coming years. When combined with the 2018/19 tree canopy benchmark, these datasets can be used to not only identify areas in greatest need of greening investment, but also identify areas with the greatest greening potential and inform the development of long term greening policies (Wu et al., 2008; Bravo-Bello et al., 2020). Extending the classification into a full semantic classification of the urban environment meant that the final output could also be applied to detail surface water run-off and flood modelling during high rainfall events (Sugg et al., 2014; Sarkar Chaudhuri et al., 2017; Hung et al., 2018).

The final project output, consisted of a 50cm resolution semantic classification of the urban environment, comprised of five classification classes (developed in collaboration with government end-users):

  1. Impermeable – impermeable ground surfaces such as pavement, concrete and roads.
  2. Permeable/Plantable Space – permeable ground surfaces including bare soil and grass.
  3. Impermeable Above-ground Object – Above ground features that are deemed to be impermeable in nature including buildings, tree canopy overhanging roads and other infrastructure between 0.25-3m in height with a relative NDVI below 0.15.
  4. Permeable Above Ground Object – Above ground features that are deemed to likely be permeable in nature including trees and vegetation between 0.25-3m in height with a relative NDVI above 0.15.
  5. Water body – Including the ocean, lakes, rivers, dams and large pools.

4.2       Methodology

Figure 7 outlines the methodology used to produce the permeable vs impermeable surface classification. The project utilised two key datasets, a set of LIDAR-derived raster inputs including a normalized digital surface model (nDSM) and intensity map, as well as four band multispectral imagery (R:G:B:NIR) captured near simultaneously to the 2018/19 LIDAR dataset. These datasets were combined with secondary, vector feature datasets used to refine the classification, including LIDAR-derived building footprints, tree canopy cover and waterbodies and AI-derived road extents. The semantic classification was carried out in two stages. Stage One (Fig. 7) consisted of classifying ground surfaces as either permeable or impermeable using an end-to-end trained Feature Pyramid Network (FPN) (Lin et al., 2017). Stage Two utilized the secondary vector feature datasets in a rule-based, Object Based Image Analysis workflow (Blaschke, 2010) to produce a refined sematic classification (Fig. 7).

4.3       Results

Classification Accuracy

The performance and accuracy of the FPN classifier was assessed using a manually labeled validation dataset which consisted of 10 separate tiles, each with an area of 1 km2. Ground and above ground objects within each validation tile were manually classified using an Object Based Image Analysis workflow, utilising our high-resolution imagery as a reference (spatial resolution of 7.5cm). In total, over 27 million pixels were manually labeled, corresponding to an approximate area of 7km2. A direct comparison between the classification of each pixel in the validation dataset and classified dataset was used to assess the performance of the FPN classifier. It is important to note that the data used for validation was not part of the data used to train the FPN classifier. The FPN-derived ground classification has an overall accuracy of 89.45% and an overall Kappa Index of Agreement of 0.78 (Excellent; Richards, (1999)).

Permeable vs Impermeable Surface Distribution

Figure 8 shows an example of the output classification, as well as the distribution of permeable and impermeable surfaces within the 2018 LIDAR survey extent, which covers the majority of Adelaide’s urban areas and corresponds most closely with the capture date of the aerial imagery used in Stage 1 of the classification. 37% of Adelaide’s urban area is permeable ground (plantable space) and 19.8% is impermeable surface (e.g. roads and pavement), and when combined with above ground impermeable surfaces (e.g. buildings), 44% of Adelaide’s urban area is impermeable and would contribute to surface water accumulation. At the suburb scale (Fig. 9), the proportion of total suburb area that is plantable space, ranges from 78.0%% in Parafield (3,459,041 m2 plantable space) to 6.4% in Mile End South (42,671 m2 plantable space), with an average of 27% across all suburbs. The areas with the lowest proportions of plantable space are concentrated within a ~4.5 km radius of the Adelaide CBD (Fig. 9). Within the 2018 LIDAR survey area, the three land ownership categories that contribute the most to the total amount of plantable space are Private land (30.8%), State Government land (20.0%) and Local Government land (16.2%).

4.4       Key Project Outcomes

Mapping the amount and spatial distribution of plantable space across the most heavily urbanized areas of metropolitan Adelaide provides much needed spatial and temporal context to the 2018/19 tree canopy benchmark and the value of the urban forest in certain areas.

By directly comparing the amount of tree canopy coverage within a suburb, as well as the proportion that suburb that is plantable, it is possible to understand the ‘greening potential’ of that suburb as well as the relative importance of preserving existing trees. Figure 10 shows a bivariate plot of the proportion of plantable space versus the overall tree canopy cover for all suburbs wholly within the 2018 LIDAR survey area. Each of the four regions define the vulnerability and potential of the urban forest in each suburb. ‘Ideal’ suburbs have high tree canopy cover and high proportion of plantable space, meaning lots of existing trees and lots of space to plant more. ‘High Greening Potential’ suburbs have low tree canopy cover and high plantable space, i.e. few trees but ample space to plant more given effective planting strategies. ‘At Risk’ suburbs have high tree canopy cover, but very low proportion of plantable space, meaning that the existing trees need to be protected as there is little plantable space to replace trees lost to urban infill. Finally, ‘Code Red’ have low tree canopy cover and low plantable space. 76 suburbs are ideal, 151 suburbs have high greening potential, 44 suburbs are at risk and 38 suburbs are code red. It is these 38 suburbs that should be targeted in greening initiatives and policy development/reform as they have very few trees and very little space to plant more. Whatever the vulnerability or potential of the urban forest, private landowners will play a central role in the success of any planting initiatives given that the majority of plantable space resides on private land.

5.       Case Study 3 – High resolution tree canopy change detection and    classification, City of Unley, 2018-2021

5.1       Project Summary

City of Unley commissioned Aerometrex to capture LIDAR across the LGA in April 2021 to produce a full urban forest assessment and provide an updated tree canopy coverage. This could be compared to the 2018 benchmark to quantitatively the measure the change in tree canopy cover over three years. A detailed investigation into the types of changes that were occurring to the tree canopy was used to gain a better understanding of what processes were contributing most to measured tree canopy gain and/or loss.

5.2       Methodology

All analyses used within the 2018/19 benchmark study were repeated, with the same input parameters and resolutions in order to produce datasets that were directly comparable to previous results. Additionally, Aerometrex utilised newly developed image segmentation and OBIA workflows to produce high-resolution (10cm) tree canopy coverage maps, for both the new 2021 LIDAR data and the historical 2018 dataset, to classify the changes occurring to the tree canopy across City of Unley. Isolated occurrences of tree canopy loss and gain were classified as Tree Planting and Tree Removal respectively. Areas of gain and loss greater than 4m2 and joined to an area of no-change (pre-existing tree canopy), were classified as Tree Growth and Pruning Reduction respectively. Areas of change less that 4m2 in area and joined to pre-existing canopy were left unclassified to remove noise associated with tree movement between and during captures. High resolution aerial imagery was used as a validation dataset and to provide an understanding of the changes in the urban environment that were driving tree canopy loss and gain. The spatial distribution of tree canopy losses and gains were assessed using standard and weighted hotspot analyses based on the Getis-Ord Gi* statistic (Ord and Getis, 1995; Getis and Ord, 2010).

5.3       Results

City of Unley Tree Canopy Cover

The repeat analysis of the urban forest across City of Unley showed that the average change in canopy height was 0.4m, the overall tree canopy coverage (measured at 1m resolution) has increased by 1.37% to 27.99% since 2018 and the largest increase by area in tree canopy coverage occurred in canopy between 3m to 10m in height.

As with the 2018 study, Residential and Private land contribute most to tree canopy cover. When measured at 10cm, the overall tree canopy coverage is refined by 4-5% for each epoch, 21.66% and 23.05% for 2018 and 2021 respectively, but critically the tree canopy change is +1.39%, only a 0.02% difference when compared to the 1m resolution dataset.

Tree Canopy Change Classification

Very high-resolution tree canopy maps for both epochs provided a robust comparison for the classification of tree canopy change (Fig. 11). Of the total gain in tree canopy cover, existing tree growth contributed five times the amount of new canopy (306,142 m2) than newly planted trees reaching 3m in height (62,109 m2), and solely offsets the total tree canopy loss, both by pruning reduction and tree removal (172,673 m2 and 127,244 m2 respectively). The average contribution of new tree canopy by a newly planted tree reaching 3m in height is 3.8 m2 and the total area of increase due to tree planting is the equivalent of 16,344 new trees at 3m in height. The total contribution of existing tree growth and total canopy loss is the equivalent of 80,563 and 78,925 new trees at 3m in height respectively.

5.4       Key Project Outcomes

The effect of data resolution on greening target assessment


Implications on Urban Forest management in City of Unley

Comparisons between the overall tree canopy cover and change in tree canopy cover measured at two different resolutions (1m and 10cm) show that the value for the overall tree canopy coverage can be 4-5% larger when measured at coarser resolutions as compared to fine resolutions, however the change in tree canopy cover is constant, independent of resolution. These results indicate that the precise wording used when defining an LGA’s greening target can have an effect on what resolution data is appropriate to assess that target. For example, greening targets such as “reaching 30% tree canopy cover by the year 2030” require high-resolution datasets which provide precise, refined tree canopy coverages. In comparison, greening targets such as “increasing tree canopy cover by 30% by the year 2030” can be assessed using coarser resolution datasets as the change over time is accurately measured regardless of resolution.

The large-scale contribution of new tree canopy cover due to the growth of existing trees, as compared to newly planted trees, suggests that preserving and protecting existing trees is of critical importance alongside planting initiatives if City of Unley is going to continue to increase the amount of tree canopy coverage in the future.

The average area of tree canopy lost per tree removal is 11.8 m2, three times the average contribution of new canopy due to newly planted trees. This highlights the importance of protecting existing trees, because as the contribution of existing tree growth is reduced through tree removal, losses in tree canopy cover will rapidly outpace the contribution of newly planted trees. Weighted Hot Spot Analysis shows that the four most statistically significant clusters of large magnitude canopy loss occur at sites of urban development (Fig. 12). This would suggest that urban development and associated urban infill pose the greatest threat to the urban forest within the City of Unley.

6. Conclusions

The three case-studies presented here showcase the power of geospatial data such as Airborne LIDAR to produce city-wide urban forest assessments which quantify the amount and spatial distribution of both tree canopy cover and plantable space, as well as explicitly measuring and classifying tree canopy change with time, across both private and public land. City-wide LIDAR-derived tree canopy and plantable space studies provide critical spatial and temporal context to the process of valuing the urban forest, whether that value is defined by monetary value or socioeconomic benefits. Large-scale urban forest assessments will play an increasingly important role in helping environmental management experts and policy makers at all levels of government build greener, cooler, more sustainable cities and increase the community’s resilience to the ongoing negative effects of climate change.

The results of each of these analyses highlight two critical considerations for environmental management experts across Adelaide:

  1. Private landowners will play a critical role in helping build a greener, more sustainable future for Adelaide as the overwhelming majority of both existing tree canopy cover and available plantable space resides on privately owned land.
  2. Preliminary tree canopy change detection and classification at the LGA level over the last three years following the 2018/19 benchmark suggests that growth of existing trees contributes a much larger amount of new tree canopy than newly planted trees. Given the time lag between planting and major tree canopy contribution, it is critical that local governments preserve existing tree canopy cover by protecting existing trees from urban developments and infill.

Acknowledgements

The author would like to thank the following people and organisations for their invaluable contributions to, and support of, the ongoing development of datasets designed to build greener, cooler, more sustainable communities – Ellie Zoghi, Fabrice Marre, Alex Rixon, Blair Pellegrino, James Cameron, Kat Ryan, Ben Leonello, Bec Taylor, The Department for Environment and Water, Green Adelaide, City of Unley, Adapt West, Resilient East, Resilient South and Adapting Northern Adelaide.

References

Abdollahi, A., and Pradhan, B., 2021, Urban Vegetation Mapping from Aerial Imagery Using Explainable AI (XAI): Sensors , v. 21, doi:10.3390/s21144738.

Amati, M., Boruff, B., and Caccetta, P., 2017, Where should all the trees go? Investigatin g the impact of tree canopy cover on socio-economic status and wellbeing: Melbourne: Innovation Australia Ltd.

ASPRS, 2019, LAS Specification 1.4.:

Berland, A., 2012, Long-term urbanization effects on tree canopy cover along an urban–rural gradient: Urban Ecosystems, v. 15, p. 721–738, doi:10.1007/s11252-012-0224-9.

Blaschke, T., 2010, Object based image analysis for remote sensing: ISPRS Journal of Photogrammetry and Remote Sensing, v. 65, p. 2–16, doi:https://doi.org/10.1016/j.isprsjprs.2009.06.004.

Bravo-Bello, J.C., Martinez-Trinidad, T., Valdez-Lazalde, J.R., Romero-Sanchez, M.E., and Martinez-Trinidad, S., 2020, Analyzing Potential Tree-Planting Sites and Tree Coverage in Mexico City Using Satellite Imagery: Forests , v. 11, doi:10.3390/f11040423.

Brilli, L., Chiesi, M., Brogi, C., Magno, R., Arcidiaco, L., Bottai, L., Tagliaferri, G., Bindi, M., and Maselli, F., 2019, Combination of ground and remote sensing data to assess carbon stock changes in the main urban park of Florence: Urban Forestry & Urban Greening, v. 43, p. 126377, doi:https://doi.org/10.1016/j.ufug.2019.126377.

Cheela, V.R.S., John, M., Biswas, W., and Sarker, P., 2021, Combating Urban Heat Island Effect—A Review of Reflective Pavements and Tree Shading Strategies: Buildings , v. 11, doi:10.3390/buildings11030093.

Coops, N.C., Tompalski, P., Goodbody, T.R.H., Queinnec, M., Luther, J.E., Bolton, D.K., White, J.C., Wulder, M.A., van Lier, O.R., and Hermosilla, T., 2021, Modelling lidar-derived estimates of forest attributes over space and time: A review of approaches and future trends: Remote Sensing of Environment, v. 260, p. 112477, doi:https://doi.org/10.1016/j.rse.2021.112477.

Dalla Corte, A.P., Souza, D.V., Rex, F.E., Sanquetta, C.R., Mohan, M., Silva, C.A., Zambrano, A.M.A., Prata, G., de Almeida, D.R.A., and Trautenmüller, J.W., 2020, Forest inventory with high-density UAV-Lidar: Machine learning approaches for predicting individual tree attributes: Computers and Electronics in Agriculture, v. 179, p. 105815.

Dark, S.J., and Bram, D., 2007, The modifiable areal unit problem (MAUP) in physical geography: Progress in Physical Geography: Earth and Environment, v. 31, p. 471–479, doi:10.1177/0309133307083294.

Dong, P., and Chen, Q., 2018a, Brief Overview of Remote Sensing, in Weng, Q. ed., LIDAR Remote Sensing and Applications, London, CRC Press, p. 1–13.

Dong, P., and Chen, Q., 2018b, LIDAR for Forestry Applications, in Weng, Q. ed., LIDAR Remote Sensing and Applications, London, CRC Press, p. 63–105.

Donovan, G.H., Landry, S., and Winter, C., 2019, Urban trees, house price, and redevelopment pressure in Tampa, Florida: Urban Forestry & Urban Greening, v. 38, p. 330–336, doi:https://doi.org/10.1016/j.ufug.2019.01.014.

Elmqvist, T., Setälä, H., Handel, S.N., van der Ploeg, S., Aronson, J., Blignaut, J.N., Gómez-Baggethun, E., Nowak, D.J., Kronenberg, J., and de Groot, R., 2015, Benefits of restoring ecosystem services in urban areas: Current Opinion in Environmental Sustainability, v. 14, p. 101–108, doi:https://doi.org/10.1016/j.cosust.2015.05.001.

Enviro Data SA, 2021, Urban Heat and Tree Mapping of Adelaide Metropolitan Area:, https://data.environment.sa.gov.au/Climate/Data-Systems/Urban-Heat-Mapping/Pages/default.aspx (accessed July 2021).

Fuller, R.A., and Gaston, K.J., 2009, The scaling of green space coverage in European cities: Biology letters, v. 5, p. 352–355.

Getis, A., and Ord, J.K., 2010, The Analysis of Spatial Association by Use of Distance Statistics BT  – Perspectives on Spatial Data Analysis, in Anselin, L. and Rey, S.J. eds., Berlin, Heidelberg, Springer Berlin Heidelberg, p. 127–145, doi:10.1007/978-3-642-01976-0_10.

Grey, V., Livesley, S.J., Fletcher, T.D., and Szota, C., 2018, Tree pits to help mitigate runoff in dense urban areas: Journal of Hydrology, v. 565, p. 400–410, doi:https://doi.org/10.1016/j.jhydrol.2018.08.038.

Hagar, J.C., Yost, A., and Haggerty, P.K., 2020, Incorporating LiDAR metrics into a structure-based habitat model for a canopy-dwelling species: Remote Sensing of Environment, v. 236, p. 111499.

Holt, S.J., 2020, LIDAR derived tree canopy coverage metrics across Adelaide, South Australia. Report 2: Metropolitan Adelaide.:

Holt, S.J., 2021, Permeable and Impermeable Ground Surface Classification Adelaide, South Australia.:

Holt, S.J., 2019, Quantifying first order urban tree attributes within The City of Mitcham Council, Adelaide using airborne LiDAR and aerial photography.:

Hung, C.-L.J., James, L.A., and Hodgson, M.E., 2018, An automated algorithm for mapping building impervious areas from airborne LiDAR point-cloud data for flood hydrology: GIScience & Remote Sensing, v. 55, p. 793–816, doi:10.1080/15481603.2018.1452588.

Hurley, J., 2020, Where will all the trees be?-an assessment of urban forest cover and management for Australian cities:

Jacobs, B., Mikhailovich, N., and Delaney, C., 2014, Benchmarking Australia’s Urban Tree Canopy: An i-Tree assessment, Final Report: Institute for Sustainable Futures, UTS.

Jones, N., and Davies, C., 2017, Linking the Environmental, Social and Economic Aspects of Urban Forestry and Green Infrastructure BT  – The Urban Forest: Cultivating Green Infrastructure for People and the Environment, in Pearlmutter, D., Calfapietra, C., Samson, R., O’Brien, L., Krajter Ostoić, S., Sanesi, G., and Alonso del Amo, R. eds., Cham, Springer International Publishing, p. 305–313, doi:10.1007/978-3-319-50280-9_23.

Kardan, O., Gozdyra, P., Misic, B., Moola, F., Palmer, L.J., Paus, T., and Berman, M.G., 2015, Neighborhood greenspace and health in a large urban center: Scientific Reports, v. 5, p. 11610, doi:10.1038/srep11610.

Klobucar, B., Östberg, J., Jansson, M., and Randrup, T.B., 2020, Long-Term Validation and Governance Role in Contemporary Urban Tree Monitoring: A Review: Sustainability , v. 12, doi:10.3390/su12145589.

Kuo, F.E., and Sullivan, W.C., 2001, Environment and Crime in the Inner City: Does Vegetation Reduce Crime? Environment and Behavior, v. 33, p. 343–367, doi:10.1177/0013916501333002.

Van Leeuwen, M., and Nieuwenhuis, M., 2010, Retrieval of forest structural parameters using LiDAR remote sensing: European Journal of Forest Research, v. 129, p. 749–770.

Leite, R. V et al., 2020, Individual Tree Attribute Estimation and Uniformity Assessment in Fast-Growing Eucalyptus spp. Forest Plantations Using Lidar and Linear Mixed-Effects Models: Remote Sensing , v. 12, doi:10.3390/rs12213599.

Li, A., Glenn, N.F., Olsoy, P.J., Mitchell, J.J., and Shrestha, R., 2015, Aboveground biomass estimates of sagebrush using terrestrial and airborne LiDAR data in a dryland ecosystem: Agricultural and Forest Meteorology, v. 213, p. 138–147.

Lin, T.-Y., Dollár, P., Girshick, R., He, K., Hariharan, B., and Belongie, S., 2017, Feature pyramid networks for object detection, in Proceedings of the IEEE conference on computer vision and pattern recognition, p. 2117–2125.

McPherson, E.G., 1998, Structure and sustainability of Sacramento’s urban forest:

McPherson, E.G., Simpson, J.R., Xiao, Q., and Wu, C., 2011, Million trees Los Angeles canopy cover and benefit assessment: Landscape and Urban Planning, v. 99, p. 40–50, doi:https://doi.org/10.1016/j.landurbplan.2010.08.011.

Nowak, D.J., Hirabayashi, S., Bodine, A., and Greenfield, E., 2014, Tree and forest effects on air quality and human health in the United States: Environmental Pollution, v. 193, p. 119–129, doi:https://doi.org/10.1016/j.envpol.2014.05.028.

Ord, J.K., and Getis, A., 1995, Local Spatial Autocorrelation Statistics: Distributional Issues and an Application: Geographical Analysis, v. 27, p. 286–306, doi:https://doi.org/10.1111/j.1538-4632.1995.tb00912.x.

Ordóñez-Barona, C. et al., 2021, International approaches to protecting and retaining trees on private urban land: Journal of Environmental Management, v. 285, p. 112081, doi:https://doi.org/10.1016/j.jenvman.2021.112081.

Parmehr, E.G., Amati, M., Taylor, E.J., and Livesley, S.J., 2016, Estimation of urban tree canopy cover using random point sampling and remote sensing methods: Urban Forestry & Urban Greening, v. 20, p. 160–171, doi:https://doi.org/10.1016/j.ufug.2016.08.011.

Payton, S., Lindsey, G., Wilson, J., Ottensmann, J.R., and Man, J., 2008, Valuing the benefits of the urban forest: a spatial hedonic approach: Journal of Environmental Planning and Management, v. 51, p. 717–736, doi:10.1080/09640560802423509.

Rashid, M., Hilton, J., Khan, N., Sutherland, D., and Moinuddin, K., 2020, A Report on WRF Software Development (Preliminary).:

Richards, J.A., 1999, Image Classification in Practice, in Remote sensing digital image analysis, Springer, p. 381–433.

Rogers, K., Andreucci, M.-B., Jones, N., Japelj, A., and Vranic, P., 2017, The Value of Valuing: Recognising the Benefits of the Urban Forest BT  – The Urban Forest: Cultivating Green Infrastructure for People and the Environment, in Pearlmutter, D., Calfapietra, C., Samson, R., O’Brien, L., Krajter Ostoić, S., Sanesi, G., and Alonso del Amo, R. eds., Cham, Springer International Publishing, p. 283–299, doi:10.1007/978-3-319-50280-9_21.

Roy, S., Byrne, J., and Pickering, C., 2012, A systematic quantitative review of urban tree benefits, costs, and assessment methods across cities in different climatic zones: Urban Forestry & Urban Greening, v. 11, p. 351–363, doi:https://doi.org/10.1016/j.ufug.2012.06.006.

Sarkar Chaudhuri, A., Singh, P., and Rai, S.C., 2017, Assessment of impervious surface growth in urban environment through remote sensing estimates: Environmental Earth Sciences, v. 76, p. 541, doi:10.1007/s12665-017-6877-1.

Schmitt-Harsh, M., Mincey, S.K., Patterson, M., Fischer, B.C., and Evans, T.P., 2013, Private residential urban forest structure and carbon storage in a moderate-sized urban area in the Midwest, United States: Urban Forestry & Urban Greening, v. 12, p. 454–463, doi:https://doi.org/10.1016/j.ufug.2013.07.007.

Selmi, W., Weber, C., Rivière, E., Blond, N., Mehdi, L., and Nowak, D., 2016, Air pollution removal by trees in public green spaces in Strasbourg city, France: Urban Forestry & Urban Greening, v. 17, p. 192–201, doi:https://doi.org/10.1016/j.ufug.2016.04.010.

Silva, C.A., Klauberg, C., Hudak, A.T., Vierling, L.A., Liesenberg, V., Carvalho, S.P.C. e, and Rodriguez, L.C.E., 2016, A principal component approach for predicting the stem volume in Eucalyptus plantations in Brazil using airborne LiDAR data: Forestry: An International Journal of Forest Research, v. 89, p. 422–433, doi:10.1093/forestry/cpw016.

Song, X.P., Tan, P.Y., Edwards, P., and Richards, D., 2018, The economic benefits and costs of trees in urban forest stewardship: A systematic review: Urban Forestry & Urban Greening, v. 29, p. 162–170, doi:https://doi.org/10.1016/j.ufug.2017.11.017.

Sugg, Z.P., Finke, T., Goodrich, D.C., Moran, M.S., and Yool, S.R., 2014, Mapping impervious surfaces using object-oriented classification in a semiarid urban region: Photogrammetric Engineering & Remote Sensing, v. 80, p. 343–352.

Tan, Z., Lau, K.K.-L., and Ng, E., 2016, Urban tree design approaches for mitigating daytime urban heat island effects in a high-density urban environment: Energy and Buildings, v. 114, p. 265–274, doi:https://doi.org/10.1016/j.enbuild.2015.06.031.

Wang, V., Gao, J., and Schwendenmann, L., 2020, Assessing changes of urban vegetation cover and aboveground carbon stocks using LiDAR and Landsat imagery data in Auckland, New Zealand: International Journal of Remote Sensing, v. 41, p. 2140–2158.

Wang, E., Kang, N., and Yu, Y., 2018, Valuing Urban Landscape Using Subjective Well-Being Data: Empirical Evidence from Dalian, China: Sustainability , v. 10, doi:10.3390/su10010036.

Wang, K., Wang, T., and Liu, X., 2019, A Review: Individual Tree Species Classification Using Integrated Airborne LiDAR and Optical Imagery with a Focus on the Urban Environment: Forests , v. 10, doi:10.3390/f10010001.

Wu, C., Xiao, Q., and McPherson, E.G., 2008, A method for locating potential tree-planting sites in urban areas: A case study of Los Angeles, USA: Urban Forestry & Urban Greening, v. 7, p. 65–76, doi:https://doi.org/10.1016/j.ufug.2008.01.002.