Abstract
With a growing number of urban tree inventory methods and diversifying use of tree inventory data by city authorities and researchers, there is a need to evaluate, review, and critically assess the inventory methods available. This study reviewed studies using urban tree inventories at single-tree level as their data source. Based on this, a bibliographic overview was established and a typology of contemporary urban tree inventory methods was created and used as a framework for evaluation and discussion of the measurement type and accuracy achievable with different methods. The authors found that data from urban tree inventories are currently being employed in research with an increasing number of focuses across a geographical scope that spans all continents except Africa. Four main types of urban tree inventories were distinguished: satellite-supported methods, airplane-supported methods, on-the-ground scanning or digital photography, and field surveys. Compiling results across studies and evaluating the parameters collected by these inventory methods and their accuracy of measurement revealed that the technology itself and current data processing methods limit the reliability of the data obtained from all methods except field surveys. The study authors recommend further technological development and scientific testing before these methods can replace field surveys.
In recent decades there has been increased interest in urban tree inventories, resulting from growing problems with pest and disease attack on the urban tree stock (e.g., Raupp et al. 2006) and growing awareness among decision-makers of the multiple ecosystem services trees provide in the cityscape (Hubacek and Kronenberg 2013). The surge of interest in urban tree inventories has been supported by rapid methodological and technological developments (e.g., i-Tree 2012). City authorities, especially in North America and Europe, have increasingly started to perform tree inventories (e.g., Keller and Konijnendijk 2012; Sjöman 2012). Parallel to this, inventories have become valuable data sources for researchers studying the environmental, social, and economic services provided by urban trees. Scientific uses of data from urban tree inventories include studies of storm-damaged trees and identification of species and dimensions most affected (Jim and Liu 1997); risk assessments (Mattheck and Breloer 1994; Lonsdale 1999); identification of species composition and diversity (Raupp et al. 2006; Sjöman et al. 2012); modeling of local climate (Nowak et al. 2001; Yokohari et al 2001; Dimoudi and Nikolopoulou 2003; Nowak et al. 2006a); impacts on air pollution, urban heat island effects, and stormwater runoff (McPherson et al. 1997); assessment of the economic benefit of urban trees (Maco and McPherson 2003; i-Tree 2012); and monetary evaluation of individual trees (CTLA 2000; Cullen 2002; Randrup 2005).
While the urban forest is widely regarded as encompassing all woody vegetation in a city (Randrup et al. 2005), urban tree inventories have mostly focused on information at the single-tree level rather than stands or groups of trees in places such as woodlands or parks (Sjöman 2012; Östberg et al. 2013). The focus on information at single-tree level reflects the motives and goals of city authorities regarding urban forestry programs, where hazard management, traffic safety, arboricultural management, species choice, replanting decisions, and monitoring changes in the urban forest have been found to dominate (Keller and Konijnendijk 2012). However, scientific studies with different objectives and different traditions of urban tree management and planning each have their own specific information needs (Schipperijn et al. 2005). Smiley and Baker (1988) argue therefore that before deciding on what type of data to collect from each tree, the ‘why’ needs to be answered. The present does not only include the “why”, but also the ‘how’ (i.e., how the tree inventory was performed and what technical aids were used to support data collection).
In response to the growing and diversifying use of data on single-tree level from urban tree inventories, the variety of inventory methods has expanded rapidly (Smiley and Baker 1988; McBride and Nowak 1989; Schipperijn et al. 2005). While field surveys offered the starting point, rapid technological development has meant that data at the single-tree level can now also be obtained from different types of ground scanning and digital photography (e.g. Buhyoff et al. 1984; Patterson et al. 2011; West et al. 2012), as well as a variety of satellite- and airplane-supported technical aids (e.g., Jutras et al. 2009; Ardila et al. 2012). Aerial methods have a range of applications for land-use and vegetation inventories at coarser scales (e.g., Mausel et al. 1992; Holopainen et al. 2006; Jutras et al. 2009). However different, all methods have their limitations regarding the data parameters that can be collected at single-tree level and their measurement accuracy, a fact that needs to be taken into account (Abd-Elrahman et al. 2010; Ardila et al. 2012). Therefore, Keller and Konijnendijk (2012) call for more research on the status of urban tree inventories and the accuracy and validity of data that can be obtained from different types of inventories. Against this background, the objectives of the present study were to:
Provide a bibliographic review of previous studies in which urban tree inventories at single-tree level are used as the primary data source.
Establish a typology of contemporary urban tree inventory methods and identify the type of data and the accuracy of measurements collected at single-tree level using different methods.
Evaluate the suitability of different urban tree inventory methods for data collection at singletree level.
METHOD
Search Strategy
The study was conducted as a review restricted to peer-reviewed scientific literature.
Two widely recognized databases were used, Web of Science and Scopus. Aiming for high sensitivity, as recommended by Pullin and Stewart (2006), the search was restricted to a single search string with few search terms, namely: tree* invent* urban*. The search terms were considered among the categories ‘Title, abstract, keywords’ (Scopus) and ‘Topic’ (Web of Science).
The search generated 154 hits. After the initial search, two rounds of selection were undertaken. First, articles were included or excluded based on their title and abstract. The remaining papers were reviewed and evaluated for their relevance. A total of 57 studies met the inclusion criteria, which were:
urban tree inventories used as the main data source
specification of the inventory method and the technical aids applied and the type of parameters collected at single-tree level
published prior to December 31, 2012
published in English
Among the articles that did not meet the inclusion criteria, those specifically studying the accuracy of measurements obtained from one or more inventory methods (but not applying the inventory data in further studies) were identified (n = 9) and used to support the accuracy assessment, which will be explained later. The authors understand that limiting the search to peer-reviewed scientific papers published in English may also have excluded a number of interesting studies published as national reports and papers, but these publications were excluded to assure a systematic search and high data quality.
Data Extraction and Analysis
A standardized data extraction spreadsheet was used to ensure controlled data retrieval. In the case of doubts and queries about whether data from an article or the entire article should be included, a discussion was held among the authors and a consensus decision was reached.
Bibliographic information was extracted on: a) year of publication, b) publishing journal, and c) geographical region in which the study was conducted. The main focus in information extraction was: d) information about the subject-specific focus of the study (henceforth, ‘study focus’), e) the inventory method and type of technical aids applied, and f) the data parameters collected from each tree (e.g., diameter at breast height, tree height, species).
In total, 152 parameters were identified across the 57 studies. However, many were redundant, like variations on the same type of measurement (e.g. crown width, canopy width, canopy drip line width, crown diameter, crown spread, width of crown). After correcting for redundancies, a total of 59 parameters were grouped into 15 types of tree information parameters and three types of data-based parameters [e.g., various items of information on planting site in one group and various abiotic and biotic damage in another group (Appendix)]. Similarly, the inventory methods used to collect the data were clustered into overall typologies, reflecting differences with respect to terrestrial versus aerial methods, mode of transportation, and type of technical aid used to collect the data parameters from each tree.
A comparative analysis was then made of the relationship between study focus, inventory method, and the 15 types of tree parameters collected. The data-based parameters were excluded from the analysis. Finally, the accuracy of the measurements obtained from each inventory method was evaluated for the 15 parameter groups. For each inventory method, the accuracy evaluation was restricted to the parameter groups collected by use of the method and/or tested by studies addressing the accuracy of the measurements. In the evaluation, the findings of studies specifically assessing the accuracy of measurements obtained by different inventory methods was the starting point. Due to the inclusion of both qualitative parameters (e.g., maintenance need) and measurable parameters (e.g., DBH and crown diameter), and the heterogeneity in studies in terms of methodology and reporting of data (some studies reporting deviation from field survey data (e.g., Park et al. 2010) and others omitting this information (e.g., Rutzinger et al. 2011), quantitative criteria for evaluation of accuracy levels were not deemed suitable. Thus the evaluation of accuracy comprised an interpretation of findings across the studies addressing the individual inventory methods, with the accuracy divided into a qualitative scale with four levels: 0 = not possible or with very low precision; 1 = low precision; 2 = intermediate precision; and 3 = high precision, meaning the accuracy corresponded to that obtained from direct measurements on site (field surveys).
RESULTS
Bibliographic Overview
More than half the studies that met the inclusion criteria were conducted in North America (58%, n = 33), while articles originating from Asia accounted for 18% (n = 10), Europe 16% (n = 9), Australia 5% (n = 3), and South America 4% (n = 2). The search did not identify any publications from Africa.
The studies were published in as many as 36 different journals. Arboriculture & Urban Forestry (named Journal of Arboriculture until 2006) was the journal with the most publications, with 26% of the articles (n = 15), followed by Landscape and Urban Planning (n = 4), Urban Forestry & Urban Greening (n = 3), and Remote Sensing (n = 3). The remaining 32 journals were represented with one article each.
Articles were published from 1984 onwards, and two breakpoints in the flow of publications could be distinguished. Up until the mid-1990s, the publications were few and scattered, but from then on, publication stabilized with 1–2 articles per year, nearly all of which originated from North America (Figure 1). A second breakpoint occurred in the late 2000s, where the annual publication rate increased to 3–15 articles. This increase coincided with a noticeable rise in the number of articles reporting research from South America, Australia, Europe, and Asia (Figure 1).
Number of publications on urban tree inventories per year and their geographical origin.
Over the years, the number of different aspects examined in studies increased steadily, to a total of 14 (Table 1). Nearly all studies focusing on the economic benefits of urban trees (grouped under ‘Cost-benefit analysis’) and environmental aspects (grouped under ‘Biodiversity’, ‘Canopy cover’, and ‘Climate/greenhouse/CO2 storage’) were published after 2001. However, studies addressing transmission of volatile organic compounds (VOCs) by urban trees were an exception from the temporal pattern in studies on the environmental aspects of urban trees. The growing number of publications from 2008 onwards is largely due to an increase in the number of methodological studies focusing on ‘test of tree inventory methods’ and ‘digitalization of tree shape’, and is also due to the emergence of studies in which urban tree inventory data on single-tree level are applied to assess ‘Biodiversity aspects’ (n = 3) (Table 1).
Number of articles on urban tree inventories grouped according to study focus and year of publication.
Typology of Urban Tree Inventory Methods
Four types of inventories for data collection at single-tree level with distinct characteristics were distinguished: 1) Satellite-supported methods, 2) airplane-supported methods, 3) on-the-ground scanning or digital photography, and 4) field surveys with direct manual measurements and/or visual inspection. As shown in Table 2, each of these types is supported by different means of technical aid.
Typology of urban tree inventory methods and related technical aids used in collection of data at the single-tree level.
Satellite-supported methods can collect information from very large areas (Cook and Iverson 1991; Small and Lu 2006). Very High Resolution (VHR) images, as well as Panchromatic and multispectral images taken by equipment on satellites (e.g., the QuickBird satellite), can be used to extract information on urban trees. Satellite-supported infrared (IR) scanning images of the wavelengths reflected by vegetation can also be used to collect information at single-tree level (e.g., Jansen et al. 2006). Satellite-supported data collection at single-tree level was used in two of the studies included in the review (Ardila et al. 2012; Cavayas et al. 2012) and to identify trees for subsequent field surveys in two other studies (Thaiutsa et al. 2008; Ningal et al. 2010) (Table 2).
Airplane-supported methods also enable data collection over large areas (Ryherd and Woodcock 1990; Mausel et al. 1992). Like satellites, airplanes can be equipped with appropriate devices, such as an IR scanner or different types of cameras (Goldberg 1981; Mausel et al. 1992; Andarz et al. 2009). However, only two of the studies reviewed adopted these methods (Miller and Winer 1984; Jutras et al. 2009) (Table 2).
Compared with the aerial methods, data collection and processing from on-the-ground digital scanning (Patterson et al. 2011) or photography (Abd-Elrahman et al. 2010) are restricted to rather small areas because each scanning/photography image is restricted to a single tree or a small group of trees. Although this technology is developing rapidly, it is still time-consuming. Of the 57 studies included in the review, five had applied methods in this typology (Abd-Elrahman et al. 2010; Park et al. 2010; Patterson et al. 2011; Rutzinger et al. 2011; West et al. 2012) (Table 2).
Field survey methods comprise direct measurements and/or visual inspection of individual trees by field staff (Adkins et al. 1997; Martin et al. 2011; Östberg et al. 2012). Although field surveys are labor-intensive and time-consuming, 46 of the 57 papers reviewed applied this method to collect their data, making it the most common of the four types. In the search for papers, a subtype of field survey was identified where the data collection is limited to visual inspection of appearance and damage (to inform assessment of hazard status) from a car driven at low speeds. This method has been called windshield survey (Bassett 1976; Rooney et al. 2005) or drive-by survey (Pokorny 2003). However, this subtype was not applied in any of the articles meeting the inclusion criteria (Table 2).
Relationship Analysis
Analysis of the relationship between study focus, inventory method, and parameters collected from each tree showed that data on the three parameters, crown size/density, tree size (other than crown size and DBH), and species information had been collected by studies falling into 12 different study focus classes, while the corresponding number for DBH was 11. Data on the remaining parameter groups, especially tree age information and tree coordinates, were collected by studies with fewer focus classes (two studies each) (Table 3).
Type of tree inventory method used for each of the objectives and number of studies using the methods for each of the objectives. Type of tree inventory method: A, satellites; B, aerial photos; C, ground scanning/photos; D, field survey (windshield method not included). The numbers of studies using the methods for each of the tree inventory parameter groups are listed as subscripts.
Turning the perspective to the study focus, studies testing urban tree inventory methods, tree vitality, arboricultural management, and tree architecture and amenity values had collected a much wider range of parameters, 13, 13, 12, and 11, respectively, compared with studies falling within the remaining 10 study focus areas (Table 3).
Concerning the inventory methods, field surveys had been applied for data collection by all 14 types of study focus areas and as the method for collection of all 15 tree parameter groups (Table 3). In comparison, data collection by use of satellite-supported methods was restricted to the parameter groups ‘tree location’, ‘tree coordinates’, and ‘tree appearance’, and this type of data collection was limited to studies testing the method. Airplane-supported methods were restricted to the testing and use of aerial photos as an aid to tree species identification and estimation of urban tree canopy cover. Ground scanning and digital photography was tested as a method to collect a wider range of tree parameters, and also applied to collect data on crown size/density, DBH, and other size parameters in studies focusing on CO2 storage and/or digitalization of tree shapes.
Tree Inventory Method Accuracy
In the papers reviewed, testing and methodological discussions of measurement accuracy were restricted to 3–5 parameter groups for each inventory method, with on-the-ground digital photography methods and satellite-supported Google Maps being the exceptions. Regarding the latter, accuracy of measurement had been tested for eight parameter groups (Table 4).
Rating of the accuracy of measurements at single-tree level obtained from different inventory methods. The rating is based on synthesis of results from studies focusing on methods into four levels of precision: 0 = not possible to measure or very low precision, 1 = low precision, 2 = moderate precision, 3 = high precision. Field surveys were used as a reference for high precision.
Species information has been identified as the most important data parameter (Östberg et al. 2013). However, only satellite-supported images and on-the-ground costume graded photogrammetry had been tested as alternatives to field inventory for species identification. Abd-Elrahman et al. (2010) found that inventories using Google Maps as data source were unable to identify tree species, while species could be identified from close-range photogrammetry solutions of tree images taken with consumer-grade cameras.
Studies on the accuracy of measurement mostly concentrated on the three size parameter groups ‘Crown size/density’, ‘DBH’, and ‘Tree size other than crown size and DBH’. The following summarizes the findings of studies testing these key parameters, while Table 4 provides the results of the rating of measurement accuracy for these as well as other parameters tested by the different inventory methods.
The different satellite-supported methods were either unable to measure tree size parameters or provided low levels of accuracy. Abd-Elrahman et al. (2010) found that Google Maps were unable to determine tree height and DBH, while crown diameter measurements on average deviated by 1.67 m from field data. Only crown-size data had been extracted from other types of satellite images. For example, Ardila et al. (2012) used geographic object-based image analysis of QuickBird images and found crown diameter to be generally overestimated (no numerical values reported). However, many trees, especially with crown diameter <4.7 m, were not identified.
In studies applying data from airplane-supported LIDAR (Light Detection and Ranging), Jutras et al. (2009) found that the average Pearson prediction coefficient (r) for DBH and crown volume was >90% when analyzed in artificial intelligence multilayer perception network scenarios. In the same study, prediction coefficients of tree height had low accuracy (<90%). Other studies of airborne laser and aerial photos had not been able to establish DBH and tree height information (Table 4).
Concerning on-the-ground scanning, Park et al. (2010) found that modeling of 3D point cloud data from a terrestrial laser (Leica HDS6000 TLS) could reconstruct crown diameter with high accuracy, with only 0.05 m (2%) difference from field survey data. Reconstruction of tree height also had high accuracy, with an average of 0.22 m (3%) deviation from field survey data. However, the reconstruction of DBH (trunk size) at 1 m above ground had low accuracy, with a deviation 0.16 m (22%) from field survey data (Park et al. 2010). In models based on measurements obtained from on-the-ground mobile laser scanning (Optech LYNX System), Rutzinger et al. (2011) found that the DBH on average deviated by 0.8 cm and crown diameter by 0.87 m from field survey data. The paper did not provide sufficient data to recalculate these results to percentage deviation.
In a test of close-range photogrammetry solutions of tree images taken by consumer-grade cameras, Abd-Elrahman et al. (2010) found crown width to deviate by 0%–28% from field measurements. Quantification of DBH was only possible for two out of four trees and varied by up to 19% from field measurements. Quantification of tree height was only possible for three out of the four trees included in the study, and deviated by up to 12% from actual tree height (Abd-Elrahman et al. 2010). Using UrbanCrowns image analysis software, Patterson et al. (2011) found that the error in estimates of crown volume increased with increasing tree size.
DISCUSSION
This review showed that the geographical scope of scientific studies relying on data from urban tree inventories at single-tree level has expanded from being primarily North American during the 1980s and 1990s to now spanning all continents except Africa. The dominance of North American research during the 1980s and 1990s (mostly published in Arboriculture & Urban Forestry) reflects the overall expansion in the field of urban forestry, which developed in North America during the 1960s as an integrative, multidisciplinary approach to the planning and management of all forest and tree resources in and near urban areas. It was not until the mid-1990s that the concept was adopted in Europe and elsewhere, following pioneering work by the U.S. Forest Service on quantifying and modeling urban forest benefits in the Chicago Urban Forest Climate Project (Konijnendijk et al. 2006). It is clear that this work in Chicago stimulated the adoption of urban forestry and urban tree research in Europe and elsewhere, supported by networks such as COST Action E12 on Urban Forests and Trees, which ran from 1997–2002 (Konijnendijk et al. 2005; Schipperijn et al. 2005; Konijnendijk et al. 2006; Konijnendijk et al. 2007), and the re-initiation of an urban forestry unit under the International Union of Forest Research Organizations (IUFRO) in 1986.
In 1983, Smiley and Baker rightfully argued that not until the ‘why’ has been clearly defined can specific data be collected at single-tree level. The findings of the present review indicate that the ‘why’ question remains relevant and may be more imperative than ever, not only for city authorities and urban tree managers (Smiley and Baker 1988), but also for the scientific community. The analysis reveals how data from urban tree inventories at single-tree level have found applications in research with a steadily growing number of different study focus areas. The fact that the 57 articles included in the review are published in 36 journals with differing scientific scope underscores the use of urban tree inventory data across disciplinary borders.
A notable current trend is the use of data at single-tree level in studies focusing on the ecosystem services provided by urban trees (including their economic value), as a supplement to studies linked more directly to daily arboricultural management and planning by city authorities. In fact, all cost-benefit studies and nearly all studies on the regulating environmental aspects of urban trees have been published since 2001, thus coinciding with, and are probably prompted by, the popularization and formal definition of the ecosystem services concept by the United Nations Millennium Ecosystem Assessment 2001–2005 (Millennium Ecosystem Assessment 2000). In a wider perspective, one can therefore argue that the concept of ecosystem services has facilitated an overall change in the scientific use of urban tree inventory data from an applied focus (Sudol and Zach 1987; Smiley and Baker 1988; McBride and Nowak 1989) to the current situation, where it also includes more strategic research (e.g., Dimoudi and Nikolopoulou 2003; Nowak et al. 2006b), as defined by OECD (1994). However, the work in Chicago during the early 1990s pioneered such research and the development of current models, such as STRATUM (now i-Tree Streets) and UFORE (now i-Tree Eco). The very recent increase in studies focusing on technological development and modeling studies also adds to this gradual shift towards more strategic considerations. However, most of the studies reviewed here focus on regulating ecosystem services related to the urban heat island effect, CO2 sequestration, storm water runoff, urban biodiversity, and VOC gases. An interesting perspective for the future would be to explore how tree inventory data at single-tree level can support more strategic research into the cultural ecosystem services provided by urban trees (e.g., non-material benefits related to cultural heritage, social cohesion, recreational experiences, and aesthetic experiences).
The main contribution of the present study is that it goes beyond the ‘why’ question and establishes a typology of ‘how’ urban inventories are conducted (at least, those applied as the main data sources in current peer-reviewed research in the area) and ‘how’ the different contemporary inventory methods and technical aids affect the parameters that can be collected at single-tree level and their accuracy of measurement. Four main types of urban tree inventory methods were identified. With 46 out of 57 articles, the ‘classical’ field surveys, where ground staff carried out direct measurements and visual inspections, dominated. One concern in relation to this inventory method is that it is labor-intensive and generally limited to public trees, simply because of the difficulties in obtaining access to trees on private land (Nowak 2008). Satellite- and airplane-supported methods are less labor-demanding and have the potential to provide easy ‘access’ to trees on private land. It is therefore exciting to see the many recent publications using/testing satellite- and airplane-supported methods (e.g., Jutras et al. 2009; Arroyo et al. 2010; Ardila et al. 2012), ground scanning (e.g., Park et al. 2010; Rutzinger et al. 2011), and digital photography methods (e.g., Patterson et al. 2011; West et al. 2012) for data collection at singletree level. The application and testing of these technologies may give an indication of coming advances.
Key Parameters and their Accuracy of Measurements
In total, 15 parameter groups were applied in the studies reviewed. Studies focusing on tests of tree inventory methods, tree vitality, arboricultural management, and tree architecture and amenity values in particular cover a wide range of parameters (11–13). This corresponds well with the number of parameters typically collected in city authority inventories. For example, Keller and Konijnendijk (2012) found that six major cities in North America and Europe collected data on 8–20 parameters. However, the number and type of parameters varied greatly between studies with different focus areas, with only species information and the three size parameters (crown size/density, DBH, and tree size other than crown size and DBH) being consistently determined. These parameters also obtained a high rating in a recent Delphi study in which city officials, arborists, and academics evaluated the relative importance of tree parameters for inclusion in large-scale urban inventories (Östberg et al. 2013). In addition, these parameters are essential input in the modeling and quantification of the ecosystem services that urban trees provide to the community and their economic value (i-Tree 2012). These four parameters therefore appear to be fundamental, and it is not surprising that size parameters have been the focus of tests of inventory methods that are less labor-intensive than field surveys (i.e., methods supported by satellite images, aerial photos, terrestrial laser scanning and mobile scanning, and on-the-ground digital photography methods). While satellite-and airplane-supported technical aids have found a wide range of applications in land-use and vegetation inventories at coarser scales, current image resolution and laser equipment limit the accuracy of the measurements to the level of single trees. Ground scanning by use of terrestrial laser provided the highest level of accuracy of tree size measurements. However, Park et al. (2010) found that the data processing time varied from 39 to 87 minutes per tree, indicating that it is questionable whether the method is less labor-intensive than the direct measurements and visual inspections applied in field surveys. In contrast, the accuracy of species information data has scarcely been tested, and the test results for satellite-supported images show low levels of precision (Cavayas et al. 2012). In terms of tree health status parameters (categorized in this study as ‘Damage’; ‘Hazard’; ‘Insects, pest, fungi’; and ‘Vitality’), the satellite- and airplane-supported methods and on-the-ground scanning methods again appear to have significant limitations, although their accuracy has not been tested in the papers reviewed. This constitutes an important barrier to application of these inventory methods by city authorities, for whom tree vitality assessment and hazard tree management are among the main reasons for inventory (Keller and Konijnendijk 2012; Thomsen 2012).
LIMITATIONS OF THE REVIEW
The four main types of tree inventory methods identified can be combined to overcome the shortcomings discussed above. This has been done in several studies on emissions of VOC gases by urban trees (Geron et al. 1995; Benjamin et al. 1997; Drewitt et al. 1998; Karlik and Winer 2001; Zhihui et al. 2003; Wang et al. 2003; Wang et al. 2005), where aerial methods have been combined with field surveys to collect high-quality data for extrapolation. In this regard, satellite-supported methods appear to be the typology most frequently combined with another inventory method, primarily field surveys (Diem and Comrie 2000), but also historical documentation (Mickler et al. 2002). While such ‘mixed inventory methods’ were beyond the scope of this review, they offer interesting perspectives that future research ought to explore (e.g., extrapolate inventories to larger areas). Furthermore, the review was restricted to articles written in English and published in peer-reviewed scientific journals. However, tree inventory methods and findings at the local or national level may also be published as reports, guidelines, and ‘gray’ literature, and thus a number of studies in which urban tree inventory methods are described and/or used as data sources may have been overlooked. An obvious example of this is the windshield method applied in practice (Pokorny 2003; Rooney et al. 2005; Escobedo and Andreu 2008), but not as part of the peer-reviewed research included in this review. With this exception, to the knowledge of the study authors, the scientific literature provides a reliable profile of the current status of urban tree inventories. Thus this review provides a useful comparative analysis of the different methods used to collect data at single-tree level.
CONCLUSIONS
This review shows that the scope of scientific studies relying on data from urban tree inventories at single-tree level has recently expanded in geographical and scientific terms. A notable current trend is the emergence of studies focusing on the ecosystem services provided by urban trees, as a supplement to studies more directly linked to arboricultural management and planning practices. The comparative analysis of contemporary inventory methods showed that available technology and current data processing limit the reliability of data obtained from satellite- and airplane-supported inventory methods and on-the-ground scanning or digital photography. The authors therefore recommend further technological development and scientific testing before these methods replace field surveys in urban tree inventory programs at single-tree level.
- © 2014, International Society of Arboriculture. All rights reserved.
LITERATURE CITED
Jump to section
Related Articles
Cited By...
- No citing articles found.