From this review, it can be seen that there are many opportunities for the use of RS in the management of trees, and that RS can offer solutions to resolve long-term operational issues in both tree management and research. Traditional methodologies that remain widely used, have seen limited development in the last 100 years, however, these methods remain the basis for the majority of modern tree assessments. These traditional methods are widely used to complete tree assessments for a range of purposes, from environmental studies, to protection of the public for health and safety surveys or the collection of data for forest inventory. A key limitation of these largely manual methods, are that they are both time and, financially costly techniques that only provide a limited amount of information about the subject trees.
These techniques are also limited by the physical capacity of the surveyor. Simply due to the limitations of where a surveyor can access a tree, there is a tendency to rely on the same, ground-based perspective for data capture and to measure only what the surveyor can physically touch. Either this means that large areas of the tree are not measured, or there has to be elements of data extrapolation to fill in missing information. In addition, with the continued use of the traditional methods, there is also a disproportionate reliance on the skills, knowledge and experience of the
individual surveyor for qualitative observations. This has been shown to introduce operator bias or subjectivity into the field data, upon which all subsequent measurements and calculations are based. It is recognised that there is a need to improve upon these traditional methods, and indeed, tree managers indicate the potential acceptance of new techniques and methodologies, predominantly where these new approaches can directly aid the management of the environment. However, there frequently remains some disconnection between the theoretical exploration of a problem, and the translation of the theory into practical solutions that can be employed in any real operational sense.
For successful tree management and academic research, there is often a need to record information about trees over time, in particular for capturing an evidence base for the quantification of change. This review shows that there are several RS solutions that can be employed to fill the requirements of tree managers and researchers in capturing accurate, repeatable and objective environmental data. This RS approach is valuable for investigations that will look at trees across a range of scales, from individual tree assessment, up to wider landscape applications of trees in woodland or forest settings. RS is shown to be able to provide opportunities to investigate elements of tree structure that otherwise would be unobtainable to the traditional surveyor, and furthermore, allows the capture of data that will lead to more insightful investigations and conclusions reached about the tree’s condition. This review identifies more than one RS method that can be employed in this manner, specifically, the use of proximal hemispherical imagery for individual tree assessment and recommends the use of ALS LiDAR for the investigation of trees in their wider landscape setting.
This review has identified that there are particular requirements for the successful application of these RS methods. These range from ensuring that hemispherical images are captured from beneath tree crowns looking towards the zenith viewpoint, to identifying a recommended method for the reduction of data noise in LiDAR investigations and considerations for successful ITC delineation and the classification of trees using analytical metrics. Moreover, this search of the literature on the subject also indicates there are opportunities to improve on the current practice of traditional methodologies and provide a technique that is readily accessible for a field operative to be able to classify tree structure. Similarly, previous research involving ITC delineation has frequently relied upon the acceptance of arbitrary thresholds for resolving pairwise matching problems, and again, this provides the prospect for a new method for the identification of tree locations and extents, with a quantification of successful matching between two ITC datasets. Furthermore, the review also shows a need to develop a combined technological solution used for identifying and characterising tree structure, which utilises techniques that are transparent, repeatable and will indicate the amount of confidence in the application of the data for tree managers or researchers. Ultimately enabling the end-user to gain new insights on the overall condition of the trees under assessment, as a prelude to informing the decision making process for tree management or further academic research.
3
Field Sites, Methods &
LiDAR Data Specification
For this research project, a substantial amount of field-based data collection was undertaken to provide a robust GR data set to be used in the validation of RS data. This necessitated the use of a series of different field methodologies, each with unique specifications and operational requirements. The fieldwork relates to a series of both direct and indirect measurements of the subject trees, using several data capture methods. These methodologies are essential for the completion of the experimental elements of this investigation, and are not covered elsewhere within this thesis. Hence, these methodologies are described to provide transparency, and, when read with the methodologies in later chapters, to enable other investigations to repeat the procedures used. These field-methodologies drawn from a variety of sources, and are either based on forestry and arboricultural industry specific best practice, or are techniques from the academic research community.
Several of these field data collection methods were used in combination in order to achieve the required objective. For example, a tree may be manually surveyed and proximally photographed. As such, each of the subsections of this chapter should not be considered isolated methods completed in sequential order, but a general record of field methodologies and data collection techniques, used concurrently where required. The trees used within the investigation were recorded across a series of different field sites each with unique combinations of tree species, age, geolocations and
topographic exposure, and in a range of differing physical conditions. Therefore, an overview of typical land use and local environmental conditions at each field site is also described.