Tree Risk Assessment
A bibliography is collection of published literature. Our task was to identify research publications that focus on arboriculture, excluding equipment and climbing methods except as it related to tree risk assessment. Literature from silviculture and pomology was included as secondary sources. Our bibliography focused on articles from peer-reviewed journals (see sidebar at end). To a lesser extent, we included relevant book sections and conference proceedings, significant publications produced by government agencies, and a few magazine articles covering professional practice. Self-published manuscripts, conference hand-outs, posters, and web pages were excluded.
This bibliography includes 700 citations. About half are from peerreviewed journals; and a quarter from key scientific papers related to tree risk assessment. Citations are from over 100 journals published internationally and in 15 countries. Authors are from around the world: North America, United Kingdom, Europe, Canada, Australia, and Asia.
The full bibliography is available through the ISA's website. A literature review of key scientific papers will be published in Arboriculture & Urban Forestry, ISA's scientific journal. There are four primary subject categories; the first two are briefly discussed in this article.
- tree biomechanics
- identification and assessment of structural defects
- risk assessment
- risk reduction or abatement
Scientists have developed a large body of literature about tree biomechanics. Over a third of the scientific articles and books we collected related to that area. While many of the experiments were performed on individual trees, most of the trees were in stands where their structure and exposure to wind was affected by nearby trees.
Articles and books by Claus Mattheck, Lothar Wessolly and others in the 1990's introduced biomechanics to the arboricultural world. Arborists began to describe trees as engineered structures, using equations and terms such as modulus of rupture, applied bending moment, and lever arm.
Biomechanics experiments are designed to quantify the forces imposed on trees and their ability to support the load or fail. Often models are developed to simplify complex computations and interactions. Those models must be tested for validity through scientific experimentation and replication. Few of the models used in arboriculture have undergone that scrutiny.
To apply mechanical and engineering formulas, certain material properties must be known. Forest researchers have measured a variety of wood properties for important lumber species. Key wood properties have been measured for some ornamental species, but not most, and it is unclear how those properties vary in different growing conditions, climates, and exposures.
Most tree failures occur during winds, so the biomechanics of wind and trees is an important area of research. Wind effects on trees have been extensively studied in conifer forests, but hardly at all for shade trees in urban environments. Cullen assembled a bibliography of wind and trees for tree care professionals (Cullen 2002).
Some models and experiments consider trees as non-porous, rigid structures intercepting a constant wind force (static modeling). Others think the dynamic movements caused by wind gusts are important to tree failure and must be considered in modeling tree response (dynamic modeling) (Baker and Bell 1992; James, Haritos, and Ades 2006). A recent wind literature review not only describes the effects of wind on trees, but also how plant movement in wind affects wind dynamics (de Langre 2008). There is a general trend among the literature to use fluid mechanics to describe wind and to model trees as flexible and porous rather than static, rigid structures.
Much of what we know about how trees fail comes from poststorm tree damage surveys (see Duryea et al. 2007 for a summary; Cutler, Gasson, and Farmer 1990; Kane 2008). Ice accumulation causes damage to trees, and the degree of damage varies by species and location (Hauer, Weishen Wang, and Dawson 1993; Luley and Bond 2006). As an example, Norway maple (Acer platanoides) is relatively resistant to ice damage, while silver maple (A. saccharinum) is extensively damaged (Sisinni, Zipperer, and Pleninger 1995; Hauer, Weishen Wang, and Dawson 1993).
Another source of information about tree failures is the International Tree Failure Database (formerly the California Tree Failure Database). Summaries of reported failures have been used to develop tree failure profiles for only two species, however: coast live oak (Quercus agrifolia) (Edberg and Berry 1999), and Monterey pine (Pinus radiata) (Edberg, Berry, and Costello 1999).
Tree risk assessments rely on identifying and assessing structural condition to assess failure potential. 'Defect' is the term we commonly use to identify a condition or characteristic that is structurally weak or contributes to a structural weakness. Yet, there has been limited scientific study of what characteristics are important, or how to translate what we see into likelihood for failure.
There are many publications that describe how to perform a tree risk assessment. All have similar components: visually assess tree structure, describe defects, evaluate the likelihood of failure, and note what would be damaged if the tree failed. There are no scientific publications, however, that evaluate, test or compare the procedures or methods. We do not know which, of the methods or procedures are most important or accurate.
In most tree inspection protocols, the arborist is instructed to systematically view the tree from top to bottom and move 360o around it. But are there other methods that give acceptable results? Rooney et al. (2005) compared walking and windshield inspections. During the windshield inspection, the arborist assessed trees as the vehicle was driven along the street at an average speed of 3.06 km/h (1.9 mph). Both sides of the trees were examined if there was vehicular access; if not, only one side was inspected. A comparison of the hazard ratings assigned trees during the windshield inspection with ratings of the same trees in a walking inspection indicated that the windshield inspection accuracy increased as the tree hazards become more severe (rated 10 and above on a 3-12 scale). The authors noted that, "If the trees are reasonably maintained, the windshield survey could be used just to locate quickly developing hazardous conditions such as hanging branches or recent storm damage, or for an annual update of street side conditions."
As long as arborists have been examining trees we have used external characteristics to give us clues about internal conditions and assess structural stability. Mattheck (1994) described this process as Visual Tree Assessment (VTA), which is widely used in tree risk assessment.
How accurately does a visual assessment represent internal conditions? Experiments to answer that question are limited. Nor do we have data identifying which characteristics are most likely to result in tree failure except under a few specific conditions.
Hickman and others (1995) evaluated and rated 695 oaks (Quercus wislizneii and Q. lobata) for eleven components including site factors, tree structure and vigor, and target value. Seven years later they reexamined the trees to identify which had failed or died. They found that three factors - decline (leaf cover and color), trunk condition and lean - were most closely related to failure, and decline was the most important predictive characteristic.