Can professionals gauge likelihood of failure? – Insights from tropical storm Matthew

Abstract

Visual risk assessment remains the primary means of gauging urban tree safety and is a key facet of storm preparation and response. While past research has investigated the reproducibility of risk assessment methodologies (i.e., precision), few, if any, studies truly address the accuracy of current inspection practices – especially with regard to the characterization of likelihood of failure. In 2016, Hurricane Matthew made landfall in the Southeastern United States as a lower-intensity tropical storm, impacting several urban sites where tree risk assessments had been conducted in the recent past. After the storm, 2069 trees on 5 properties were revisited to assess storm damage. The vast majority (93%) of trees survived Matthew intact, with 6% of the assessed population suffering partial (i.e., branch) failure and the remaining 1% experiencing whole-tree failure. Failure rates differed by species, with age, and given the presence of external defects. The presence of dead branches (P-value < 0.001), deep planting (P-value < 0.001), severe stem-girdling roots (P-value = 0.020), and previous wounding (P-value = 0.016) were associated with increased likelihood of failure. The original risk assessments were fairly accurate: 94.1% of trees assessed as having an “imminent” likelihood of failure were damaged in the storm. In contrast, 38.8% of trees rated as “probable”, 15.3% of tree rated “possible”, 0.0% of trees rated “improbable” with regard to likelihood of failure were damaged during the storm.