Tree tilt monitoring in rural and urban landscapes of Hong Kong using smart sensing technology

Abstract

Urban trees are beneficial to our environment and important to human inhabitants. However, they are exposed to natural and anthropogenic stressors, such as strong windstorms, extreme wind events and accidents; inducing tree falling which can cause personal damages, economic losses and infrastructural destructions. The current study is the first of its kind, presenting a tree monitoring system, and using smart sensing devices installed on more than 8000 trees in Hong Kong's rural and urban landscapes. A description of the key components of the system, followed by big data analysis and three case studies of strong wind events over the past 2 years, are presented. A network of smart sensing devices was deployed to develop a large-scale, long-term, smart tree monitoring framework; to help identify potentially hazardous trees in urban areas, particularly during extreme weather events. The changes in tree tilt angle under natural wind loading were recorded. Patterns and responses of tree tilt angles were analyzed, with prediction using time series models based on the Seasonal Autoregressive Integrated Moving Average (SARIMA) and Extreme Gradient Boosting time series forecasting (xGBoost). The results showed the highest correlation for 1-hour forward forecasting, by applying xGBoost model on tree tilt data and weather observations (R2=0.90). On the other hand, SARIMA model produced one-step-ahead prediction with correlation (R2) ranging from 0.77 to 0.93, while lower correlation (R2 ≤ 0.55) was observed for long term prediction (15 days) of the tree tilt angles. Finally, a dashboard and mobile applications of tree monitoring systems were developed, to transfer knowledge and engage the public in understanding associated hazards with tree failures in the urban area.