Estimating the leaf area of an individual tree in urban areas using terrestrial laser scanner and path length distribution model

Urban leaf area measurement is crucial to properly determining the effect of urban trees on micro-climate regulation, heat island effect, building cooling, air quality improvement, and ozone formation. Previous works on the leaf area measurement have mainly focused on the stand level, although the presence of individual trees is more common than forests in urban areas. The only feasible ways for an operational non-destructive leaf area measurement, namely, optical indirect methods, are mostly limited in urban areas because light path is constantly intercepted by surrounding buildings or other objects. A terrestrial laser scanner (TLS), which can extract an individual tree by using its unique distance information, provides a possibility for indirectly measuring the leaf area index (LAI) in urban areas. However, indirect LAI measurement theory, which uses the cosine of an observation zenith angle for path-length correction, is incompatible for an individual tree because the representative projected area of LAI changes while the observation zenith angle changes, thus making the results incomparable and ambiguous. Therefore, we modified a path length distribution model for the leaf area measurement of an individual tree by replacing the traditional cosine path length correction for a continuous canopy with real path length distribution. We reconstructed the tree crown envelope from a TLS point cloud and calculated a real path length distribution through laser pulse-envelope intersections. Consequently, leaf area density was separated from the path length distribution model for leaf area calculation. Comparisons with reference measurement for an individual tree showed that the TLS-derived leaf area using the path length distribution is insensitive to the scanning resolution and agrees well with an allometric measurement with an overestimation from 5 m² to 18 m² (3–10%, respectively). Results from different stations are globally consistent, and using a weighted mean for different stations by sample numbers further improves the universality and efficiency of the proposed method. Further automation of the proposed method can facilitate a rapid and operational leaf area extraction of an individual tree for urban climate modeling.