Estimation of plant area index and phenological transition dates from digital repeat photography and radiometric approaches in a hardwood forest in the Northeastern United States

Long-term, continuous digital camera imagery and tower-based radiometric monitoring were conducted at a representative hardwood forest site in the Northeastern United States, part of the AmeriFlux network. In this study, the phenological metrics of the leaf area index (LAI), plant area index (PAI) and associated transition dates (e.g., timing of the onset of leaf expansion and the cessation of leaf fall) were compared using 4-year of data from Bartlett Experimental Forest. We used digital repeat photography (DRP) imagery collected using two different methods (“canopy cover” and “phenocam” approaches), together with above- and below-canopy measurements of photosynthetically active radiation (PAR). The growth-period LAI estimated from canopy cover images (LAICANOPY) and the above and below canopy PAR measurements (LAIfPARₜ) were within approximately the same range, in term of magnitude, as previous results for multiple comparative methods, although growing-season LAICANOPY was slightly lower (3.11m²m⁻² to 3.35m²m⁻²) than LAIfPARₜ (3.19m²m⁻² to 3.67m²m⁻²). In addition, we derived phenological transition dates from PAICANOPY, PAIfPARₜ, and color-based metrics calculated from the phenocam imagery (green (GCC) and red (RCC) chromatic coordinates). The transition dates in both spring and autumn differed somewhat according to method, presumably due to the vegetation status detection abilities of each vegetation metric. We found that LAI estimation from canopy cover images may be influenced by automatic exposure settings, which limits the ability to detect subtle changes in phenology during the transition phases in both spring and autumn. Particularly in autumn, the color-based metrics calculated from the phenocam imagery are decoupled from leaf area dynamics and thus PAI. While above and below canopy PAR measurements could yield the better indicators for estimating LAI, its seasonal dynamics, and associated phenological transition dates in long-term monitoring, we argue that there are obvious benefits to the multi-sensor approach used here.