Aspen Bibliography

Leaf area measurements based on hemispheric photographs and leaf litter collection in a deciduous forest during autumn leaf-fall

Document Type


Journal/Book Title/Conference

Agricultural and Forest Meteorology





First Page


Last Page


Publication Date



Predicted relationships between leaf area index (LAI) and gap frequency using the Poisson, binomial and Markov theoretical models of canopy geometry (Nilson, 1971) were used to estimate LAI from hemispherical photographs taken during the autumn leaf-fall period in a maple-aspen forest in southern Ontario, Canada. Both the binomial and Markov models require specification of a parameter to describe the clumpiness of the leaf distributions. For the binomial model, the parameter value given by Baldocchi et al. (1985) for an oak-hickory forest in Tennessee was used, while for the Markov model, a means for estimating the required parameter was developed based on conditional gap probabilities deduced from the hemispheric photographs. LAI estimates derived from the theoretical models were compared with values obtained from leaf-litter collection as the canopy went from fully leafed to leafless (LAI range 0–∼5.1). Comparisons were obtained for 6 days over this period. Estimates from the models were based on photographs taken at the bottom of the canopy at nine locations within a 15 × 15-m plot. Analysis of photographs was performed using a video camera and a commercial image analyser. Poisson model-derived results were appreciably less than the LAI from leaf-litter collection for values of the latter greater than ∼2, while results from the binomial and the Markov models compared favourably with leaf-litter collection for LAI>2. The apparent failure of the Poisson model at the relatively larger LAI's was attributed to the expected clumpiness of leaf distributions in deciduous forests. For LAI's less than ∼1, the estimates from all models exceeded the leaf-litter collection values, reflecting the influence of tree branches on the gap frequencies recorded in the hemispheric photographs. An estimate of 0.5 was obtained for the woody element area index based on the mean collection LAI estimates. For the Markov model, this difference showed little trend with decreasing leaf area, but for the binomial model, the corresponding difference tended to increase as LAI decreased, probably reflecting a change in the true clumpiness parameter from the full canopy value as leaf area decreased. Estimates of leaf area density profiles were also obtained. These were derived from hemispheric field-of-view photographs taken at seven levels from a tower within the forest. Only the equivalent of one photograph at each level for any measurement day could be obtained, which increased the inherent uncertainty of the estimates relative to the entire canopy estimates, which were based on nine photographs. With a little subjective smoothing of the LAI profiles produced by application of the Markov model, physically consistent leaf area density profiles were produced which appeared reasonable based on the few measurements of such profiles that have been reported.