Estimation of canopy properties in deciduous forests with digital hemispherical and cover photography

Rapid, reliable and meaningful estimates of forest canopy are essential to the characterization of forest ecosystems. In this paper the accuracy of digital hemispherical (DHP) and cover (DCP) photography for the estimation of canopy properties in deciduous forests was evaluated. Leaf area index (LAI) estimated from both these photographic methods and from light transmittance data derived from DHP were compared with direct measurements obtained by litter traps (LAI(LT)) and an AccuPAR ceptometer. Also, comparison with different gap fraction methods used to calculate LAI in DHP and LAI-2000 PCA were performed. We applied these methods in four forest stands of Quercus cerris, two stands of Castanea sativa and four stands of Fagus sylvatica, the most common deciduous species in Italy, where LAI(LT) ranged from 3.9 to 7.3. Both photographic methods provided good indirect estimates of LAI(LT). The DCP method provided estimates of crown porosity, crown cover, foliage cover and the clumping index at the zenith, but required assumptions about the light extinction coefficient at the zenith (k), to accurately estimate LAI. Cover photography provided good indirect estimates of LAI assuming a spherical leaf angle distribution, even though k appeared to decrease as LAI increased, thus affecting the accuracy of LAI estimates in DCP. In contrast, the accuracy of LAI estimates in DHP appeared insensitive to LAI(LT) values, but the method was sensitive to photographic exposure and more time-consuming than DCP. The studied stands were characterized by higher within-crown clumping than between-crowns clumping; only the segmented analysis of gap fraction for each ring of the fisheye images was found to provide reliable and useful clumping index in DHP. The 1-azimuth segment method employed in PCA poorly detected clumping in dense canopies. The correlation between transmittance estimates by DHP with values measured at noon with the AccuPAR ceptometer was linear and significant, although the variability observed in reference measures suggested that results obtained with the ceptometer should be treated with caution. We conclude both photographic methods are suitable for dense deciduous forests. Cover photography holds great promise as a means to quickly obtain inexpensive estimates of LAI over large areas. However, in situations where no direct reference measurements of k are available, we recommend using both DHP and DCP, in order to cross-calibrate the two methods; DCP could then be used for more routinely indirect measurement and monitoring of LAI. (C) 2012 Elsevier B.V. All rights reserved.