Evaluation and correction of optically derived leaf area index in different temperate forests

Abstract: In recent years optical techniques for rapid LAI measurements have been developed, but few studies have been performed to evaluate the accuracy of optical estimation of LAI in mixed deciduous-evergreen forest stands. In this paper, we assessed the accuracy of digital hemispherical photography (DHP) and the LAI-2000 for the estimation of effective LAI (Le) by comparison with litter collection LAI (LAIlit) in four mixed deciduous broadleaf and evergreen needleleaf forests and one deciduous needleleaf forest. We also evaluated the relative contribution of major error sources to the determination of LAI by optical methods, including the woody-to-total area ratio (α), the element clumping index (ΩE) and the needle-to-shoot area ratio (γE). Additionally, incorrect automatic photographic exposure has been considered for DHP. DHP Le underestimated LAIlit by an average of 44-70% in different forests, and the difference between LAIlit and DHP Le after correction for the automatic exposure, α, ΩE and γE ranged from 1% to 21% in five forest stands. In contrast, LAI values from LAI-2000 were more similar to the direct litter collection LAI. The LAI-2000 Le underestimated LAIlit by an average of 13-40% in these forests, while the accuracy of the best estimates of LAI using LAI-2000 methods is over 93% after considering α, ΩE and γE. The error caused by automatic exposure to DHP Le is larger than other factors in all forest stands, and the γE was the main uncertainty to LAI-2000 Le in most forest stands. Moreover, optical LAI (both DHP and LAI-2000) was significantly (P < 0.01) correlated with LAIlit, especially the corrected LAI obtained by the LAI-2000 (R2 = 0.83, RMSE = 1.04). Our results demonstrate that the above factors affect the estimation of LAI by optical methods, thus the species composition of a forest stand should be seriously considered in order to improve the accuracy of LAI by optical methods.