Key messageWe evaluated the error caused by optical measurements of leaf area index using a direct method in a mixed broadleaf-coniferous forest in China.AbstractIndirect optical methods to measure leaf area index (LAI) have been previously developed, but it is difficult to evaluate the accuracy of these methods in a mixed broadleaf-coniferous forest. In this study, the LAI in a mixed broadleaved-Korean pine (Pinus koraiensis) forest in China was estimated directly by litter collection (LAIlit) for the purpose of evaluating optical LAI measurements using digital hemispherical photography (DHP) and LAI-2000. With the DHP method, we corrected a systematic error due to incorrect automatic photographic exposure. With both DHP and LAI-2000 methods, we studied the influences of zenith angle selection schemes (0°–45°, 30°–60°, 45°–60° and 0°–75°) on the effective LAI (Le) measurement. In addition to optical Le, we also investigated other major factors influencing the determination of LAI, including woody-to-total area ratio (α), element clumping index (ΩE) and needle-to-shoot area ratio (γE). A significant correlation (P < 0.01) was observed between optical (DHP and LAI-2000) and litter collection methods, but DHP Le underestimated LAIlit by 61 % on average based on different zenith angle ranges, and Le at 45°–60° agrees better with LAIlit (R2 = 0.75, P < 0.01 and RMSE = 4.5), and the accuracy was enhanced by 21 % on average after considering α, ΩE and γE and was further improved by 36 % after correcting for the error due to exposure. In contrast, LAI-2000 Le underestimated LAIlit by 32 % on average based on different zenith angle ranges, and Le in rings 1–3 is closer to LAIlit (R2 = 0.80, P < 0.01 and RMSE = 2.1) than those in other rings (e.g., 3–4, 4 and 1–5), and after correcting for α, ΩE and γE, the difference between LAI-2000 LAI and LAIlit was less than 6 %. Although DHP Le underestimated LAI-2000 Le by an average of 43 % at different zenith angle ranges, significant correlations between them were found (minimum r = 0.787, P < 0.01). We confirm the accuracy of the best estimates of LAI using DHP and LAI-2000 methods are to be over 94 % after considering woody materials and foliage clumping within shoots and the canopy. Meanwhile, the litter collection method is useful for estimating LAI in a mixed broadleaf-coniferous forest, if the specific leaf area for all major species and the average leaf age for evergreen coniferous species are known.