Abstract: Based on the simulated data produced by PROSAIL model and field test data, the authors respectively analyzed many kinds of factors which affect some representative inversion methods, such as VI, second derivative, model inversion and the directional second derivative The factors taken into account include background, directional variation of canopy reflectance and random errors and so on The results demonstrate that under various conditions, there are huge differences araong the inversion results resulted from these methods Vegetation index NDVI has the lowest accuracy, could not avoid those influences The accuracy of model inversion is better than that of VI method, but is still influenced by the variety of backgrounds The second derivative can partly remove the influence of backgrounds, but it is also limited by the anisotropy of canopy reflectance By comparison, the directional second derivative can improve inversion accuracy through removing the influences from the soil background and directional variation of canopy reflectance simultaneously As second derivative is sensitive to noise, the elimination of noise is crucial to the inversion result

Key words: leaf area index, inversion method, directional second derivative

摘要： 以PROSAIL模型模拟数据和地面实测数据为基础，分别分析了土壤背景、冠层反射率非各向同性以及随机噪声等因素对几类代表性反演方法的影响（植被指数法、二阶微分法、模型反演法以及方向性二阶微分法）。结果表明在不同条件下，各类反演方法的反演精度差别较大。植被指数NDVI对几种因素的滤除能力都较差，反演精度最低；模型反演精度高于植被指数方法，但会受到土壤背景的影响；二阶微分方法虽然能部分消除土壤背景的影响，但受冠层反射率非各向同性的限制。文中提出的方向性二阶微分法能较好地消除土壤背景和冠层反射率非各向同性的影响，反演精度较前者有所提高，但二阶微分方法易受噪声影响。

关键词: 叶面积指数, 反演方法, 方向性二阶微分