基于BP神经网络的广东省针阔混交异龄林立地质量评价

目的针阔混交异龄林的地位指数计算一直是立地质量评价中的难点，国内外对针阔混交异龄林的地位指数模型的研究较少，为建立更精确的针阔混交异龄林地位指数模型，把神经网络模型引入针阔混交异龄林的立地质量评价。方法以广东省针阔混交异龄林为研究对象，建立基于神经网络方法的林分优势高模型以及针阔混交林地位指数模型，除年龄因子外，加入了海拔、坡度、坡向、坡位、土壤厚度、腐殖层厚度等立地因子，另外考虑针叶树种与阔叶树种的断面积比重对针阔混交异龄林样地地位指数的影响，并建立针阔混交异龄林的地位指数的计算模型。结果针阔混交林的地位指数的最大值为21.4 m，最小值为6.1 m，平均值为13.7 m，中位数为13.6 m，标准差为3.2 m，地位指数的最大值与最小值的差值为15.3 m。结论从结果中可以反映出，广东省地貌复杂且破碎，山多平地少，立地状况差异较大；另外由于广东省林地树种及树种比例具有复杂多样性的特征，导致基准年龄的差异较大。故地位指数的变异较大。本研究在计算针阔混交林的地位指数时，加入了海拔、坡度、坡向、坡位、土壤厚度、腐殖层厚度等立地因子，提高了针阔混交林地位指数的预估精度。研究结果为针阔混交异龄林地位指数的计算提供了精度更高的方法。 

Site quality evaluation of uneven-aged mixed coniferous and broadleaved stands in Guangdong Province of southern China based on BP neural network

ObjectiveThe calculation to the site index of uneven-aged coniferous and broadleaved mixed stands has always been a difficult point in the evaluation of site quality. At home and abroad, there are few studies on the site index model of uneven-aged coniferous and broadleaved mixed stands. In order to establish a more accurate site index of coniferous and broadleaved mixed stands, the model introduces the neural network model into the site quality evaluation of uneven-aged coniferous and broadleaved mixed stands.MethodIn this study, uneven-aged coniferous and broad-leaved mixed stands in Guangdong province were used as study object, the stand dominant tree height model and site index model to uneven-aged coniferous and broadleaved mixed forests were estabished based on neural network. Besides the age factor, altitude, slope, slope direction, slope position, soil thickness and humus layer thickness were added. The influence of the basal area ratio of coniferous and broad-leaved species on the site index were also considered, then the site index model of coniferous and broadleaved mixed stands was estabished.ResultThe results showed that the maximum value of the site index to the uneven-aged coniferous and broadleaved mixed stands was 21.4 m, the minimum value was 6.1 m, the average value was 13.7 m, the median was 13.6 m, and the standard deviation was 3.2 m. The difference between the maximum and minimum values of the site index was 15.3 m.ConclusionFrom the results, it can be reflected that the landforms in Guangdong Province are complex and fragmented, and there are more mountains and less flat land, and the site conditions are different one another. In addition, due to the complexity and diversity of tree species and tree species ratio in Guangdong Province, then it lead to quite different in the reference age. Therefore, the site index is not the same one another. In the computation of the site index of coniferous and broadleaved mixed stands, the site factors such as altitude, slope, slope aspect, slope position, soil thickness and humus layer thickness were added to improve the prediction accuracy of the site index of coniferous and broadleaved mixed stands. The results provide a more accurate method for the calculation of the site index of uneven-aged coniferous and broadleaved mixed stands.