古田山自然保护区四个常绿阔叶树种叶面积的估算模型

Leaf Aare Estimation Models for Four Evergreen Broadleaved Tree Species in Gutianshan Nature Reserve

通过对浙江古田山自然保护区4个主要阔叶树种——甜槠、木荷、青冈和石栎的调查采样,测定了4个物种共1160张叶片的叶长、叶宽和叶面积,基于此构建4个物种叶面积与叶长和(或)叶宽的8种常用回归模型。结果表明,模型SA=a(LW)+b具有最高的决定系数(R^2=0.97-0.99)和最低的均方差(MSE=0.52-1.59);幂指数模型SA=aW^b是所有单变量模型中拟合效果最佳的模型。

Investigation was conducted on 4 major broadleaved species such as Castanopsis eyrei, Schima superba, Cycloblanopsis glauca, and Lithocarpus glaber in Gufianshan Natural Reserve, Zhejiang province. Leaf length （L）, width （W） and leaf area （LA） of 1 160 leaves of 4 species were determined. 8 common regression models were established. The results showed that the model SA = a（LW）＋b had the highest determination coefficient of （R2 = 0.97-0.99） and the lowest mean square error （MSE = 0.52-1.59）, and thus was the most accurate model for estimating LA. The result also demonstrated that among all the univariate models, the power regression model SA = aWb might have the best fitting effect due to the highest determination coefficients and the lowest mean square error.