基于阶跃函数的红树林凋落物变化模型研究

由于凋落物随时间变化而存在峰值,按一般的回归方法或分段拟合方法均不能建立理想的凋落物随时间变化的模型。利用阶跃函数,解决了分段模型一直无法解决的两个问题:一是变点的数学确定方法,另一个是变点的连续性问题。建立了基于符号函数的阶跃函数模型,并以此为基础,提出了具有峰值的凋落物耦合模型。为了检验凋落物模型的效果,以雷州半岛红树林为例,结果表明,决定系数0.8以上,回归模型达到显著性水平。

Study on the model for litter-fall varying in mangrove forest based on a unit step function

The quantity of detritus changes monthly and there is a peak value in its changing curve.A model for litter-fall varying in mangrove forest can not be established by usual regression methods or fragmented fitting.The two questions to establish the model in common ways,one is about how to determine the transform point,the other how to make the curve continuous at the point,have been solved by a unit step function.The model of the unit step function based on a sign function was developed,and according to the model,the other litter-fall ingrated model with the peak value was created.The model for mangroves detritus was tested in the Leizhou Peninsula,and the conclusion shows that the determinative coefficient is more than 0.8 and the model reached a marked level.