论文部分内容阅读
L ANDIS模型是模拟自然和人为干扰下森林景观变化的空间直观景观模型。模型把景观概念化为由相同大小的像元或样地组成的格网。在每一个像元上 ,模型要求输入物种和年龄组信息。但是 ,由于研究区一般由成千上百万个像元构成 ,不可能通过实际调查获取每一个像元上的物种和年龄组信息。因此 ,采用了一种基于小班的随机赋值法从森林调查数据中获取每一个像元的物种和年龄组信息。该方法是一种基于概率的方法 ,会在 L ANDIS模型模拟的物种和年龄组信息的输入中引入不确定性。为了评价由基于小班的随机赋值法所引入像元尺度上的不确定性对模型模拟结果的影响 ,用蒙特卡罗模拟法进行不确定性分析。对 L ANDIS模型模拟的每一个物种 ,用众数年龄组发生频率来定量化单个像元上年龄组信息的不确定性 ,用所有像元上的众数年龄组平均发生频率来定量化年龄组信息在像元尺度上总的不确定性。平均发生频率越高 ,不确定性越低。为了评价基于小班的随机赋值法对景观尺度上模型模拟结果的影响 ,计算了每一个物种在整个研究区内的面积百分比和聚集度指数。变异系数越大 ,不确定性越高。对所有物种 ,年龄组信息不确定性在模型模拟的初期是比较低的 (平均发生频率大于 10 )。种子传播、建群、死亡和火干扰使模型结果的不确定?
The L ANDIS model is a spatial and intuitive landscape model that simulates changes in the forest landscape under natural and man-made disturbances. The model conceptualizes the landscape as a grid of pixels or plots of the same size. On each pixel, the model asks for species and age group information. However, since the study area is generally composed of millions of pixels, it is not possible to obtain species and age group information on each pixel through actual surveys. Therefore, a species-based and age-group information for each pixel was obtained from forest survey data using a small-class random assignment. This method is a probability-based approach that introduces uncertainty into the input of species and age group information simulated by the L ANDIS model. In order to evaluate the impact of the uncertainty of pixel dimension introduced by small class based random assignment on model simulation results, Monte Carlo simulation is used to analyze the uncertainty. For each species simulated by the L ANDIS model, the population age group frequency was used to quantify the uncertainty of the age group information on a single pixel, and the age group was quantified using the average occurrence frequency of all age groups on all pixels The total uncertainty of the information on the pixel scale. The higher the average frequency of occurrence, the lower the uncertainty. In order to evaluate the effect of random assignment based on small classes on model simulation results at landscape scale, the area percent and aggregation index of each species in the entire study area were calculated. The greater the coefficient of variation, the higher the uncertainty. Information uncertainty for all species and age groups is low at the beginning of the model simulation (average frequency of occurrence is greater than 10). Seed propagation, colonization, death and fire disturbance make the model uncertain?