基于多元回归与人工神经网络的橡胶园树龄遥感反演研究

树龄是生产管理、森林生态系统研究如叶面积指数（LAI）反演，净初级生产量（NPP）估算的重要参数之一。以国营新盈农场为例，研究橡胶园树龄与美国陆地卫星TM影像之间的关系，分别应用多元回归和人工神经网络的方法建立了橡胶园树龄遥感反演模型。研究结果显示：1）TM波段和橡胶园树龄呈显著负相关，其中近红外（B4）、红外波段（B5）与树龄的相关系数最高，分别达到-0.70和-0.69；2）人工神经网络模型能克服建模数据非正态、非线性及共线的影响，能够明显的提高模型的预测精度，绝对预测误差（ε≤6年）的百分比为81.2％，远高于回归模型的69.2％； 3）由于橡胶树生长特性、自然灾害等因素的影响，多元回归模型和人工神经网络模型都存在估计偏差，即高估小于25龄的橡胶园树龄和低估对应老龄胶园的树龄。研究结果表明，利用人工神经网络的方法进行橡胶园生物物理参数遥感反演具有良好的应用前景。

Estimation of Rubber Plantation Age Using Statistical and Artificial Neutral Network Approaches with Landsat TM Data

Tree age is an important factor for production management and predicting forest biophysical characteristics such as leaf area index(LAI),net primary production(NPP)for ecosystem modeling.The study explored the relationship between rubber(Hevea brasiliensis)Plantation age and Landsat Thematic Mapper(TM)imagery,and investigated the multivariate regression and artificial neutral network(ANN)approaches to perform the predictive modeling of the age of rubber Plantation in Xinying state farm.The results indicate that: 1)all the TM bands were negatively correlated with stand age,of which the near-infrared and infrared band(B4 and B5)had the strongest correlation coefficients;2)the ANN prediction models,which had the absolute prediction error(ε≤6years)of 81.2% and much higher than regression models(69.2%),could explain more difference in tree age than multivariate regression models because of their ability to take into account nonlinear,non-normally distributed data;3)both the regression and ANN models showed an overestimation of young tree age and underestimation matvrre tree age due to the growth characteristics and disturbance of typhoon and cold weather.The results suggest that ANN approach may be of significant value when using remote sensing data to model certain rubber tree variables in Hainan Island.