摘要
利用Jarvis和Ball两种常规气孔导度模拟模型和人工神经网络模型(ANN)对中国北方城市常见的6种绿化植物的夏季气孔导度变化动态进行了模拟。结果表明:贴梗海棠(Chaenomeles speciosa)和龙爪槐(Sophora japonica)的最优模拟模型为Ball模型,木兰(Magnolia liliflora)最优模拟模型为Jarvis模型,而日本樱花(Prunus yedoensis)、金钟花(Forsythia viridissima)、紫荆(Cercis chinensis)的最优模拟模型为ANN模型。这表明ANN模型可以有效的改善气孔导度环境响应行为的数值模拟。环境因子的多元逐步回归结果和人工神经网络模型敏感性分析表明PAR是夏季气孔导度环境因子中最敏感的要素,而Ta则是其中最不敏感的要素。
Two common-used simulation modes of stomatal conductance, the Ball model and the Jarvis model, and the artificial neutral network （ANN） were employed to simulate the dynamics of summer stomatal conductance of six common greening species in cities of northern China. Results show that, the Ball model is the best one for the simulation of Chaenomeles speciosa and Sophora japonica, and the Jarvis model is the best for the simulation of Magnolia liliflora; while the ANN model is the best one for the simulation of Prumts yedoensis, Forsythia viridissima and Cercis chinensis. This indicated that the ANN model could improve the simulation of stomatal conductance. The multiple regression and sensitivity analysis show that PAR is the most important and sensitive factor of the stomatal conductance in summer; while Ta is the most not sensitive one.
出处
《生态学报》
CAS
CSCD
北大核心
2007年第10期4141-4148,共8页
Acta Ecologica Sinica
基金
国家教育部博士点基金资助项目(20030422030)
山东省优秀中青年科学家科研奖励基金(2005BS08010)~~
